high quality anime ripping

[Lev]

hey,

i want to rip anime in ultra high quality.. 300 mb per episode in a resolution of 640 x 480.. what filters should i use?
i mean is there still need for filters like warpsharp or smart smoother with these high bitrates?

[h00z]

Warp sharp and smart smoother help to reconstruct the original image, they have nothing to do with bitrate.

They way I do my anime is this (and it's a round-about way, but it works for me :):

I make my .d2v file in DVD2AVI, then convert it in VFAPI. I then use AVISynth and GreddyHMA to create an .avs file that I can open in Nandub. Once in Nandub is use Smart-Smoother at 3/25 and Warp Sharp at 40 (of course resize and crop as well). Then just encode as you normally would.

I also always use a true (LAME encoded) VBR MP3 for the audio track. But hey... That's just me :)

[obelix]

Hi h00z,

why do you convert it with VFAPI? This should give you awfully low framerates.
I prefer DVD2AIV-project, GordianKnot -> AVISynth frameserver. This avs you can open with Nandub where you can apply your filters.
I'm not sure if I'm right, but it looks to me as if you are cascading frameservers.

[fisix]

he needs to perform inverse telecine (IVTC) on the
stream, and doom9's guide suggests this process to
get very good results. there are other ways, but this
way is reasonably fast, and you can look at the stream
frame by frame if you want to find places where
plug and chug ivtc might stumble.

[h00z]

Yep. What he said :)

Actually my bottleneck is the application of the filters. The settings I use increase encoding time dramatically, but I have had comments about the fact that some of my rips actually look better than the source DVD (due to the image reconstruction).

On my dual Duron 1ghz machine I encode at ~7-8fps using my settings. Not speedy, but it is worth it for the final quality.

[Lev]

h00z y do you use AVISynth and GreddyHMA to convert the avi to avs and not just use the .avi vfapi created?

and in what order do you use your filters?

[h00z]

Yes, I use AVISynth and GreedyHMA to decimate the VFAPI AVI from 30fps to 24fps (Yes, IVTC makes a HUGE difference in anime). A sample .avs file goes like this (the bold '3' is important):LoadPlugin("D:\downloads\Avisynthv105\GreedyHMA\GreedyHMA.dll")
AVISource("D:\vobs\cowboy.bebop\1\cb1.1_d2v-vfapi.avi")
return GreedyHMA(0,0,3,0,0,0,0,0)Then my filters go like this (ignore the beginning resolution, I just grabbed a screenshot to show the filters and settings :)):

http://www.hooz.org/misc.images/filters.gif

Does that help?

[Lev]

thnx h00z
for pal (25 fps) i don't have to use AVISynth and GreedyHMA right?
i use the smart deinterlace filter..

[charlesky]

IVTC is essential if you want quality. De-interlacing will just produce a 29.97fps file, whereas all you need is 24fps, ie you're wasting bits on 6 frames each second that shouldn't be there. You'll also see artifacts from the deinterlace filter on many of the frames.

Personally, I create the .d2v file, load it into TMPGEnc and run it through the IVTC panel there (24fps flicker priority). I then save the .tpr project and convert it with VFAPI to load into Nandub for filters and encoding. TNPGEnc's IVTC filter is pretty powerful, but the main advantage is that it allows you to correct the stream manually, a function that often seems to be needed with anime.

Most anime is shot on film at 24fps, so that should be your target framerate, but anime studios are notorious for getting "clever" and will sometimes put in 29.97fps pans and CG segments that are shot on video (grrr).

I encode my stuff at around 1900kbps, which puts 2 24min episodes on a 700M CDR. The quality is very acceptable, but I wouldn't call it 'ultra-high': it's still possible to see a bit of fixed-pattern and mosquito noise if you look closely.

[h00z]

GreedyHMA is proving to be a better IVTC solution (not to mention *MUCH* easier) than TMPGEnc.

As for bitrate, I just use Nandub's bitcalc for my anime. For individual episodes (24 minutes +/-) I shoot for 170mb final size, normal movies ~645mb. That might seem small, but with the filters and variable bitrate of Nandub's SBC... I consider it VERY high quality.

Of course, YMMV ;)

[h00z]

Check this screenshot of one of my Cardcaptor Sakura rips. Pay special attention to the ghosting around her hair (my rip is the top frame, original DVD source on the bottom of course).

http://www.hooz.org/misc.images/quality.jpg

The quality of the rip actually looks a little better than the source after filters. What do you think?

[h00z]

I need to clarify that the "ghosting" I was talking about is due to the compression used by the MPEG2 encoder when making the DVD. Some of that will come back when encoding into DivX (MPEG4), but with the filters you at least get to start from a clean slate so to speak, rather than adding more compression artifacts on top of the existing artifacts.

With these particular filters you get as good (or maybe a little better) image quality with your final file as you have on the DVD IMHO.

I plan on applying my methods to the new DivX 4 CODEC in VirtualDub to see how well it works and also to see if it is any faster. Overall though, wise use of filters with Nandub's SBC and DivX 3.11 is the best way to rip anime that I have seen.

[manono]

Hi-Since you're working with PAL material (something you should have stated in the beginning), then you can ignore much of what h00z and charlesky said about NTSC material IVTC, not that it's wrong, but it doesn't apply to you. I think you might want to use GreedyHMA anyway for PAL material, if for no other reason than it's so much faster than the VFAPI method. GreedyHMA has settings specific to PAL material. Assuming you have the plug in and know how to set it up, then GreedyHMA(1,0,1,0,0,0,0,0), assuming Top First (the usual case), or GreedyHMA(1,0,1,0,1,0,0,0)(if you'd like to use the vertical filter (recommended, and it'll deinterlace any remaining interlaced frames, and smooth a bit as well)), should work pretty well. It performs a 2:2 pulldown (keeping the 25fps).

The reason you might not want to use just the Smart Deinterlace filter is that it will give you all kinds of ghosting, and won't do a complete job. As for h00z's recommendation about using Warp Sharp and the Smart Smoother, that's up to you, but, as he said, it will slow things way down. My own feeling is that with a high enough bit rate, you don't need them, as you'll get less crud outside the lines needing repair. He's putting 4 episodes/CD and is bit rate constrained (depending on resolution-couldn't tell from the screenshot, though it looks like 512x384), and I believe they probably do help quite a bit. Personally, I put 3/CD at 512x384, and don't use them. You plan on putting 2/CD but at 640x480, and when you have episodes with lots of action, your bit rate may not even be high enough, and you may wish to use the filters. The only way to know for sure is to experiment, as we all have done. Whatever works best for you is what's best.

[neo]

first time I post in a year...
but I really want to be sure: is ivtc-ing an anime a good thing or just a mean to reduce space. For a movie I am sure that every cam shoot usually at 24 fps and every conventionnal theater is designed to project 24fps movies made thanx to these cams.
but for animes, this is not the case since they are drawn then shoot.
I heard somwhere that to have a good efficiency and productivity for TV series, japanese made only 15 draws for one second. It seems low but the eye can't see the difference. Of course try a 15 fps output with TMPG and you will see how bad it looks. If that is true I think that they have a sort of telecine for animes 15->30 fps. That would mean that on hi-mo scenes all frames would be interlaced to preserve continuity.
yes it can be bullshit, but just try to run ivtc with the opening of Evangelion and look at the sequence with the blue sky, the boy's head
and the two girls' body in dark scrolling aroud him. you will notice in TMPG that all these frames are interlaced. how can you perform IVTC with every frame depending of her neighboors. you can't without breaking the fluidity of the move = choppy avi
It is not big deal cause you won't notice it unless you really pay attention on scrolling scenes.
Last year, I had the same problem with cow-boy bebop but I kept IVTC since with old methods I couldn't apply deinterlace filters. Believe me, with the best settings I couldn't get rid of all interlaced frames, with hi-mo they came back. But I can't say it was choppy.
IMO ivtc is a bad thing if you want high quality but it is good if you just want to pack your cd with the max number of avis. Finally, the real thing which gets rid of interlacement is deinterlacement filters. I don't use smart deinterlace (never used it, maybe I make a huge mistake), I prefer to use TMPG deinterlace (double) without IVTC.

Maybe I am really wrong, that's why I am posting. What do u think?

[neo]

I have problems with encoding those *$ù*ing fades.
General quality is good,
but when there is a fade effect between two sequences, it gets really ugly. Again it is from the opening of evangelion where you have a lot of these effects at the beginning.
does someone know how to get rid of this?

Another problem is what I call the ghost problem. Imagine you have an uniform blue background, and a litlle dark object moves from the left to the right. Usually, you get a trail of little black pixels behind the object, what I call his ghost. i found that it was due to the interpolation between keyframes, after each keyframe the ghost effect is reset. So to minimize it, I should put a small time between keyframes which I, of course, don't want. Do you have a suggestion?

here are my settings:VirtualDub.audio.SetSource(0);
VirtualDub.audio.SetMode(0);
VirtualDub.audio.SetInterleave(1,500,500,1,0);
VirtualDub.audio.SetClipMode(1,1);
VirtualDub.audio.SetConversion(0,0,0,0,0);
VirtualDub.audio.SetVolume();
VirtualDub.audio.SetCompression();
VirtualDub.audio2.SetSource(0);
VirtualDub.video.SetDepth(24,24);
VirtualDub.video.SetMode(3);
VirtualDub.video.SetFrameRate(0,1);
VirtualDub.video.SetIVTC(0,0,-1,0);
VirtualDub.video.SetRange(0,0);
VirtualDub.video.SetDivX(956,10);
VirtualDub.video.SetQualityControl(2,16,30,50);
VirtualDub.video.SetMotionDetection(8,14,300,300);
VirtualDub.video.SetCrispness(20,0);
VirtualDub.video.SpaceKF(30);
VirtualDub.video.InternalSCD(100);
VirtualDub.video.SetMinKBPS(320);
VirtualDub.video.SetCurveFile("F:\\evavob\\hue.stats");
VirtualDub.video.SetCurveMcFactor(0);
VirtualDub.video.SetCurveCompression(0,2);
VirtualDub.video.SetCurveFilter(250,3000);
VirtualDub.video.SetCurveCredits(0,250);
VirtualDub.video.SetLumaCorrectionAmp(1,10,30);
VirtualDub.video.SetCurveRedist(1);
// VirtualDub.video.CalcCurveCompression();
VirtualDub.video.SetCompLevelsMain(2,7);
VirtualDub.video.SetCompLevelsA(300,3,16);
VirtualDub.video.SetCompLevelsB(300,4,16);
VirtualDub.video.SetCompLevelsC(300,5,16);
VirtualDub.video.SetCompLevelsD(300,6,16);
VirtualDub.video.SetCompLevelsE(300,7,16);
VirtualDub.video.SetCompLevelK(2,31);
VirtualDub.video.SetBitsReservoir(0,35,27,70,45,0);
VirtualDub.video.SetLowBrCorrection(1,1);
VirtualDub.video.NoAVIOutput(0);
VirtualDub.video.GenStats("",0);
VirtualDub.video.SetEncodingControl("");
VirtualDub.video.filters.Clear();
VirtualDub.video.filters.Add("smart resize (1.1)");
VirtualDub.video.filters.instance[0].Config(640,480,0,3,616,480,0x000000,"0 6 7 0 560 240 100 0 1 1 1 4 4 2 2");
VirtualDub.video.filters.Add("smart smoother (1.1)");
VirtualDub.video.filters.instance[1].Config(5, 25, 0);
VirtualDub.video.filters.Add("temporal smoother");
VirtualDub.video.filters.instance[2].Config(3);
VirtualDub.video.filters.Add("smart smoother (1.1)");
VirtualDub.video.filters.instance[3].Config(5, 25, 0);
VirtualDub.video.filters.Add("hue/saturation/intensity (1.0)");
VirtualDub.video.filters.instance[4].Config(-2, 1, 0, 42, 63);
VirtualDub.subset.Delete();
VirtualDub.brc.Set( 0, 645 );
VirtualDub.brc.Set( 1, 1 );
VirtualDub.brc.Set( 2, 128 );
VirtualDub.brc.Set( 3, 0 );
VirtualDub.brc.Set( 4, 1 );


thanx for any help

[ssjkakaroto]

i have the exact same problem with those trails and i have no idea how to get rid of them, if you see the ending of Final Fantasy Unlimited encoded by soldats you'll see that they can get rid of them really fast, i've asked how they do that on their forum but so far no response. any help would be appreciated.

[Dave]

Just out of interest, Hooz, how well do these filters work when used in conjunction with, say, a Hollywood movie whose source is a little poor?

[nephilim]

Wow, neat process you got there, h00z. Though I ran into a problem with (I assume cb1 is Cowboy Bebop) the same Anime. One thing to note is that in most Anime the field order is top first, hence GreedyHMA(1,0,3,0,0,0,0,0) is accurate. That's for those who cut, paste, and ask questions later, hehe.
How well overall does TMPEG IVTC compare to the myriad AVISynth modules? I'd think if you could somehow skip the VFAPI the speed increase would be well worth it. (I'm lazy. Skimming 40,000 frames an episode really kills my brain, so I'm talking pure automatic IVTC quality)
Serious kudos, tho. Now I can finally do Lain some justice.

[manono]

You don't have to make a VFAPI. You can do it using just AviSynth. But if you want to use the Smart Smoother and Warp Sharp, you'll have to use Full Processing in Nandub and insert the filters, as there is as yet no AviSynth versions of them, which will slow things down some, but it'll still be faster than frameserving with a VFAPI. However (from the GreedyHMA readme):

3 - Deinterlace (Force Video) but with frame dropping. Use for NTSC video
if you still want to decimate. With video source there really are no
proper fields to drop, but this will drop the ones looking most like
dupes.

4 - Auto Pulldown with frame dropping. The best (most automatic) setting.
Works in most cases, at least if I get all the bugs out.

5 - Pulldown only (Force Film) with frame dropping. This will give the best
results if you have 100% properly mastered NTSC film source with no
video sections and not too many edits. It can adjust for most scene
changes and changes in pulldown cadence, but not mixed up fields.

So I don't use the 3 setting-I prefer GreedyHMA(1,0,4,0,1,0,0,0), and the vertical filter will remove any remaining interlaced frames. It works.

[ssjkakaroto]

manono, does the vertical filter really removes frames? cuz the greedyhma readme only says it removes noise and deinterlacing artifacts

[manono]

Hi-sorry-I misspoke myself-I should have said it removes some noise and deinterlaces. I have found that using the 4 setting by itself may leave behind a very few interlaced frames in difficult source material such as Lain (the 5 setting leaves behind lots), so I put in the vertical filter. You don't necessarily have to use Smart Smoother and Warp Sharp if your bit rate is high enough (say, 3 episodes/CD-R at 512x384), but experiment and decide for yourself.

[killerSheep]

I disagree. Often, the DVD source itself has a certain amount of noise in it that these filters get rid of. The artist drew each frame using solid, non-noisy color, and to reproduce this with DivX, any DVD inconsistencies must be ironed out first. Look higher in the thread at the pics from h00z for Sakura if you don't believe me.

[manono]

Hi- I'd say it's a sort of yes and no kind of situation. And you're certainly right in saying the filters can remove the crud, and I have looked at h00z's pics several times (and I did cover myself by using the word "necessarily"). I've experimented with the filters myself. But for a couple of reasons I don't generally use them:

1. There's a time vs. benefit at work here. Encoding an episode that may take an hour just using GreedyHMA, may take up to 2.5 hours with the filters applied.

2. The more important reason is that I don't watch my anime one frame at a time. Sure you can see the stuff if you stop the action and look closely at the monitor. But you just can't see it at full speed. Even more so if you have a vid card with video out and watch on TV (as I do).

So- if you're encoding all 26 episodes of Cowboy Bebop (for example), the extra time involved using the Smart Smoother and Warp Sharp isn't worth it to me, for what I think is a whole lot of extra time, with not much extra benefit. But that's just my opinion, and I understand your point of view.

[killerSheep]

Yeah, you are right it takes alot of time. I guess it's kindof a personal decision. When I watch an episode encoded using the filters, I can tell the difference between filtered and unfiltered. The colors just seem more uniform and "solid" looking.

But, for alot of people, the time factor involved simply isn't worth it. I guess I attribute my methods to a certain amount of perfectionism I've never been able to get rid of :p

Either way you'll get nice quality overall if you have the correct SBC settings.

[jrmillerUT]

Can u tell the filters were applied when viewed with your monitor at 640x480 and the Divx avi at fullscreen while being comfortably away from the monitor? I've noticed that filters help my compressability moreso than changing the way the movies look from far away.

[dragonlz]

A lot of the Anime DVDs i encoded still look partially interlaced when I IVTC with GreedyHMA(0,0,3,0,0,0,0,0) and even if I turn the verticle filter on(0,0,3,0,1,0,0,0) it wouldn't make much of a difference. Deinterlacing after that doesn't help that much. I don't know if anyone else had this problem, but I found a really simple solution to it.

When I use
InverseTelecine(40,10,15)
then,
SmartDeinterlace(2,15,true,true,true)

it gets rid of all of the interlaced artifacts.
Then apply those filters h00z recommended if you feel it's necessary.

Hope it helps someone.

[ssjkakaroto]

dragonlz try using GreedyHMA(1,0,3,0,1,0,0,0) and see if that helps, GreedyHMA(0,0,3,0,1,0,0,0) didn't work with Dual

[dragonlz]

I tried that too, didn't work for a lot of the DVDs(eg. Initial D, Inuyasha, etc). :(
but i'm glad the combination of IVTC and SmartDeinterlace worked.

[nephilim]

Yeah, that works fine for most Anime. In my experience different titles tend to need radically different procedures. While GreedyHMA works better for noisy, and (amazingly enough) high motion titles, a smooth source with fine borders like Lain or your more recent titles will react better to the standard IVTC AVISynth plugin. For filesize most will benefit from a smoother to get rid of 29.97FPS jumps and other deinterlacing woes.
Greedy will still leave jagged artifacts on fine borders no matter what you set for Vertical Filter, sometimes when they are perfectly still. There can be a lessening of the effect with bilinear resizing, and a compressibility check can tell you how much.
As with everything, the best thing you can do is play around with settings and filters until you develop your own set of rules to rip by. (No matter how much it kills your head)

[OUTPinged_]

>"IMO ivtc is a bad thing if you want high quality but it is good if >you just want to pack your cd with the max number of avis"

that was the most stupid thing i have ever heard

neo, you have to reconsider you way of anime ripping.

and more: wasting bits is bad even if you have a plenty of space.

just keep the damn vobs in that case

[dragonlz]

@OUTPinged_

There is a situation where you don't want to IVTC: when the IVTC filters don't take out the interlaced frames(meaning they are taking out the correct ones). Believe it or not, that happens quite often with anime:rolleyes:

But yeah other than that there's no point to NOT IVTC.

[poopity poop]

Check out my site:
http://web.syr.edu/~tjmyers/animefilters.html
It explains a very good filter set to use to create high quality DVD rips of anime

I've since then realized the settings quoted on there are a little too extreme and smooth out the beautiful background of anime and are WAY too high for movie textures.

For anime use 5-7 diameter and 20-25 STR. Also I use warpsharp at a setting of 40 for anime.

Like always... there is NOT JUST ONE GOOD OVERALL SETTING! You must look at what the filter does and decide for yourself what looks good, what doesn't what can I change, how does this filter work, what if I do this, etc?

also if anyone accually reads down this far...MY credentials are that I've created the highest quality SBC Kenshin encodes on DALnet to date for the size of ~170megs/episode. Sorry ruri_chan, your encodes just don't stack up. And if you recall, unless there was someone who made the exact same discovery as I did I was the first one to figure out the fitler combination of smartsmoother and warpsharp noted on this old nandub thread:
http://rilanparty.com/vbb/showthread.php?s=&threadid=7867


Anyway I just wanted to say to all my Kenshin fans on DALnet, I'm REALLY REALLY sorry for not encoding the rest of the series when they come out, I'm not bringing my comtpuer back next semester, get them while you can before this tuesday(DEC 18th) on my FTP and fserves, after that you'll accually have to trade for thatm *gasp* !

[poopity poop]

I believe I must apologize for my last post, I ddidn't want to see full of myself, even though I accually am. Its just that I'm not going to have my comtpuer next semester, I'm not going to encode, etc. Its jsut rough on me and I just wanted to see that I made my mark on the encoding world or whatever..prolly sounds corny. Whatever, enjoy ripping you guys, and don't distribute and SHIT, or I'll be upset :) don't release until its perfected

[neo]

>that was the most stupid thing i have ever heard

it is so funny, to hear people answering you like you are shit without a good damn argumentation, I have explained my problem with solid FACTS and some suppositions and instead of helping me, you say I suck?
why do you post if it is only to say that your way is the best without any argument.
I am sure you only understood or readed the sentence you quote.
I said it: I am not sure of what I am saying but since I had very bad results with most of animes and I couldn't understand how it could be possible to ivtc such sequences, I tried to make some theories that I posted before on this thread.
since nobody answered WITH arguments, I believe I am right.
if you don't agree, fine, so tell me what sequence to input in tmpg to deinterlace the american release of evangelion.
or just tell me why you always have to run a deinterlace filter after ivtc.
if you can't just shut up and pass your way, man.
Again, I would be really glad to ivtc animes (from tv not theaters)just like a movie but I don't think it is possible.

and about wasting bits, how many bits do you gain with ivtc?
did you try to see the difference?

and thanx dragonlz

[takeru]

hey, i got a good challenge for you people. lets see who can encode the opening title sequence for fushigi yuugi with 64-320 vbr mp3 audio, 512x384, 23.976fps, AND maintain a filesize of 25megs or less. AND, the scene with tamahome and hotohori showing off their moves together MUST be unblocky! :D this is not as easy it may seem either.

[dragonlz]

I think it's possible, how long is it? I think I managed that for flame of recca opening scene, which contains more action.

[poopity poop]

I've never seen the opening but I'm sure it has be to a lot of action. I think the best way to do this would be to use MM4, take the rally high action scene and do multiple encodes on it to just take out the blockyness and then aplice it in, of course do the audio seperately. Also use Gknot to edit the stats file to maximize quality, watch the bitrate as it encodes to maek sure its doing what you want it to do. I've found that at motion 299 a bitrate of 3500 has no blockyness. So just make sure not to use curve comrpession so that at that motion its geting at least 3500.
Also 25 megs allows you to prolly set the max bitrate to around 4000-5000, with an average bitrate above 1100, so that makes it pretty easier. I think a harder challenge is to fit it to 20megs. :) I just was playing around with VCD encoding and I did a 9 pass vbr to the first opening of Kenshin at a resolution of 352x240 and during the maximum motion there was no blockyness, and the file was around 24 megs. OF course the resolution was small, and this was mpg-1, but it is possible

[OUTPinged_]

2all mr. smartasses about my prev. post: 1.read the damn quote. 2.read mine answer. 3.shut up.

neo: you wrote "dont use ivtc if you want quality". that _was_ stupid.

no IVTC means 30fps with some 24fps frames blended/jaggy. they are taking 10-30% bitrate more that well iVTC'ed anime. and all the 24fps motion is f*cked.

poopity: do you have any idea why warpsharp filter wont work on one of mine pc's? it just freezes the system instantly do death :((( (windows98 sucks)


2all: all mixed (FILM/NTSC) dvd sources should be IVTC'ed if there is mostly FILM going. if tmpg isnt picking right frames for you (and you _can_ manually), just set it up correctly. read the damn giude at least once. NTSC sequences wont ivtc, so you need to apply deinterlace filter anyway. you can make tmpg just field drop those frames, you can set the deinterlace filter for the whole movie, that will f*ck all the other frames but you can be sure no interlacing slips by. the quality of filter will determine how much of edges will be f*cked in that case. you can deinterlace that CG separately, for better image use area based deinterlace, for fluid motion use 30>120>24fps decimation. that can be tiresome tho. alot of QC is needed. anyway dont show youself smarter that anyone else. all of these questions were discussed alot long ago. ppl argued and came to tmpg+area based deinterlace combo as giving best quality/effort ratio. do what you want man. noone really cares :P unless you will make a full QC for each episode you wont get anywhere near good release groups quality.

and making 50minutes/700mb is a bit waste in most cases, unless you have a 1ghz+ pc.

[takeru]

i agree with outpinged. i have no idea dragonlz why you would say ivtc is a waste in anime. it saves a good amount of bits for quality overall. in my own method i use smart smoother to blend out leftover interlace lines in hard to ivtc anime's, and there is a good amount of these unfortunately. so far only hand maid may and tylor have gotten excellent results from ivtc. less than 1% of frames left per ep with any interlacing, if any even. anyone know of a place that would host big files? i wanna post some samples but can't find anywhere willing to do it for free...

just upgraded from a p2 400 to an athlon xp 1800+, damn i am so happy :D

[poopity poop]

OUTPinged_ --> Not sure why warpsharp doens't work, that's a little strange. Its worked in win98, and winXP. maybe you have a bad order of filters that vdub/ndub doesn't like. Also make sure that you are not croping on the warpsharp filter; only crop on the resize filter. I've gotten some crashes/messed up files when I've cropped on anything but. If those things don;t work just upgrade to XP, its working fine no problems

And I really haven't been following the posts about ivtcing and such...but whoever said ivtcing is a waste on anime is a fool. There are two ways you can get rid of interlacing: ivtc(the best and most proper) or deinterlace filter(discard field 1 or 2). Blending or duplicating is looks ridiculous. But you can accually get good deinterlacing results if you discard field 1 or 2...try it I'm serious

of course you have to watch your frame rate, because ivtcing in tmpeg gives you 24fps...which is optimal

[dragonlz]

i have no idea dragonlz why you would say ivtc is a waste in anime.
Look, please read my posts carefully before saying dragonlz said this, dragonlz said that. I _never_ said ivtc is a waste in anime or otherwise. Actually I think you'd be stupid not to use it _IF_ it works with that particular anime that you are encoding.
I'm just simply stating that sometimes IVTC does not work very well, for some of the anime I ripped, it took out the normal frames and left the interlaced ones. How well does that work? not very apparently.
Think about it, if you are left with the interlaced frames, how the hell are you going to get "good" quality results? even if you de-interlace it afterwards, it'll still leave half blurred results. In that case, and only that case, it'd be dumb to keep using IVTC. It's simply better to keep all of the frames instead of discarding the progressive ones.

[takeru]

poopity poop: how is blending rediculous? even after proper ivtc, theres still most likely some frames that still have interlace lines. most animes you can't fully get rid of interlace lines. and yes, i do switch between odd and even fields for better results also if you ever read my method in the past. i rely mainly on the automatic deinterlace feature of tmpg and follow up with manual to do this. i blend to get rid of the extra lines ivtc'ing and deinterlacing leaves behind.

note: i meant blend feature in smart deinterlace filter in nandub, not smart smoother :P man i have to sleep more.

[manono]

You still following this thread? Have you tried the 2D Cleaner yet (anyone else experimented with it)? I think it's slightly better than Smart Smoother, as well as being slightly faster (at least with radius 1-it really bogs down with higher radii, but then so does Smart Smoother). For what it's worth, Nicky seems to like it better.

[poopity poop]

I just wanted to say:

DAMN this is a large thread for doom9 ha ha

takeru --> I agree. I take my statment back.. After reading some different things and having some prpblems with SVCD encoding, and seeing some gay interlaced frames sneak by on my own encodes. If I had to do it again, I would have used an area based interalcing filter to get rid of that. Its so annoying to see something you've worked so hard on and an interlaced frames sneaks by real fast ha ha

[triffid]

Yep, hate those little interlaced frames...i have found not one anime that doesnt have at least a few scenes where a guy's arm is interlaced, mouth movements when talking are interlaced, etc. I just said "Screw it" and run Gunnar Thalin's area based deinterlace filter on everything. Here's my filter list:
-Area based deinterlace
-resize bilinear to 512/384 (i never tried smart resize, but a greater encoder than I says it's not worth the extra encode time)
-smart smoother (forget the settings, but they're standard for anime ripping, probly the same as Poop's
-warp sharp at 40
-vobsub

With those filters the video that goes into the encoder is near flawless, often better than the DVD itself. I've noticed that US Manga Corps especially does a poor job on their DVD quality, leaving in quite a bit of grainyness.

This coupled with my oddball settings produce very good quality episodes at small filesizes, depending on the amount of action in the original video...for example my encode of Trigun DVD1, in which the first episode was 200MB while the second ep was 115MB, both excellent quality in both low and hi motion.

This thread helped me out a bit, pushed my encodes to higher levels. Thanks a lot, esp. Poop. If anyone wants help encoding anime I'm always on DALnet and willing to help.

PS Are there any great reasons to use smart resize instead of the built-in vdub resize filter? If there is I wana know em. Thanks :D

[poopity poop]

smart resize is just has more options for aspect ratio's and other things. But I just grab my calculator and figure AR->resolution conversions by hand its faster.

Also watch out for the 512x384 resolution...if you go below 150 megs that's an awful high resolution for that file size. I would say if you go below 150 use 400x300.

for a 23 minute episode:
30-60megs--> 320x240
61-150megs--> 400x300
150+ --> 512x384

Also be careful of warpsharp being too high. I've since then brought the setting down to like 20-25. And smart smoother to the defualt settings of 5/25 or sometimes 5/20. You want to remove the crap with out ruining the hand painted backgrounds :)

[triffid]

thanks for the reply poop

anyway i agree that in general under 150mb is too small for 512x384, however:

in my encodes i set min bitrate around 350-375 among other settings. On very few episodes does it dip below 150, and when it does it is, im guessing, due to the compressability of the action in that episode. I dont force it to go to that size (in fact, i use the bitrate calculator to set 200MB X number of eps on the DVD)

The max BR is near always over 1050 on my encodes. Therefore im guessing that if it needed the extra filesize, it woulda used it, but didnt.

for the record my 115MB trigun ep has no noticable differences in quality from the other episodes.

Summary: I agree with you in general, but im not gona fix something that already works. :)

thanks for responding :)
-triffid

[OUTPinged_]

You guys still didnt get the point.

ok, we were talking about "high quality" ripping :/

if youd ask some ppl with a fast downlinks and huge collections of stuff, what will they tell ya:

300meg/ep = excellent quality possible.
230meg/ep = very good/good quality
180meg/ep = ok/tolerable quality.
145meg/ep = crap, wont take that.
xxxmeg/ep = crap if was ripped by a dumbass :P


you can always bring the resolution down, but it doesnt get you that much spare bits, and 145m eps will suck anyway, the picture wil be way too blurred or the res will be to low. first sucks, second sucks too, remember sorrow 200mb 320x encodes :P

what about IVTC: use tmpeg damnit. avisynth plugins are good only if you dont care about quality much. spare a bit of your time lol.

as for "impossible to ivtc" titles: i have IVTC'ed Lain with tmpeg on a autoamtic settings of tmpeg and no interlacing was left. just dont act stupid ppl. you get avisynth's speed, you get a crappy picture. period.

[triffid]

filesize has quite a bit less to do with quality than you think when you consider the capabilities of nandub's SBC options.
The quality of my current encode is coming out at around 180MB per ep, but looks better than any 250MB eps that I have seen.

This is a mistake that i see a lot of people make. They automatically assume that if something is small it is low quality, and if something is big it is high quality. All the filesize has to do with is bitrate and keyframes. Bottom line: filesize is useful as an indicator of quality in encodes done with Flask, not nearly as much in Nandub.

PS I have also IVTC'ed Lain with TMPG's automatic settings and it left interlacing.
I'd be glad to know exactly how you did it.

(BTW in case you didn't know, I agree with you that anime needs to be IVTC'ed)

[ssjkakaroto]

hi, has anyone tried Decomb for IVTC yet? it seems it works really well with anime but i haven't tried it myself.

[Hanty]

Outpinged >

I would argue that size is a direct reflection of encode quality.
I happened to talk with a great encoder ripping NTHT DVD1 for a DAL channel. (Nope, not going to mention the group, not important)

He used a bitrate of 1130 and the filesize came out as 126 MB.
At a bitrate of 2000 the filesize came out as 126MB.
Further at a bitrate of 3000 the filesize came out as 126MB.

BUT there was less artifacting and other crap in the higher bitrate versions.

Granted, NTHT is very dark and thus comes out undersized
(Same goes for Boogiepop) but nevertheless...
There are always exceptions, newbies especially tend to look at
filesizes and equate a large one with good quality. Big mistake.

[poopity poop]

I agree with hanty.

Take a look at my Kenshin encodes especially the later ones and you will see that if you double the file size you don't really get better quality.

If an episode is encodes properly:
4 episodes/CD(165-180megs/episode) = 95% DVD quality
2 episodes/CD(330-360megs/episode) = 99% DVD quality

I don't soppose many people will disagree with me. (...of course the only one's that do respond, will be the one's that do disagree...)

[dragonlz]

Damn, this thread is still going..

I'd agree to poopity poop on the size issue. I've done at least 20 anime series and 4 eps/CD seems to work really well. I'd say it's around DVD quality. Ofcourse 2 eps/CD would produce better quality but it's not that noticable and you are watching the show, not every pixel of a single frame. If you are doing a lot of rips.. you also have to consider the fact if it's worth it to use twice as many CDR's to store them.
Honestly, when you have a collection > 300 CDRs of anime. You'll have to find a good balance between size and quality.

as for "impossible to ivtc" titles: i have IVTC'ed Lain with tmpeg on a autoamtic settings of tmpeg and no interlacing was left. just dont act stupid ppl. you get avisynth's speed, you get a crappy picture. period.
You should think before posting. Do you think we are dumb enough not to try TMPG before saying it's not possible to IVTC? Ofcourse not, there are apparently some anime titles that you haven't encountered before. Maybe more experience will help to explain your ignorant comments. Also, avisynth is good because it's fast, but fast does not mean worse quality. My experiences lead me to believe avisynth is better at IVTCing than TMPG.

[OUTPinged_]

ok, here is some thoughts for a dumb ppl who still dont get it.

IF YOU DIDNT DO SOMETHING WRONG, your encode will always look better when you will spend more bits into it

more bits equals to:

1. lower DRFs. if someone will argue on that, read the manual you dumbass.

2. higher resolution. now we will see a smart people who know better than everyone else that anime should be encoded in 320x so that ppl buy dvd and support bla bla bla.

3. less source blurring. you can blur different, you can use a XXX filter, you can use a crisness modulation, it doesnt matter. the filesize will get lower in general.

so you cant tell me a 320x 80mb encode will look better than 720x 400mb encode.

for those smart ppl who are comparing 180 to 230mb encodes come the next part.

more bits doesnt equal to:

1. better encode. cause some encodes are done by some smartasses who dont do IVTC and deinterlace stuff with blurring. they make encodes with 256kbit audio which is not dynamically compressed cause "it is done so on dvd and i want it to be of "100% dvd quality"


2. less noise impact. ok, you can keep noise and kill 50% of your bits for its sake. that wont help your encode. temp.smooth it, spart.smooth it, do whatever you cant to get rid of that shit.

3. less interlacing. some dvds are just impossible to reconstruct fields. and even deinterlacing the whole thing leaves the picture dirty. its your choise what you will do, but more bits just wont help you at all. interlacing sux. period.

to the guy who said avisynth's ivtc is better that TMPEG's:

you remind me of ppl who were telling me flask is the best. tmpeg's routines are:

1. processing stuff in RGB. (if you cant figure it out why it matters, just dont comment)
2. optimised for a correct IVTC, not "speed"
3. they are not perfect and terribly slow too. still tmpg is the only tool that will make you sure that no FILM parts will be deinterlaced.


2eps/cd vs 4eps/cd. ha-ha-ha. buy yourself a glasses man. :P


poopity: raise the resolution then :) you do compres stuff, ne? same as with mp3. some things sound good at 128kb, some dont. but noone is shouting "make everything in 128k, cause i cant hear the difference on my 2$ headphones" well excluding the smartasses of course.

btw alot of ppl cant encode an ep properly, so they have to use more bits. so no 4ep/cd encodes for them :-E

Hanty : size vs quality relation isnt direct. i didnt write that. "artifacting and crap" dont relate that matter. btw, bitrate did not change anything in there, you ahve got the same file, ne? put codec errors aside.

ssj: tried it. 1.7 sux. 2.2 is ok to use. still not as good as tmpg. but if you dont have time for it, decomb2.2 is a way to go. interpolation is F-ed tho :(

triffid:

>filesize has quite a bit less to do with quality than you think
no, it has exact impact on it as i think. see earlyer crap.

>The quality of my current encode is coming out at around 180MB per >ep, but looks better than any 250MB eps that I have seen
buy a better glasses then. dont make people laugh at you. and dont post stupid things.

>They automatically assume that if something is small it is low >quality, and if something is big it is high quality.
not automatically. "xxxmeg/ep = crap if was ripped by a dumbass", remember? and there are alot of these. ph3ar =)

>PS I have also IVTC'ed Lain with TMPG's automatic settings and it >left interlacing.
>I'd be glad to know exactly how you did it.
np man, just use auto deinterlacing optins. check all, correct manually. i am doing lain rips atm (sloowly), so i guess you will be able to check them after some time.

still, you can either deinterlace or blur+decimate. if someone could write a field muxing decimation plugin for a 30fps>deinterlaced smooth 24fps pans, it would be great :(

for smartases: deinterlaced pan doesnt mean smooth :P

poopity again:

>for a 23 minute episode:
>30-60megs--> 320x240
>61-150megs--> 400x300
>150+ --> 512x384
man, dont make such a statements, they are way to varying.

btw, a question to you: do you know a filter which will smooth the effect of xsharpen without blurring the image? ssmoothers dont help :(((

more to triffid:

>for the record my 115MB trigun ep has no noticable differences in >quality from the other episodes.
from the other 115 mb trigun eps? ;) or other eps encoded by you? make youself clear man.

just kidding

btw people, try out ssmoother HQ. its slow, but it kix.

[Hanty]

Yes, bitrate did change things in there, without affecting filesize.

[OUTPinged_]

that is funny then. bitrate just controls BR size. absolutely nothing else. the filesize was left unchanges so i guess it was 2x drf all the way. have anything except codec errors changed?

try encoding that thing with antishit on and lower (1130) bitrate.

[dragonlz]

@OUTPINGED_

I'm sorry to say, after reading your post, all i have to say is HUH? 90% of your post is total bs. i'll just quote a few lines from your post:
to the guy who said avisynth's ivtc is better that TMPEG's:

you remind me of ppl who were telling me flask is the best. tmpeg's routines are:
1. processing stuff in RGB. (if you cant figure it out why it matters, just dont comment)
2. optimised for a correct IVTC, not "speed"
3. they are not perfect and terribly slow too. still tmpg is the only tool that will make you sure that no FILM parts will be deinterlaced.
1. processing stuff in RGB doesn't make it "better quality", it all depends on how the algorithm is defined.
2. what the hell is "correct IVTC"? If you mean perfect IVTC, it doesn't exist. How do you know TMPG is optimized for "correctness" and avisynth is optimized for speed? First of all avisynth has several different IVTC plugins that you've never used in your entire life.
3. again no proof, just bs

If I remind you of one of those people that told you flask was the best, then you remind me of one of those people who was asking "what is a divx?" 'nuff said.

[takeru]

yes, tmpeg doesn't always do ivtc perfectly. a few anime's i encoded tmpeg got confused when there were too many parallel lines in the scene, and kept switching back and forth between even and odd fields each frame even though the scene never even needed any field changes. manually fixing these parts wasn't fun...

[OUTPinged_]

>after reading your post, all i have to say is HUH?

yeah, you have to say that much :/

>. processing stuff in RGB doesn't make it "better quality",
read your quote of my post :) "if you cant figure it out why it matters, just dont comment"

>. what the hell is "correct IVTC"?
its when a tool can make you a correctly ivtced FILM stream. have you tried to use any AVS plugin on plain FILM source, just for testing?

"incorrect" there will mean when plugin is selecting a wrong field, altho right field is available nearby.

>"? If you mean perfect IVTC, it doesn't exist
rtfm man. read youre posts haha. perfect ivtc = all the frames are restored. you can make it easy if source is 100% FILM. FF is a inverse pulldown process too if you dont know that.

and dont tell me i wrote non-telecined sources can be restored with ivtc completely. i'm not 100years old yet.

>How do you know TMPG is optimized for "correctness"
i asked the developers and read what they wrote. tmpegs routines were written with being as good at guessing which filed to use as possible. it still has some speed optimise afaik.

>and avisynth is optimized for speed?
first, there are several iVTC DLLs for avs. wizard's is optimized for fastest pure FILM source (with small NTSC deinterlacing routine) restoring. greedy is optimised for real time hdtv ivtc in a first place. decomb is in a development but it is selecting the same fields as wizard's dll and applying area based deinterlace on top of "bad" frames.

>3. again no proof, just bs

./me hands dragonlz a cookie

good guess man, all this thread is a big bs, everyone is comfy with his tools and is defending them at no cost.

and what exactly is the bs in that?

>not perfect and terribly slow too
cheers to takeru - check the "not perfect" part. and if mr.dragonlz is considering tmpg fast i wonder what kind of hw he has :/


>tmpg is the only tool that will make you sure that no
>FILM parts will be deinterlaced

ok, here comes more:

consider another tool which IVTC's FILM parts ok (except FF, it doesnt catch pattern changes :), can calc threshold as good and isnt coded for SGi workstation. filter chains dont count.


and more about decomb: there could be written a tool which is better than tmpg. tmpg is still missig frame candidates on hard parts. decomb seems like a good candidate, but they should at least fix smart deinterlacing and make it more configurable.

takeru: try to use avs dlls, they would suck at that even more. the program really cant get a lok at the whole picture, it just calcs nearby pixels and that case cant be helped. you can use blending tho.

[poopity poop]

Pooping feels nice :)

[manono]

and more about decomb: there could be written a tool which is better than tmpg. tmpg is still missig frame candidates on hard parts. decomb seems like a good candidate, but they should at least fix smart deinterlacing and make it more configurable.

As you mentioned-it is still in development. When finished, Decomb should be better than TMPGEnc and a heck of a lot faster. But the deinterlacing algo is and has been configureable. If you find (as I have) that the default setting (15 ATM) is sometimes too strong, and deinterlaces some good frames turning them to crap, then you might try 30, which I have found for anime to sometimes be better. The help file tells how to configure the FieldDeinterlace.

Since you seem to know what you're talking about (if you strip out the flame portions of your posts), feedback from your tests would be much appreciated down in the AviSynth Forum.

Since we're all in this for the same thing I think (a better and easier and faster IVTC-Deinterlace for anime), then lets keep this discussion on the higher more polite level it deserves (can't believe I just said that-I love a flame war as much as the next guy, but there are too many knowledgeable guys contributing to this thread to get sidetracked by non-issues).

[OUTPinged_]

>When finished, Decomb should be better than
>TMPGEnc and a heck of a lot faster.
good news. the current (v2.2) is pretty good at hitting the right fields, still not as good as tmpg but close. but it still misses interlacing with low (bare to none) luma difference. guess there isnt much to do about anyway.

the smart deinterlacing is not working "right" in interpolation mode, but it seems mr. Graft knows that so we will hope it will be fixed.

>>decomb seems like a good candidate, but they should
>>at least fix smart deinterlacing and make it more configurable.
>But the deinterlacing algo is and has been configureable
field selection is not. and i would like to tweak thresh of "giving up" on some titles. the default is set wisely, but it seems no tool can be as perfect as the dvd remastering guy's stupidity.

>. If you find (as I have) that the default setting (15 ATM)
>is sometimes too strong, and deinterlaces some good frames
>turning them to crap, then you might try 30, which I have found
>for anime to sometimes be better.

hmm... i didnt get it. to me it felt like lowering thresh would make deinterlaced areas smaller. *checking docs* hmm... *checking more docs* :/

ok, i suck. this means i cant figure out if a frame has less deinterlacing or the motion map denoising works as it should :/

anyway, interpolation dont work in 2.2 and low threshold shouldnt kill my current encode (with blending and medium resolution (608x)). more feedback will be as soon as it'll finish (5h).

and mamono, dont forget flaming dumbasses is fun :) contribute too man :-) the size vs quality part deserves a good laugh.

[manono]

Hi OUTPinged-
You gotta keep checking the AviSynth Forum or his website. He dumped Decomb 2.2 and there have been 2 versions since then. I didn't understand why he dumped it (I don't mean I disagreed, but that the reason was over my head), but the things you mentioned could be it.

So you're not saying that it "over deinterlaces" and turns some good frames bad, right? You're saying that it "under deinterlaces" for your current material and some interlaced frames slip through? Then you're right-lower the threshhold (but maybe check if some good frames turn "bad", i.e., lines turn jagged and maybe the scene will start to "shimmer". He's still tweaking this thing, and is turning out improved versions daily. One thing I discovered while testing recently (which perhaps you already knew) is that sometimes the DVD itself is screwed up, and often you can't blame the IVTC unless you've checked the source.

Heck-even some of your statements go over my head.

dont forget flaming dumbasses is fun

Stop that-these are all honorable people. You know-he developed this thing as a way to improve his video capture encoding (as did Tom Barry (GreedyHMA)), but from the responses to his thread I think he's discovering that there may be more anime encoders than vid cap guys around. So ask him (as I have) to work next on a better encoding-post processing filter for smoothing-denoising to help us clean up our encodes better. I guess you've discovered that Smart Smoother-2D Cleaner combined with Warp Sharp-Xsharpen-Unsharp mask just don't do much in medium to dark areas. I'm getting tired of putting in all this time encoding and finding them not perfect.

[OUTPinged_]

>He dumped Decomb 2.2 and there have been 2 versions since then. I >didn't understand why he dumped it

first, he messed up all the deinterlacing part.

second, sometimes it misses a completely clean field and choses heavily interlaced (and deinterlaces it).

ok, that may look not so awful but if mr. graft is saying its like that, then it should be :(

>So you're not saying that it "over deinterlaces" and turns
>some good frames bad, right? You're saying that it
>"under deinterlaces" for your current material and some
>interlaced frames slip through?
both wrong. that is why it was funny :) stil, my test scene (from GiTS) had some freaky frames (note for dumbasses: i was testing _deinterlacer's_ threshs, not ivtc's) with very smooth gradients between fields. guess i had to choose something more regular for that.

>sometimes the DVD itself is screwed up, and often
>you can't blame the IVTC unless you've checked the source
yeah, i blame deinterlacers in these cases. these idiots in ADV are lazy and cant get a decent progressive master so they use pure interlaced vhs's and are maikng a lame attempts to recover that stuff.

well they are better than me at that, but i aint get paid for my efforts :)

>Stop that-these are all honorable people.
the word dumbasses wasnt used for the developers(in most cases). it was used to label some "smart" rippers in this thread who dont read what they do write after what they write and after they see me commenting their bs.

>but from the responses to his thread I think he's discovering that >there may be more anime encoders than vid cap guys around
wrong, but vid cap guys are happy with current soft afaik. and anime rippers were silent before (a year ago)because they all were smartasses and did anime rips in 352x so they didnt have to ivtc. now their pride of making 80mb 352x episodes is ruined and they are flaming all over the board.

lol i just love to insult that kind of people.

>Smart Smoother-2D Cleaner combined with Warp
>Sharp-Xsharpen-Unsharp mask just don't do much
>in medium to dark areas

so true. the most disappointment for me was a warpsharp, after poop has nearly worshiped it. it sux in long term and high thresh. and you dont need it on a clean stuff to usually.

the lesser evil of them all at the moment seems to be spartialsoftenmmx with (3,4,20,false,false,4,4,6,8) thresholds. at least that does work on animes with not detailed bg. my dnr skills suck so i use the threshs these software developer guys recommend. still that stuff F's up in dark areas too. and it is terribly slow. and sometimes i wonder if it is working at all. i was trying to get the source to rip off the dnr part but didnt succeed. Mathias's page is kinda down and links to old board dont bring me anywhere too.

[manono]

Hi OUTPinged-

did anime rips in 352x so they didnt have to ivtc.

Funny you should say that. Ordinarily I don't download others' work, because it's generally so crappy. But recently I got some, and it's often still what you said. You'd think that people that are putting stuff out there to share would take more care in their rips. If it's video cap-fansub stuff I can almost understand it, but if it's off a DVD, then there's just no excuse. But I don't think any of the people that visit Doom9 are guilty of making that crap, and especially those that have contributed to this thread. Like I said before-these are honorable people trying to make the best possible rips given the still imperfect tools we have to work with (I wasn't referring to the mastering houses before when I made the remark about being honorable, but to the people here).

but vid cap guys are happy with current soft afaik.

I'm not so sure, otherwise why would both DGraft and TBarry release IVTC-deinterlacers to improve the situation, when their primary interest is video caps? Neither realized the importance of their tools for anime rippers until it was brought to their attention. I'm sure there are many vid cap guys also trying to make the best possible encodes, and vid cap encoding presents problems similar to and often greater than ripping anime from DVDs. Blight has recently discovered Decomb (as he did GreedyHMA earlier), and he's a heavy hitter in the vid cap field. Decomb will be incorporated into DVD2SVCD soon and a lot of SVCDs are made for vid caps.


the most disappointment for me was a warpsharp

Well, p p has backed off of the high settings he recommended originally (and I credit him for that). Part of the problem is not only do you have the mosquito noise, block noise, line noise from the original source material (DVD), but the damn encoding adds more of its own, and at the moment there's no post processing filter to clean the stuff up. DGraft realizes the problem, and may tackle it when Decomb gets finalized (that's why I suggested we bug him about it).

It looks as if no one else wants to talk to you anymore. Does it make you wonder why? Look-I didn't see anything obviously wrong about the statement about making good rips with smaller file size (and I'm not talking about 320x240 80MB episodes and neither were they). I think the point was that with proper filtering you can rip to the same resolution and get away with a somewhat smaller file size. Personally, I don't put 4 eps/CD either (I put 3 at 512x384), and I'm inclined to think that all things being equal, 3/CD will look better, but for sharing, the smaller size is faster to transfer. To each his own, but I don't think it's cause for your violently sarcastic responses (you're pretty good at it, but this isn't the place for it). What's the cliche? You can catch more flies with honey than with vinegar? So be nice-invite them back in :)

But things are looking up, and in the not too distant future we should have some better tools to work with. Nice talking to you OUTPinged.

[dragonlz]

OUTpinged_
It was funny reaing your responses. I'm kinda tired of responding to your arrogant bs but here it is anyways.

>>. what the hell is "correct IVTC"?
>its when a tool can make you a correctly ivtced FILM stream. have >you tried to use any AVS plugin on plain FILM source, just for >testing?
>"incorrect" there will mean when plugin is selecting a wrong field, >altho right field is available nearby.
My point is that there's no "correct IVTC" or "incorrect IVTC". The algorithms are always aimed to select the right fields. It can be called "better IVTC" and "worse IVTC" but you can't refer to a sucky IVTC prog. as "incorrect" as you don't know how IVTC algorithms work.

>rtfm man. read youre posts haha. perfect ivtc = all the frames are >restored. you can make it easy if source is 100% FILM. FF is a >inverse pulldown process too if you dont know that.
When I talk about perfect IVTC, i don't mean force film. And when was the last time you've seen a 100% FILM source? Most of them are 97%-99%. IVTC programs do catch most of the telecined parts, but they are not perfect. Selecting the wrong fields, missing the right fields occasionally happen and therefore making the final result not "perfect".

>tmpegs routines were written with being as good at guessing which >filed to use as possible.
right, developers will ofcourse say their stuff is the best. I agree TMPG has a powerful IVTC engine, but I don't think it's the best out there.

>>and avisynth is optimized for speed?
>first, there are several iVTC DLLs for avs. wizard's is optimized >for fastest pure FILM source (with small NTSC deinterlacing routine) >restoring. greedy is optimised for real time hdtv ivtc in a first >place. decomb is in a development but it is selecting the same >fields as wizard's dll and applying area based deinterlace on top >of "bad" frames.
You wrote something very informational but gave answer to my question, avisynth dlls are not just optimized for speed(not that it's a bad thing to have). Decomb is a good example, it's a tad slower than all other tools, but produces better results so people still use it.

>./me hands dragonlz a cookie
/me takes that cookie and shove it up your ass :D

>good guess man, all this thread is a big bs, everyone is comfy with >his tools and is defending them at no cost.
This thread was not a big bs, it became a big bs after you started being stupid. You are the one "defending your tools at no cost" and disregarding facts.

>cheers to takeru - check the "not perfect" part. and if mr.dragonlz >is considering tmpg fast i wonder what kind of hw he has :/
When did I say I considered tmpg fast again? oh by the way, 1.63 Mhz Athlon XP.

[OUTPinged_]

sweet, more dragonlz fun!

>My point is that there's no "correct IVTC" or "incorrect IVTC".
lol man, read youre prev posts

>you can't refer to a sucky IVTC prog. as "incorrect"
>as you don't know how IVTC algorithms work
you are sure better know of what i do man. one more stupid line. i know pretty good as they guess stuff. i can write a sucky one myself.

>When I talk about perfect IVTC, i don't mean force film.
sure, else you would look stupid. ne?

>And when was the last time you've seen a 100% FILM source?
>Most of them are 97%-99%
lol man, you have got me. sure that fbi logo (or the one with ugly winged horse) in the beginning makes big difference.

>I agree TMPG has a powerful IVTC engine, but
>I don't think it's the best out there.
name it. decomb aside, its in early devel stage and not on par with tmpg fiels guessing yet. mr Graft will tell you the same.

>You are the one "defending your tools at no cost"
>and disregarding facts.
i like the way the word "facts" sounds. atm they are just your statements and they arent facts.

i did said tmpg is the best ivtc'ing tool (non-pro one) and you are trying to put me wrong. i dont mind but dont sound stupid and just say "here and there tmpg fails and xxx marks fieds rifgt". be more mature man. you sound like a kid sometimes.

>When did I say I considered tmpg fast again?
man that was ritorical question. and i dont care what stuff do you buy. honest.

ok, back to mamono talk.

>Neither realized the importance of their tools
>for anime rippers until it was brought to their attention
and that is exactly why i am flaming here :)

>Blight has recently discovered Decomb

heh. still decomb is promising. i have encoded a bit with it and it has only minor errors. still there is something wrong with field selection when it cant be sure 100%, and as i will be able to tell whats going on clearly i wont post it to DG.


>why would both DGraft and TBarry release IVTC-deinterlacers
sometimes they have fun with badly mastered dvds too. thats why.

>You can catch more flies with honey than with vinegar?
>So be nice-invite them back in
sure man. i just dont catch shit flies whith you-know-what. that discussion is pointless. i told that already. ie i am just flaming everyone who makes posts with statements.

look, noone is asking everything. they are just stating they are cool. ok, i state they post crap. they dont seem to like it.


>and at the moment there's no post
>processing filter to clean the stuff up.
correct. spatials suck. i was trying to separate luma and chroma threshs but it didnt help. i have asked a sshq devel to make me a tweakable thresh for each pixel "ring" around source pixel. if toying with luma/chroma threshs difference while excluding the radius1 pixels from calc will help, we will have a winner :) i have a "semi two-pass" anime only filter in mind too.

sorry guys, its 3h night and i am getting even more dumb and slow :(

laterz

[dragonlz]

>you are sure better know of what i do man. one more stupid line. i >know pretty good as they guess stuff. i can write a sucky one >myself.
yeah you are right, they are guessing stuff, what's important is how they do it. dumbasses like you can write a sucky one? is it going to be something like if (flicker>prevfram.flicker) { this->dropframe(); }? great algorithm, maybe you can release it for $$$.

>lol man, you have got me. sure that fbi logo (or the one with ugly >winged horse) in the beginning makes big difference.
now you really start to sound stupid, some movies use video for intro and some other parts. don't quote me unless you know wtf i'm talking about.

>name it. decomb aside, its in early devel stage and not on par with >tmpg fiels guessing yet. mr Graft will tell you the same.
greedyHMA, doh!

>>When did I say I considered tmpg fast again?
>man that was ritorical question. and i dont care what stuff do you >buy. honest.
lol, another stupid line from outpinged, i thought u wanted to know what kind of hw I have. what happened now?


>i did said tmpg is the best ivtc'ing tool (non-pro one) and you are >trying to put me wrong. i dont mind but dont sound stupid and just >say "here and there tmpg fails and xxx marks fieds rifgt". be more >mature man. you sound like a kid sometimes.
haha, this is getting kinda funny, you say I sound like a kid. you should read your own posts before clicking on "Submit Reply".

getting back to the issue of ivtc tool. You said tmpg is the best one, IMHO it's not. TMPG did fail on a lot of the rips for me, try it with Love Hina DVDs, you'll see what I mean.

anyways, keeping this up will be a waste of my time. last post from me in this thread.

[Hanty]

Well, then do tell what the best ictv tool is for anime ripping.

[OUTPinged_]

>greedyHMA, doh!
rofl.

man you should at least try the tool before you make statements. greedy is only for realtime captures, it is missing fileds very often.

and if you cant get good results out of tmpg it doesnt mean the tool is bad.


>TMPG did fail on a lot of the rips for me, try it with Love Hina DVDs
wierd, worked like a charm for me. did you use HK ones? tmpg is not "fire and forget" type. mess with it. it gives good results after correction.

ok, back to serious for mamono again.


>and I'm inclined to think that all things being equal,
>3/CD will look better,
at least somebody is watching his rips after making them. whee.

the flaming has started just because some ppl stated 4ep/cd rips are "better". i "disagreed" (ok, stating everyone there is idiot may sound a bit too much but i am tired of seeing ppl to respond with "you suck" to my polite posts. that wasnt the case in that forum, but things dont change over time :(

>but for sharing, the smaller size is faster to transfer.
sure, but if you can afford yourself 180mb eps, you should afford 230mb eps too.

the 50mb rm's were a good thing for a modemmers. they were 1.5 times smaller than similar quality avis. that kind of difference did matter. 33% increase in a filesize isnt that much.

4eps/cd imo should be used when:

1.ripping very compressible series.
2.cut op/ed.
3.ripping of long (52+) ep series, altho that can be doubted.

still, the 33% increase in filesize does matter.

as for ivtc, i'm not a masohist and i dont like messing with tmpg too. still, i dont see anything decent competition on anime ivtc scene except decomb. it has smart deinterlacer built in, that is a big advantage. sometimes threshholds in tmpg should be tweaked a loads of times to get a result :(

ok, as dragonz pointed, that disscussion is over. see you in future threads ppl :)

cheers to dragonlz.

respect to manono.

poop: managed to make warpsharp work good enough so it doesnt make the picture worse. still, its not a ultimate solution and perfomance drop is scary.

[takeru]

hmmm... 3eps/cd, 4eps/cd. actually i think it should just go by how long each episode is after cutting out the opening and ending. say, for 4eps/cd, if each episode was less than 23 minutes, it will fit nicely into one cd with bilinear. or, 3eps/cd with bicubic. but if the episodes are 24 minutes or longer, then its best to just use 3eps/cd to maintain quality. after ivtc, this seems good to me.

[poopity poop]

all this 4CD/ep vs 3 CD/ep won;t matter too much anymroe will it?

DVDs!! I predict encoders will make use of DVDs to release 1gig/episode with both audio's and subtitles WITH interactive menus(I recommend roxio video pack). In fact I'm surprised that adx and other SVD releases DON'T have interactive menus and both audio's It is tough to have both audio tracks with only 795 megs to work with...but that will all change, that will all change

<looks forard to DVDs!>

[jrmillerUT]

@Poopitypoop,

I agree with you totally on the DVD subject. Until prices come down significantly on the DVDR's then it will not be cost effective for plain DVD conversion. I predict that anime encoders will go the route you have suggested and I also see people encoding 2-3 full length movies onto each DVD.

Another Q, when you're using Gknot to generate an ECF only, do you load your .d2v? I guess i'm having kf probs between Gknot and Nandub that i don't usually have.(read below)

@everyone else

This is kinda off topic but i'm having a few problems with my keyframes. When i do a first pass unfiltered in nandub it picks up the scene changes no prob. (alternate method, multiplier of 33) But when i run a first pass with smartsmoother rad 7 strength 30, and a warpsharp of 40, nandub is missing many keyframes.
I looked at the stats file in GKnot and randomly noticed that the Luma values are now much much smoother due to the smart smoother, and since the alternate SCD works on luma diffs, do i need to start using the normal SCD at 99% or so? I'm encoding the new DBZ episodes and the source contains many mpeg2 macroblocks so the filters are non negotiable and i don't really want to manually insert kf's since my .d2v is at 29fps and i'm serving nandub with avisynth and decomb1.9 for IVTC puposes.

Any help would be greatly appreciated!

James

[poopity poop]

I don't usualy load my d2v, but that's because I ONLY use Gknot for ecf creation (for kf insertion)and bitrate calc. I know many of you use it to make another stats file and whatever.

If you want to load your d2v file(and are doing anime), you may have to make another d2v file, one with FORCE FILM turned on so you can see if it is inserting keyframes and stuf in the right place.
Remember(if you are doing anime this is):
your d2v should be 29.97 fps
your new ivtc'ed tpr should be at 23.976fps
which makes your stats file 23.976fps

makeing a new d2v with force film will allow you to view the correct frame when you load the d2v file into gknot.

Just a trick.

In responce to the 2-3 movies/DVD. Totally, when I get a DVD burner I will make SVCD's with about 2gigs/movie, and Divx at about 1gig/movie and maybe 2gigs for a long movie. Of course it all depends on the length, but oh man...when those DVD burners come down in price, there will be high quality, high file size, shit flying around!

[jrmillerUT]

This is kinda off topic but do you know if the X-box can play VCD's and SVCD's?

[takeru]

you can get the pioneer dvr-a03 for like 376$ now. dvd-r's you can get for like 2.95 for 4.7gig or 5.95 for 9.4gig double sided. kinda wierd though. to burn the double sided discs you have to flip it upside down to burn the other side with the pioneer :) brings back old memories of flipping 5.25" floppies to use the other side... aaaah. around april pioneer is supposed to step up production of dvd-r's, so hopefully the prices of the drives and media will drop like a rock.

[jrmillerUT]

can the DVD-R's burn CD audio media? I don't have another bay free on my comp. :-(

[UHT]

no, they cant

[Crucio]

sorry if this has been brought up already, but i encode a lot of anime with nandub. Many times i have used aviutil to go through each frame individually and fix up with paintshop pro.
Now i was wondering, i saw how smart smoother and warpsharp improved the image, but i noticed that the lines were not looking perfect. I know that the image was blown up, but i am experimenting with flaxen cartoon tool and have found that withing reasonable parameters, that filter can really enhance a video. Ofcourse if the filter is overused, the video is ruined.

My typical program list is, if im not editing frame by frame, dvd2avi, tmpeg for auto ivct, vfapi, nandub

in nandub the filters are, area based deinterlace, smart resize, smart smoother, warp sharp, the if neccessary, the cartoon tool.

peace out to the other animation freaks :)

[Crucio]

a pixel base of about 70 along with default values seems to do a good job (look at the inverse map)

[poopity poop]

I was the first one to say to use smart smoother and warp sharp together, and I think I should be the one to totally wipe it off the face of the planet....read on

I'm working on a new site that is explictly for anime filters(noise reduction). It will be ready soon, not sure when.
SMART SMOOTHER AND WARP SHARP is good, but it sucks compared to the other filters out there. BELIEVE ME! I'm testing them

From my "preliminary test"(sorry that sounded gay but whatever) the number 1 best filter to use is a close tie:(note: one one!, much much faster)
2d_celaner optimized for performance:
http://sauron.mordor.net/dgraft/2dcleaner.html
and
Smart Smoother HQ
http://sauron.mordor.net/dgraft/hiq/smoothhiq.html
There is also an area based smoother that can be used in conjuction to fill anime characters with their original color:
http://www.maven.de/ code>area smoother

I just feel its my duty to dispell that the once good method of SS and WS together is good. Its...ok.. but its so FREAKLING slow and the 2d cleaner is freaking great. PLEASE check these filters out they ARE better...I invented the SS/WS method you've got to believe me :)
--------------------------------------------------------------
Read on for the best settings I've found:

Just to spoil the findings of my tests, and because I feel obligated to have people stop using the picture destroying method of SS/WS here are the best settings I've found so far. Of course figure out what each setting does and apply it to your anime that you are encoding, but these seemed to work well in all senarios of anime(high background, large drawings of character(no detail), and normal frames)

Smart smoother HQ:
dia=7
threshold=50 (I think)
Amount=125
Weighted average on
weighted average wtih difference on

2d cleaner:
note: Did not do much testing with this yet:
defualt settings worked fine

for Area based smoother there are too many oiptions to remember without having my website in progress with me where I'm writing this but use this filter to accually fill in entire color blocks of averaged color where the animated characters have been colored in. You'll see what I mean once you get it.

JUST PLEASE DON'T USE SS/WS! Its awful compared to these fitlers, and slow!

[ThePanda]

Hi, so far I only finished reading the first page of this thread but I wanted to post this now b/c I have to go:

1. I use DVD2AVI to open the vobs, RGB mode, I crop with it, making sure to crop out those annoying flashes on the top and bottom on every scene change in many animes.

2. I load the d2v through VFAPI into VirtualDub

3. In VirtualDub, I do IVTC(adaptive) in the frame rate menu. filters: I do deinterlace(blend), temporal smoother(2 to 5 depending on fuzziness etc), resize(640x480 precise bicubic A=-0.60)

4. Then encode that at 1350kbps Divx4 2-pass w/ default maximum postprocessing.

5. Then mux the AC3 in Nandub.

This seems to give good results for me. I IVTC in addition to the deinterlace(blend) filter because I hate how every IVTC filter misses a lot of those annoying horizontal interlace lines, no matter how good it supposedly is, and I figured that it wouldn't do too much damage to the large uniformly-colored areas of most anime anyway. I remember trying SmartDeinterlace before, but I forget why I stopped using it. It sounds good, so I guess I will try it again so I can preserve better quality in most of the frames.

To the person who said that just deinterlace alone is better than IVTC for everything: I think that is wrong because I thought I remembered reading that some frames are interpolations of two others, so if you do the wrong method it won't fix that and you will be left with 2 images on one frame, which will make it look bad and take up more bits (or something like that).

Another personal note: If the DVD has both English and Japanese audio, I mux both AC3 tracks in Nandub (with the proper delay). I render the resulting AVI in GraphEdit to make an English graph and a Jap graph, and I add the DirectVobSub filter to the jap graph. I put the subtitles and the AVI w/ both languages in a subfolder called "data", then I just click the english.avi or jap.avi graph.

I hope any of this helps, I've been experimenting on the anime for awhile :)

[NinjaGara]

Poopitypoop

I tested the filters you mentioned and excellent!!!
Nothing to do with SS and WS!

If you need I-Author, perhaps purpleman.cjb.net is still hosting it.
Some week ago I downloaded it from there and I tried to use it but I got a security error message... I still dont know why...

But, by now, I have enough to do with nandub!

[takeru]

i did some testing on 2d cleaner and smoother HQ. actually, i messed with smoother HQ before you mentioned it. smootherHQ does give decent results, but its pretty slow too. 2d cleaner is very fast, but really kills fine details, so its best to use with cartoons/anime with not much detail, like simpsons or something. currently testing area smoother.

[poopity poop]

I've recieved an e-mail from klaus Post(the maker of SS HQ) and he has stated that a faster, optimized version of Smart smoother HQ is in beta and on its way. I feel that it will make it one of the best anime noise reducing tools out there. I may be beta testing it and I'll keep you posted here.


NinjaGara->
There is a "cracked" dll that needs to be copyed into your windows/system directory to replace the old one that gives you that message. Check the textg files that came with the rar you downloaded

[NinjaGara]

yes, SS HQ is very slow but gives better results than SS and WS.
I hope Klaus finish an optimized version of this filter soon.

poopity poop

You already got it!
Well, I did that! But for some reason that didnt work...
I will try it again, it seems that I made a mistake...

[poopity poop]

please set up your private messages so we can discuss the dll. I tried to private message you but you have not set it up yet :(

Cracked things and warez are not allowed on doom9.

[NinjaGara]

Well, I changed my options and now I can receive private messages!

[OUTPinged_]

Heh, its me again =)

A pity dragonz isnt there, he wrote such an interesting posts..

well whatever.

poop:

Heh, I glad you have realized that combo gives unsatisfactory results. SSHQ and 2dc are better, but! you havent tested them, they dont work good on default settings and you have to tweak them individually for each encode. It kills background on default settings. Try to reduce thresh to 1 on sshq and you will see the amount of detail lost. I personally wouldnt go over 25. 50 is overkill.

as for 2dcelaner, you shouldn state it works "good" on default too. The purpose of the filter is to blur the picture. I personally used 2dc on one encode (1x1, t21) and it gave tolerable results (BAA r1 dvd, same quality as 1st r1 eva dvd). Still it is smoothing too much.

for light luma noise you should really use temporalsoften(1,10,1), it doesnt smooth picture in that case.

My point on this is:

SPATIAL SMOOTHERS ARE EVIL! Use them only as the last resourt.

btw, i have tested a beta of sshq1.2, and the filter gives a whooping 2fps on p3 machine. Well isnt that cute? make it two times better and you will be able to actuallly use it.

[poopity poop]

I have to disagree with everythgn you said respectfully.

SSHQ works well and does not smooth out details as much as you lead on. That's the point of the weighted and weighted with difference options., and the 2_d cleaner does not smooth sharp area's at all, as long as you set them correctly. Simple as that.

I do agree with you that you must tweak each filter for each encode...sort of. I encode Kenshin, they are made be the same people and have about the same amount of noise from episode to episode, so I will pick a settings and stick with it unles I notice that they are becoming too smoothed which I doubt will happen

[breathtaker]

because this is a anime topic ill post my question here. ok i did 2 other anime dvds and this one is my third and i am useing vdub and i have alot of blur and blocks in my picture . i used sshq,warpsharp,2d_celaner and some others but they looks all the same i still cant get rid of it. what else can i do?

[poopity poop]

well that is a bit vague. Are they messed up artifacts in the encode or is it just blurry?

If its blurry its because you are using too many or too high of filters setings. I hope you are not using all of those filters. Pick one, I recomend SSHQ if you don't care about speed, or 2_d cleaner optimized.

If you are getting encoding artifacts, you should use anti-shit

[Crucio]

damn...
"ThePanda" Let me give you some pointers, ok?

First off, rip your movies in YUV. It is what they are and divx is optimized to support yuv. RGB will slow you down. Another thing. Do not crop in dvd2avi! Dvd2avi doesnot give accurate results and is slower.

Next, DONT use VirtualDubs IVCT. Tmpeg or avisynth, hell, even aviutil are better solutions than VDUB!!! Also, for anime, really you should be doing ivct manually, but if not, then use tmpeg/avisynth.

Next, the reason IVCT is important is, TV is interlaced. It displays 60 fields/sec. When those fields are combined they are 30fps. But, there are 6 extra frame inside of 30 to make up for the flicker you would get in NTSC video at only 48fields/sec. On the computer, the monitor is progressive and has a higher refresh rate. Thus, the video does not look jerky when cut down to 24 frames. It actually looks better. A good ivct will eliminate most interlacing problems too. which brings me to my next criticism:

DONT EVER EVER EVER EVER USE DEINTERLACE: BLEND!!!!!

the best filter you can use, if your too lazy to go through the anime and deinterlace/ivct manually (paintshop pro and aviutil) is to use area based deinterlace : and uncheck blend. You must set it to interpolate. If you dont, then you suck :)

anyways, back to this shizznit. Poopity poop. Do you really consider yourself an expert ( im not making fun of you im just curious) I am positive i found those filters at a site besides yours first im sorry to say :(. You were in my knowelge the first to bring them to this board though :)

I would like to know what encoding group (isonews) you belong to, maybe we can chat in #nandub on dalnet, anyways.

Have you ever gone frame by frame to fix anime?


Also, about the filters.I was encoding fatal fury and experimenting with the cartoon tool. at a base of 70 it looks very nice, also, i cant see why you dont like the warp sharp way. I have tried those other filters. they are good, but maybe an 8 direction sharponer is better. Actually If you want incredible quality, simply hook your HQ progressive scan dvd player to a HDTV upconverter, then a signal strengthener, a video stabilizer and then to a pinacle dc2000 video card.

Im just joking, not many people are that rich :(


I think that the only way that you will ever fix up a video is to work with the source. A dvd just cannot be considered a source. Some companies have good mpeg2 encoders, but even with a noise reducer, you just cant fix them. Like dbz movie 1. In the red, omg, there was so much pixelization :(

[ThePanda]

Originally posted by poopity poop
And I really haven't been following the posts about ivtcing and such...but whoever said ivtcing is a waste on anime is a fool. There are two ways you can get rid of interlacing: ivtc(the best and most proper) or deinterlace filter(discard field 1 or 2). Blending or duplicating is looks ridiculous. But you can accually get good deinterlacing results if you discard field 1 or 2...try it I'm serious

of course you have to watch your frame rate, because ivtcing in tmpeg gives you 24fps...which is optimal

You can just discard one field and then the interlacing is all gone? Does this make the vertical resolution 1/2 of what it was before, so you could encode at like 640x240 and do a resize player filter to stretch it on playback?

[manono]

Hi-

You can just discard one field and then the interlacing is all gone?

Yes, it will be all gone (but if you have some sort of blended fields source, then you still might be stuck with some ghosting). But this is the purpose of the VerticalReduceBy2 Command in AviSynth. And by resizing before encoding, you don't have to use your player to resize. But there's a huge drawback to knocking out every other field and then resizing (the fast deinterlace method in GKnot) and that is you frequently introduce the shimmer-flashing that all anime encoders hate. So I would not recommend that method.

Edit: 2 days later it was brought to my attention that the VerticalReduceBy2 deinterlacer doesn't take one of the fields, but blends the two together (thanks ws). So that may lead to ghosting, although it does get rid of the interlacing well enough. What poopity poop was proposing is possible in TMPGEnc. as well as with the SeparateFields-SelectOdd (or Even) commands in AviSynth. That also gets rid of the interlacing, but may lead to shimmer-flashing in some cases. Sorry for the misinformation.

For more on the subject, along with pictures go here (http://www.engr.orst.edu/~sheareni/deint/deintfaq1.html)

[ThePanda]

ok, thanks for the info. i'm trying smartdeinterlacer now instead of deinterlace(blend) with the settings i typed up above. hopefully it will work well

[Crucio]

THE best deinterlacer is "area based deinterlacer" use interpolate

[takeru]

Originally posted by Crucio
THE best deinterlacer is "area based deinterlacer" use interpolate

which area based deinterlacer you talking about? the one by gunnar thalin?

using area based interpolate with sshq after tmpeg ivtc gives some pretty good results, but also makes me feel like my athlon xp turned back into my p166 (1-2fps!!!)...

[poopity poop]

so cute...so cute...

The best deinterlacer is NOT area-based, or even the deinterlacer filter in vdub.
First you must understand what interlacing is, then you will slowly realize that Decomb is the best, then GreedyHMA and Tmpeg...silly people

[Dali Lama]

Poopity Poop, There is a new IVTC plugin for Avisynth called "IVTC 4". It is very good with mixed media content like anime. I achieved superior results with this than decomb. However, decomb is still the best for regular movies.

Bye,

Dali

[Crucio]

OFCOURSE i do ivct before the ABD, but ABDeinterlacer is still best. Only for lazy people though. I do ivct maually. Also, i edit use aviutil to copy and paste good frames onto bad frames

[ThePanda]

hmm now i remember why i stopped using this smart deinterlacer. it causes all white flickering around the motion when the characters move their mouths and stuff. maybe that was because i had it set to interpolate instead of blend.

[ThePanda]

I found this frame tweaker plugin (http://www.geocities.com/cplarosa/video/tweaker.htm) that could let you fix those frames on the scene changes of a lot of anime with the horizontal blackish lightning-bars on the top and bottom of the screen. It seems like it would take a ridiculous amount of time though scrolling through the video in slow motion looking for which frames it happens on, so I'll just clip out the area. Most people just leave them in I think, but they annoy me too much :D

EDIT: Crucio: AviUtil lets you do that too? Is it easy or does it take a long time?

[Crucio]

its relatively easy, its best suited for copying a frame that is better before the good one, or copying the frame to the clipboard and then pasting into paintshop pro. It can be done pretty fast though

[manono]

Hi-I wonder if you could give me more information as to why you believe that IVTC4 is better than Decomb. I haven't tried IVTC4 yet and have found Decomb to be excellent. So I'm curious. Is it because (as daxab admits) IVTC4 is slower-No, that can't be it. Is it because IVTC4 doesn't have a deinterlacer built in to catch any stray interlaced frames that sneak through-No, that can't be it. Do you find that Decomb lets more interlaced frames through-then you can tweak the setting-maybe change the default setting of 15 to 10. Does Decomb deinterlace good frames making them jaggie (much more likely)-then you can weaken the default setting of 15 to 30 or so.

From what I understand, IVTC4 is good on easy, properly telecined material (as is Decomb), but it will fail on more difficult material. But as I said, I haven't yet tried IVTC4, so I'm curious.

[western shinma]

But you haven't even tried it yet! Donald Graft even said that it "appears to work well on several torture clips that Decomb has problems with." Where did you get the impression it only works on easy material?

[manono]

Hi ws-

Dali Lama made a pretty bold assertion that I couldn't let pass. I can understand if IVTC4 is as good as Decomb in some anime material. I can not understand if he believes it to be better than Decomb in all anime material (using the latest versions of Decomb) which is what he seems to be saying.

DGraft's statement was made some time ago, and more recent versions of Decomb have since been released.

I hope Dali Lama comes back to let us know what version of Decomb he was using, and what the AviSynth script was, and if he experimented with tweaking the threshold or other settings before comparing the 2 methods. Because, otherwise I'd say it's an unfair comparison. And an unfair statement.

[neuron2]

Gentlemen,

Let us keep some perspective here. There will always be a need for different kinds of tools in our toolboxes. One screwdriver is not going to be enough for all jobs; we need a phillips head, a flat, and even sometimes a torx. And those too in different sizes. It all depends on the nature of the problem and the goals we have. So let us not begin some kind of war here over tools.

Having said that, I do want to address a few points that have come out here. I must say that my reaction to Dalai Lama's point about Decomb being better for "regular movies" was the same as that of manono. I was motivated to write Decomb by the fact that anime was not treated well by existing IVTC tools. And the early successes I had were achieved with difficult anime. Decomb does also work very well with clean telecined source, what I suppose is meant by "regular movies".

Now let's get technical. Decomb without postprocessing is a straightforward field-matching approach to recreating progressive frames. It is completely blind to patterns; it deals with each frame one at a time. IVTC4 is also a field matcher in effect, but it can overrule the result of field matching with what it thinks is the prevailing pattern. This is an interesting approach. But due to its pattern awareness it is currently limited to the patterns for which it is coded, i.e., 3:2 pulldown. That is not a problem, it is just a different kind of screwdriver, a specialized one. Being specialized, however, obviously limits its usefulness, e.g., my PAL friends cannot use it.

When IVTC4 first came out I ran it against my torture clips and I noticed that it recreated *a few* progressive frames that Decomb did not. I assumed first that it was better field matching code in IVTC4, but it later transpired that it was the pattern override that kicked in (also please note that IVTC4 fails at other points in the torture clips where Decomb succeeds). Before realizing that it was the pattern override, and still thinking it was the matching code, I went into a coding frenzy and discovered a much better matching method that allowed Decomb to recreate those few frames correctly, and this new matching is now in the released version.

Note also that Daxab advises that IVTC4 be followed by my FieldDeinterlace() to clean up the bad frames that come through IVTC4! After doing that one wonders why one shouldn't just use Telecide() to do the field matching as well. I don't think anyone has shown that IVTC4 field matching is better than Decomb's current matching, and if it isn't, how can one argue that IVTC4 handles anime better than the current Decomb?

So, it is different strokes for different folks. If IVTC4 does the job for you and you prefer to use it, fine. Sometimes the color of the screwdriver is important to some people! If you work in areas where IVTC4 does not apply, then you'll need Decomb.

Finally, for me the important things are to be useful to people through my efforts *and* to participate in a collective endeavor to find better ways to do the things we do. That involves researching *and sharing* our results. That is why every single line of my source code is available, and the same can be said for Wiz, trbarry, and others. I think it is a very bad precedent to be witholding source code and not cooperating in the collective endeavour, while benefiting from it.

[ThePanda]

Hi, you can call me crazy, but I actually prefer the smoothed and consistent look of deinterlace(blend) for anime :). If you use smart smoother or something, why not just use deinterlace(blend) and eliminate smart smoother, and just use temporal smoother? I suppose I would want to use something else for a live action movie with more details though.

Also, thanks for pointing out not to crop in DVD2AVI, VirtualDub seems to be better for it.

[Dali Lama]

Sorry that I did not reply right away, I have been studying for tests:

I like the new IVTC 4 on anime.

IVTC 4 gets rid of all of the interlaced frames in anime. Decomb does not.

They are both equal in speed on my Duron.

Decomb causes avisynth errors that can be reproduced. IVTC 4 does not.

IVTC 4 has no options to set or think about--> easy to use. Decomb makes you think about which setting to use, but once you know which one, its pretty easy from there. So ease-of-use is pretty good for both, but IVTC 4, like Donald said is catered to NTSC users, which I am.

The Version of Decomb I used was 3.1. The version of IVTC 4 I used was 1.01.

Side Note: Decomb produced beautiful results on a pretty hard movie: Crouching Tiger Hidden Dragon. It switches from progressive to interlaced and NTSC to Film. Very complicated. Therefore, I like Decomb for movies.


Do I know the technical reason that IVTC 4 works better on anime? NO. Why Decomb did beautifully on CTHD? NO. But I respect thier work and hope they continue development.

Thanks,

Dali

[neuron2]

>The Version of Decomb I used was 3.1.

As manono pointed out, that is one revision behind the Decomb with the improved field matching.

>IVTC 4 gets rid of all of the interlaced frames in anime.

That is a demonstrably false statement. Not only is it demonstrably false, but daxab recommends placing my FieldDeinterlace() after his filter to correct the bad frames that slip through. Why are you making false claims like this?

For others that may be misled by your false statement, here is a direct quote from the author of IVTC4:


It [IVTC4] is not perfect. It has to deal with inconsistencies in the source material due to noise, edits, and CG overlays (30fps animation overlaid on telecined 24fps video, seen in things like credits and anime). IVTC4 will try to handle all of these problems, but interlaced frames will emerge if things get dicey. If this is a concern, I recommend following IVTC4() with a high quality deinterlacing filter like Donald Graft's FieldDeinterlace(), from the Decomb plugin.
Finally, I'm not aware of Decomb "causing avisynth errors". Please elaborate.

[western shinma]

Dali Lama,

IVTC4 does have options as well, although it seems you were just using the defaults.

Also, there is nothing complicated about this:
switches from progressive to interlaced and NTSC to Film
That is normal, it just means there aren't any pulldown flags in the so-called NTSC sections which are actually telecined film just like the rest of the movie.

[Inindo]

Hello. This is my first post here, and I want it to be contributing something, instead of asking for help. However, I am rather new at this so please be kind.

I’ve been testing out the Area Smoother. It seems to work all right, but I don’t think it’s very practical yet. The most obvious problem is how to prevent it from eating the background. This will shade a given “area” with all the SAME pixels so it will destroy background if it decides to “smooth” it over.

Increasing Max Pixel color difference and Max Span Color difference will help. However, you will begin yo see jagged lines when a color fades into a solid color. Plus, many areas will be missed. Increasing Minimum span Width and Span Overlap seems to help. But then the jagged line problems becomes obvious at certain points.. Another method that is simply lower the noise margin. Lowering it to around 1.3 generally does the trick for me.

However, this lends another problem. Now, in order for any areas to show at all, I need to turn down all the other values to almost nothing before I see any kind of result. Otherwise, it can be an large area of obviously uniform color but still not grouped into an “area”. Even so, it will still turn into a “patchwork quilt” various areas, never really coming together into one large one. I’m still trying to find a happy medium between the two.

Another problem is that the “areas” tend to have jagged edges, which are sometimes rather noticeable. This applies to both the large uniform areas, and the small scattered ones.

I was thinking about first running a normal smoother, then Area Smoother, but that will take a huge amount of time. Any ideas?

[Crucio]

hey "thepanda"

i too like the look of the blended frames. This is only good if you have made sure there are no interlaced frames though.

Otherwise, you will get a ghosting effect

[manono]

Hi-

Yeah-I fooled with it for awhile and came to the same conclusion as you. While it might be useful in animation where the colors really are solid (The Simpsons?), in any case where one color gradually changes into a different one within the same frame, the results are horrible looking, with the jagged edge effect you noticed. I gave up.

[Inindo]

Yeah. I tried it on Cowboy Bebop: Sympathy for the Devil. Lots of dark areas, blended areas, and detailed backgrounds on this one. I tried the smoothing fiter beforehand and ie helped a little. but not much.

The problem with this is first, each scene can give completely diffrent results. Second, that it is possible to screw up a scene REALLY badly, in a painfully noticable way.

Eh, great idea in theory. I suppose it might work with some others.

[poopity poop]

This is a new topic for anime and I just have a quick question not to deter anyone from the ivtc stuff going on:

I hate seeing mpeg-4 artifacts in my anime esspecially on the subtitles. I'm asssuming anitshit-full, while slow, can eradicate these problems. I'm asking this(because I'm lazy):

Who has experimented with different antishit settings and can tell me(in a private message, so as not to deter from the ivtc discussion) the best settings for anime with subtitles that are being burned into the frame by vobsub, so as to most limit the amount of crap that gets through?
-------------------------------------------------------
Also, perhaps there should be a seperate forum for anime(DivX, SVCD, etc.), seeing as how its more difficult to encode and a lot of people are interested in it

[Crucio]

naw its cool, ill just post it here.

i have found 17-0-0 to be best for anime

also, the thing about the subs, i have found its best to just include subs for BSplayer. Theyre MUCH better that way

[poopity poop]

yes but I have an audience on DALnet that expects subs to be burnt into the frames.
I'm also going to learn everything there is to know about X(S)VCD's and basically reproduce a 2episode/CD rip of the Kenshin series that will be 10x's better than adx(switchable subs, 2 audio, menu). They were so smug when I showed them a sample, mine will be GRRRRREAT(read like tony the tiger)...sorry I'm excited about SVCD's!

[Wyldchyld]

Greetings to all! This is my first post on the Doom9 forums, and i hope to contribute as much as i'm sure i'll learn from everyone here. My principle interest is anime ripping, from DVD to DivX, from both full-length feature anime and episodal anime sources. I've been heavily experimenting with a variety of methods for a good 3 months now, and feel confident enough to bring something to the forums :). I've posted my PC's stats in my signature so as to provide a constant reference to any ripping benchmarks i may post here.

I'd like to begin my applauding the tremendous wealth of information contained in this post alone. By simply reading this one thread, any budding anime ripper can find him/herself with nearly everything they need except experience, which is a rarity of sorts even in this day and information age.I do however, strongly support Poopity Poop's suggested creation of a seperate anime-ripping sub-forum on Doom9, as i feel we could all greatly benefit from it, not to mention providing a definitive resource for this particular brand of ripping, undoubtedly one of the most difficult IMHO. I'd also like to offer my own 2 shiny pennies as to my findings w/ the two hot topics in this thread, IVTC and Filter combinations for anime. Oh, and anyone wishing to know more about my experiments may contact me, of course :).

A) Deinterlacing/IVTC'ing anime: I've experimented w/ Decomb, Telecide, Smart Deinterlacer, VDub's own Deinterlacer & IVTC algorithm, as well as manual and automatic IVTC in TMPEG. In the end, as painful as this may sound, i've found that a careful manual IVTC in TMPEG has produced the quickest and best results! Barring any major oversights during the manual process, and occasionally adding the Gunnar Thalin Area-Based Deinterlacer ("the GTABD") has produced truly remarkable results for me. Honestly, it's not even that hard, and i think everyone should give it a try!

B) Filtering anime (VDub/Nandub): I'm sorry to say, Poopity Poop, but i've been experimenting w/ SS/WS &/or 2D Cleaner for quite some time now :) On the other hand, i couldn't agree more that Donald Graft's 2D Cleaner ("the 2DC") (i use the K7-Optimized version by Jaan Kalda) produces excellent result for luma noise cleaning. As you stated earlier when criticized about your opinions resulting from experimenting with the 2DC, "the 2_d cleaner does not smooth sharp area's at all, as long as you set them correctly. Simple as that. Once again, if *careful*, you can reduce the blurring to a very minimal level; and again, like ALL DVD RIPPING (!), there are no "right" settings! If someone put a gun to my head, however, i'd say that a Threshold of 5-6 w/ radii of 3/3, 4/3, or 4/4 seems to work best. In some cases, a threshold of 9-12 can even be used, but not for higher-quality painted backdrops.

Other than the luma noise-cleaning component, i almost always use a subtle brightness/contrast modulation, which again is almost always set to -2/-6 brightness, +6 contrast, which again helps mitigate luma noise if used in concert w/ the 2DC. When resizing i've come to the conclusion that VDub's own resizer, using Precise Bicubic filtering, is the only way to go as far as i've seen. I NEVER crop more than 8 pixels all around, w/ the aspect ratio usually be within 0.1%, sometimes 0.2% (as it should *always* be!). For those who'll insist that Precise Bilinear works best, i just find that it blens WAY too much, and prefer to rely more on Nandub's "Motion-Based Crispness Modulation" option for high-motion "blurring". Otherwise the resize is usually accomodated quite well by this filter. The size increase w/ bilinear filtering isn't to be ignored, i'll give you that, but i find it an acceptable compromise. Also, given that i strive to make very high quality rips, i rarely stray from a horizontal resolution in the 540-640 range for 720 sources. Anything and well, i don't find that they become good enough backups for myself (because that's what i'm doing, in the end). I end up with 85-95%% DVD quality rips for 23-minute episodes, 2X/CD, and i'm happy with that.

For subtitles, i always rip subs, always in english unless by special request, always burn them in the frames, and i almost always apply my own blend of colours, which is something i'd love to discuss in another thread btw :sly:. For those of you who're interested, my fill colour on the main subtitle stream is usually an RGB in the neighborhood of 250/230/100, and the outline is again usually around 20/15/5. I must say that i've seen some really nice subs by http://www.elite-fansubs.com on their "Argento Soma" fansubs, which were a White/Off-White/Pearl fill w/ a Teal/Aqua outline, truly impressive. For "Secondary" or "Overlap" subtitles streams (like, say in "Neon Genesis Evangelion" when the radio plays in the background) i use a standard White/Black Fill/Outline. In response to Poopity Poop's question about reducing noise around burned-in subtitles, antishit does yield decent results for sure. I use 10-15/0/0 and it works quite well, though i can't get rid of them altogether most of the time.

I hope i'm not too late to contribute to this thread, and if so i'll just pick my socks up a join in next time. If anyone has questions of comments they'd like to ask me, please don't hesitate, but i prefer being contacted by ICQ (59582232) if not through the forum. Thanks, and cheers!

Wyldchyld :devil:
Current Project: Ghost In The Shell, Subbed, 697-702MB, DivX 3.11A Nandub SBC, 608x336-640x352.

[manono]

Hi Wyldchyld-

Welcome to the forum, and if all of your posts are as excellent as the first one, then you'll be a great addition.

I think that one of the best points you made was the one about not going overboard on the filters. From just about the day it was released I've been using the Optimized 2D Cleaner, and have slowly adjusted the settings downward so that I'm now usually using it with a threshold of 5 or 6 and a radius of 2/2 or 3/3. I may put WarpSharp on top of it with a 16 setting sometimes. But it's just too easy to blur clouds, walls, tree leaves, the ground, etc. if you're not careful.

One minor correction, though (so minor that I almost hate to bring it up, but credit where credit is due). Although Donald Graft's site hosts the 2D Cleaner, it was written by Jim Casaburi and optimized by Jaan Kalda.

[Wyldchyld]

Manono: Thanks for the kind words, and you're right, the 2DC is Jim Casaburi's, my mistake! Using Warp Sharp w/ the 2DC is not a bad idea at all, but it must definetaly be used only in moderation. My own testing has lead me never to use it past a threshold of 30, 2 blur passes. The result is definetaly an improvement, but it seems to get rid of detail really fast, and i don't think i'll be using it unless the source material is really noisy. I think i'll give your 15 threshold setting a try, since i've never used it with such a low value, but tell me, how many blur passes wouls you use w/ that threshold value? Running a quick preview on "Ghost In The Shell", a plain single-pass blur seems to work nicely. Furthermore, what are your thoughts on the added processing time incurred by using Warpsharp? Acceptable compromise? From my observations, depending on the source material and any previous filters applied, it can add 10-20% processing time on my system, so it may not always be a viable solution, no matter how pretty :).

Wyldchyld :devil:
Current Project: Ghost In The Shell, Subbed, 697-702MB, DivX 3.11A Nandub SBC, 608x336-640x352.

[Crucio]

SVCDS SUCK!!!!!!!!!

[Crucio]

:0

well, anyways, has anyone tried to rip the fatal fury oavs? they, well, IVCt very badly and i was wondering whether it was telecined at all.

I cant seem to do it manually.

O well, my current projects are Golgo13: queen bee and fatal fury

I rip for AMC on dalnet, and efnet.

Who do u guys rip for?

[manono]

Hey Wyldchyld-

I don't know what a blur pass is. Does that mean that you put the filters on twice? If so, I only do it once.

Here's my current filter chain for some El Hazards I'm working on:

Levels-100 (it'll say .39 on the outside-use the bottom left slider to lighten the picture and uncheck that box to make it RGB)
2D Cleaner-6, 2/2
WarpSharp-16
Levels-100 (use top teft slider to darken it back up and uncheck the box)
Saturate-1.15 (from Hue, Saturate, Intensity filter). I don't necessarily recommend that one-I just like things a bit more colorful-doesn't change it much.

The reason I put on the levels is that the smoothers-cleaners don't seem to do much to medium-dark areas or to the noise around the subs, so by lightening things up before smoothing-sharpening, the 2D Cleaner-WarpSharp combo can do a better job. Maybe it's just my imagination.

As for time, I'll run the first pass with only Crop-Decomb-Subs-BicubicResize(512,384,0,0.5) and Fast Recompress (resize varies depending on the quality/size I aim to get-usually 3/CD). It's done with an .avs generated in GKnot with Decomb edited in. That will generally take 53 minutes for a 24 minute episode on my system (these El Hazard OAVs are a bit longer). Then in Nandub, I'll put the filters on for just the second pass together with anti-shit 12 and Full Processing (I know-the guides say to use the same filters for the first pass as you intend to use for the second pass, so don't try this at home). That will take around 1 hour 53 minutes or so on my machine. I don't know how much of that time is added by WarpSharp alone. But before the Optimized 2D Cleaner I had to use the Smart Smoother and it was so much slower, that this way now seems fast by comparison. Filtering before resizing should improve things even more (thanks ws), but I'm lazy and impatient.

I also would like to see an anime specific Forum here. As it is, the anime posts wind up scattered all over the place. I would also like to see an IVTC-Filters forum (although those things could be covered in an anime forum, "real" movie encoders also need them sometimes).

Who do u guys rip for?

I rip for myself.

Edit: I changed a couple of my settings for this particular series. As poopity poop says, always experiment with different settings for different material and find what looks good to you. The author may not be held responsible for any damage caused by following his methods.:)

[Dali Lama]

Could you explain how you "manually IVTC" in TMPEG. I have tried to do it, following doom9's guide, but it confuses me. I tried to get a pattern and it doesn't work for the whole movie?

Thanks,

Dali

edit: Right now, I don't have any anime DVDs to test on. Could anyone send me a short clip of a scene to try out some of the methods Manono and WyldChyld gave. Thanks again.

[Wyldchyld]

Crucio: Who do i rip for? Myself mainly, but also for a few close friends of mine; i find anime to be such a unique genre, one that has even significantly affected the way i view many things in life, to a point where it's become something that i want to share with others. Given the chance, i'll probably also trade my DVDivX rips with other people, but preferrably not have them "released". On the topic of Fatal Fury, i can't help you there since i haven't even seen them! If you're wondering how to IVTC manually, i'd suggest checking out the Doom9 guide, it's how i learned to do it. (See below, my section to Dali Lama for more info).

Manono: The "Blur Pass" option is one of the 2 options in the standard Warp Sharp filter; i suspect you may be using an older version of the filter, or mayhaps it's me. In any case, here's a screenshot of what mine looks like: http://www.promisemeheaven.com/WS_2-Ops.gif. Seeeee? "Blur passes" :). I only use the filter once, as using it twice would undoubtedly yield an exponential result, being somewhat disastrous too i imagine. I think i'll try it just for fun some time... :D.

I like your idea of playing with the levels, although i haven't touched upon that too much just yet. I find the difference between using a levels filters and a bright/contrast fairly negligeable, but i agree with you that it produces nicer results, and i maintain that it seems to help "restore" in a way some of the detail blurred by spatial smoothers, that is if the damage isn't too extensive already :). As for using a levels filter before a sub filter, i haven't tried that actually; i would tend to think that noise around burned-in subs is caused simply by compressing the frames, and as such the filter would have no effect, but i don't know nearly enough about video noise to say that for sure.

I like your idea of doing 2 passes w/ different filters, even if it is true that you wouldn't be able to reap thefull benefits of the "standard" 2-pass nandub SBC encoding method, but i also haven't tried it yet. I'm guess the filesize would have a slight deviation to it, somewhere in the range of 5-10%, which is something one could easily adapt to while ripping a series, for example (even if each episode has completely different materia, the lengths are all roughly the same, usually). I'm also starting to wish i knew more about using AVISynth since everyone using it reports a great success, while i don't even understand how it works yet! LoL :p.

Dali Lama : I only followed Doom9's guide to manual IVTC'ing and it worked out great, save for the fact that it took me 3 attempts to get it right. The first time around i actually Telecined the entire movie AGAIN (yes, believe it or not, i did the exact opposite of what i'd set out to do), you can imagine the result! In the end it worked out fine, so i'm not too sure where you could've gone wrong. One thing i'd tell anyone attempting this method is to remind them that originally, when Telecine is applied to source material, it's most often applued several times (as in, with different patterns) throughout a movie. As such, once you've entered your first pattern, you'll need to scan through the entire movie AFTER the point where you first entered a pattern, find a point where your pattern is no longer effective, and apply a new one from that point on. Once i'm done ripping Ghost In The Shell satisfactorily (which is taking a damn long time for some reason, i'm never satisfied), i'll rip a sample from the GITS source DVD for you, and try to explain it, that should help.

NOTE: Anyone who'd like the GITS sample to help practice how to manually IVTC can let me know and i'll arrange it.

Also, i had to read up alot on Telecine and IVTC'ing before getting it right. Here're the most useful links i found over the past few weeks:

- http://doom9.org/mpg/tmpg-ivtc.htm (Doom9's own guide).
- http://www.inwards.com/~dbb/interlace_myths.html
- http://arbor.ee.ntu.edu.tw/~jackei/dvd2avi/ivtc/
- http://www.inmatrix.com/articles/ivtc.shtml

All of them basically cover the same thing, some are more technical about how it works, other detail a method of doing it, etc. Hopefully this gives you a leg up and something to do until i can get you this sample, which might take a few days as i said, i really want to finish this rip first :(.

Wyldchyld :devil:
Current Project: Ghost In The Shell, Subbed, 697-702MB, DivX 3.11A Nandub SBC, 608x336-640x352.

[manono]

Hi Again-

You're right-I must have an older version of WarpSharp. Mine only has one slider. Thanks for the pic.

I'd guess that the levels stuff does the same as your way with brightness-contrast. I found levels very easy as you use the bottom slider to lighten it up, and the top slider at the same setting to darken it back to the beginning. I always take the easy way out whenever possible. Damn-I thought I made it up, and here I come to find out that you're doing the same thing. :)

I'm guess the filesize would have a slight deviation to it

Nope-I aim for 701 MB with 3 episodes, and that's what I get. I have found cases with live action movies where if I put on some strong denoisers-cleaners for only the second pass that I may have up to a 1% smaller filesize.

even if it is true that you wouldn't be able to reap thefull benefits of the "standard" 2-pass nandub

I've wondered about that, but I've compared the results doing it both ways, and for the life of me, I just can't find any differences, although I admit there may be some. Sure saves a lot of time, though.

(even if each episode has completely different material, the lengths are all roughly the same, usually)

I'm not sure I understand that one completely. If you mean that the final encoded file sizes wind up the same, then I've found huge differences in file sizes among different episodes of equal quality. This depends mainly on the amount of action in the episodes. For example-Ep 23 of Cowboy Bebop, entitled Brain Scratch is filled with computer screens and TV watching, and the animators make them very noisy. And the file size wound up being huge compared to others. Episodes with them travelling through the Warp Gates a lot also eat up the bits. So how do you ensure even quality for episodes within the same CD? One way is to do the first passes for the 3 episodes/CD (or in your case 2/CD),load the stats files and then jiggle the file size-bit rate among the 3 until the 2nd Pass Compared To First Pass percentage in GKnot is roughly the same. I get signifigant differences in relative file sizes doing it that way. But again, that may not have been what you were talking about. As an experiment, you might load a few of your .avis from the same series in Nandub and then go File Information-DivX Quality, and see if the numbers given vary much. Of course, with 2/CD, even at 640x480, you may be getting numbers all in the 80s and 90s.

Anyway-I'm enjoying discussing things with you Wyldchyld. OutPinged-where are you when we need you? You like talking about these things (and bringing us back down to earth).

[Dali Lama]

Thanks WyldChyld,

I read those articles...good stuff, especially the second articles. But from that article he says to do automatic for anime episodes and manual for anime movies? Do you agree?

I have also been trying out Decomb and IVTC 4. I think Manono likes Decomb better, but I have found IVTC 4 to work well for anime. Give them both a try if you have time.

http://sauron.mordor.net/dgraft/decomb.html For Decomb

http://www.geocities.com/daxab88/ For IVTC 4

And I will patiently wait for your clip. (you don't have a clip of Rurouni Kenshin by chance? That's my favorite! and the one I will most likely try to encode, when I get back my DVD. Thanks :) )

P.S.: Manono, how has your Xvid encoding been going? And do you like IVTC 4 yet?

Ok bye,

Dali

Edit: To all, I have read somewhere in this forum that "Sub Rip" gives smaller files than "Vob Sub" (used in Gknot). If so, how is this possible? And WyldChyld, I would like to make fancy looking subs like the ones I d/l. You were talking about it in your post, but which program does this for you??

[ThePanda]

Originally posted by Dali Lama
Edit: To all, I have read somewhere in this forum that "Sub Rip" gives smaller files than "Vob Sub" (used in Gknot). If so, how is this possible? And WyldChyld, I would like to make fancy looking subs like the ones I d/l. You were talking about it in your post, but which program does this for you??

SubRip can convert the subtitles to text format, so the file size is smaller. This is good if you want to edit the subtitles if they are in broken english or something (Someone please edit the Yu Yu Hakusho subs so I can understand them, I started to do it but gave up b/c 112 episodes is too much to do alone :))
But I think .txt subtitles suck otherwise because they often mess up the formatting of the text.

[OUTPinged_]

Decomb vs ivtc4 debates:

the thing neuron mentioned just settles it. ivtc4 results may look better if dvd was mastered from VHS master tape (that happens often to r1 dvds).

tmpg gets you best results sometiimes, but you have to be really good with it (10 encodes at least). And if you are postprocessing it, that means you arent any good with tmpeg at all.

So use decomb well and it will pay back with excelent results. Just dont be afraid to mess with settings, some encodes require blend, others interpolate (this one is critical). there are some encodes where you just have to use that 30fps mode with smooth pans (Thanks DG for this one, you da man!).

poop: u wont get any good results with 16(or less)-0-0 anitishit settings as somebody mentioned. I'll be a polite boy and wont write here this guy is just talking shit and havent tested anything at all. When i have run into that problem, i had to raise settings to 21-1-0, it became better. If you use this, you will have to check for places where codec and antishit both suck ass adn are producing streams of drf32 frames. Mess with them separately. The first thresh depends on subtitle sharpness (more sharp subs tend to produce more shit). White subs woth black borders are even more evil, dont use them.


And yes, SVCD suck =P

wyld: about your position on cropping: dont be afraid man, even 2% off are tolerable. beter crop more and more and more. be more evil on these encodes, it helps alot. dont trust anybody, alot of people here dont care about results (ie heavily filter output).

on youre filtering settings: you talking here you use up to 12 threshs. but think yourself, are they filtering enough? it doesnt give you anything. even strong noise doesnt interfere with watching stuff, it only makes a movie to take less bits. now make a check on how much profit you are getting from using these filters :/

gah, it look like i am beginning to post alot of shit too. (noone reads that anyway) But you guys begun first :P

so if you have read till this place, then know all posts above are crap, and use only mine recomendations. not that they are any better, but at least they are 133t. so it makes them automatically only and the best ones. j/k man.

hmm.. what else you guys discussed here....

doh, i told ya warpsharp sux. even poop himself wrote that after all.

and mamono, if you hitting filesize that well, that just means you didnt edit your results at all. thats not good or bad, jsut dont be proud of it :-P 3-5% off filesize dont change anything really.


dali lama(see? some ppl are realizing how good they are finally): no, subtitles dont change anything at all. i got 2% difference at most at first pass. different soft gives identical subs. color/font dont matter.


THE POINT IS: THIS THREAD SUX!!! PPL ARE POSTING SHIT HERE(me)!!! DONT USE ANYBODYS ADVICE(use mine)!! DONT TAKE ANYBODY SERIOUSLY(xcept me of course)!!


disclaimer: if you think of some part of my post as of offence then you are right, it _is_ offence. flame here, thats a flaming thread, not about ripping as one may think.

\o/ (c) OUT ______o/__L,,,,,______

[manono]

Ahhhh! OUTPinged! You worry me. You're making no sense. Are you eating enough? Getting enough sleep? Hitting the pipe too often?

Good to have you back man. Keeping us on our toes. Adding some seriousness to the thread.

Is it time to start a new thread? Call it High Quality Anime Ripping-Part 2. Or maybe Anime Lies And Other BS.

[Wyldchyld]

*Soapbox*: I for one don't think this thread sux or that it's a flaming thread: not everyone here is saying "this technique sucks, this guy's full of shit," etc. There's alot of knowledge being shared, and alot of constructive criticism, too.

OUTPinged_: I gotta hand it to you, it seems like you know your stuff. There's some pretty decent knowledge in your posts... but then you go and spoil it by cutting up the thread. Kinda makes it hard for me to seriously take any of what you say... So tell me: in the end, what's your point? You're suggesting some stuff to me, but then you're sayin' not to trust anybody. So then why even post your advice here? :confused:

Manono: I'll be honest, nothing i suggest here will ever originally be my idea, i just work on a trial and error basis. I tweak & perfect things, i'm not really an innovator. I must've picked up the brightness/contrast/levels/hue thing either on some webpage i read, or by pure fluke :), so i don't want any credit here. LoL for all i know it might've been from you. What i meant by the source file lenght is the length in time (hh:mm:ss.mss); even if 2 cowboy bebop episodes will vary significantly in terms of bit demand, a quick tweak'll bring 'em back to roughly the same size, like you mentionned.

Dali Lama: I use VobSub V1.53 myself, and i think that's a really old version of it: going to try another one today. I don't know anything about filesize differences between using it or SubRip, although i know SubRip gives you more options and therefore control, options which i find i don't really care for even if some are really neat, so i've stuck w/ the quite functional VobSub. I'd say try both methods, they're pretty easy to figure out. Oh, and i'm afraid I can't really provide you with anything too exotic in terms of clips, i stick mainly to the classics. I'll probably hook you up w/ something from Gasaraki though, since that'll probably be my next project.

The mention of using auto-IVTC for series', manual for movies, is simply in terms of time consumption. Manual IVTC for an entire series can get long and tedious, even if i've found it quicker than waiting 45 minutes for the auto IVTC to finish; manual gives better results too. I must apologize though, i've never tried Decomb, i meant that i've tried Decimate. Giving decomb & IVTC 4 a try is definetaly on my to-do list now, there seems to be alot of praise towards it.

Wyldchyld :devil:
Current Project: Ghost In The Shell, Subbed, 697-702MB, DivX 3.11A Nandub SBC, 608x336-640x352.

EDIT: I just purchased DivX 5.0Pro, am toying with it right now: will post my impressions of it later on!

[Crucio]

hey, outpinged, if your as good as you say, goto #nandub on IRC and We'll be the judge of that. MAybe exchange samples and if you are a great encoder we will hook you up. Were always looking for new people who dont suck

[Crucio]

i forgot, the server is Dalnet, also, if im not there, just tell an op that Crucio sent u, or say u want to send a sample, whatever

[Crucio]

one more thing outpinged, its 1337 not 133t

[poopity poop]

False, SVCD's do not Suck.

I just bought a 3 disk DVD player, and basically switched form making SBC DivX, which is a pain in the ass to output to TV, to just plain making huge SVCD files, who cares, vbr mpeg-2 compression is great, its just as good as SBC DivX, and we can argue until we are blue in the face, the fact it, for people who have DVD players...its more practical.
I'm sure you've wanted to show anime to your non-anime friends..you like sitin' in front of a moniter, or having your comptuer close enough to the TV, or setting up some long cord with some sort of power cupling for signal loss..just make an SVCD with CCESP and stick the thing in the drive.

note: this is in the nandub section...but...its turned into an anime thread...so..yeah whatever, SVCD's are practical

[OUTPinged_]

Crucio: ok, i'm there ready for some fun. get in too and pm me.

lol @ topic: "we are the best encoders. get op status when you are skilled enough."

and its "177t", not "1337" :P tho it depends on how much leet/readable you want it to be.

poop: looks like we have lost one more ripper to svcd. =) anyway, its just you who decide, are you ok with svcd quality or not.

[ThePanda]

Can someone tell me, if you use SVCD to be played on a DVD player, do you still have to deal with deinterlacing, or can you just leave it interlaced and it will look just like the original DVD? I just ordered a dvd player, and although I don't have a problem outputting divx from the computer to the tv, I thought SVCD might look nicer because there would be no interlacing artifacts?
It sucks that you can't use AC3 audio though, right?

I would like it most if I could store mpeg4-type files on my computer with the AC3 audio and not have to deal with deinterlacing, then just output it to my tv somehow and it looks like the original (no interlacing stuff visible..). is that possible or not?

[poopity poop]

yeah well havn't lost me yet... SBC encoding is still superior being that you need approximatly half as much info wtih DivX than with mpg. But for me right now...SVCD is fine because of a big screen TV he he.

INTERLACING...
Yes you can leave the leaving interlacing works....sort of. Here's what I mean. I edited some video for my church. I captured the video total interlacing. Then I aded titles and junk using only huffyYUV so as not to lose anything. For some reason my computer coudln't keep up with a fullscreen real-time huffyYUV decompression when my TV out was on. So I accualy had to compress it using 6000Kbps DivX(which is for all intents and purposes lossless, arguably). The DivX file did have interlacing, but because the bitrate was so freaking high, the interlacing did not get smeared, and when recorded using wmp with tv out to a VHS tape, it played fine o a TV, with no interlacing. So that really only answers your question half-assed just thought people would find that interesting.
Anywa to answer your question: You really can't compress interlaced video because the codec will treat that like details and smear it. The only way to raelly do that is to have such a high bitrate that it doesn't smear(which is what I did). But that's not practical for storage purposes.
When encoding to SVCD the same principle applies. You really have to de-interlace it to compress it. then when you mux the audio in, it acually adds the scan offsets. Basically to sum everything up....you always have to deinterlace your source before compressing it, unless you are using extremely unreasonable bitrates(even then its just a hack-job).

[Crucio]

OUTPinged_ sorry i wasnt in nandub


im there now if you want to talk

[ThePanda]

poopity: thanks for the detailed reply

[neuron2]

Poop, you're not correct about always having to deinterlace to do compression. The MPEG2 format specifically supports compression of interlaced material. It can simply compress the fields individually as if they were frames. Did you know that services such as DirecTV broadcast their material (telecined as well as straight video) as MPEG2? SVCD uses MPEG2 so there is no need to deinterlace as long as you do the MPEG2 encode with interlacing support enabled.

[OUTPinged_]

heh, for some reason interlaced svcds are extremely rare and people prefer to have badly ivtc'ed encode instead of pure interlaced svcd

[poopity poop]

That's very interesting about directTV using mpeg-2 compression. Thanks for the info Graft. I guess my post was more just re-stating if you comress an interlaced DivX frame, it treats it as detail and will smear it, that's all :)

[triffid]

wow, this threads still going :D

anway, read about SVCD, svcds are great if you gota dvd player from a company that doesnt bow down to the almighty Hollywood sign (ie apex). But, we gota Sony dvd player here...gahhh no svcd for us, unless theres a special way to get it to work.

Anyway, speaking of IVTC I have been happy with decomb since I tried it, I really like skipping TMPG when i can. However, I do turn back to the tried-and-true TMPG when necessary.

ps I encode for a channel on dalnet.
#shinseiki-illuminati

yes, it is hard to spell :P

[daxab]

@triffid

You can run TMPG and then save a .tpr file and then use TPRIVTC to load up the .tpr file in a .avs file -- if such a thing is useful to you. (I posted TPRIVTC on another thread.)

[takeru]

daxab:
i don't see tprivtc anywhere. how about a link? all the threads i see are the ones you talking bits about it not being done yet.

[poopity poop]

yes I link would be great, I would love to add that method to my arsenal, please whre is that plugin?

[OUTPinged_]

lol, tpr files are working in avs, just use internal tmpeg's fapi plugin.

at least that was the way i was messing with tmpeg for a last 4 months.

why need more plugins and stuff for that?

poop: :-)

[daxab]

Hm, it took me awhile to find this. The search facility seems to ignore attachment names. Here's the thread:

http://forum.doom9.org/showthread.php?s=&threadid=15802

[poopity poop]

he he I just clicked on the number 150 (prolly now 151) in replies under this thread. When you do that it brings up a window and tallies the number of response per person.....





I'm WINNING, nanny nanny BOO BOO, stick your head in..POOP!



anyway....yeah...just found that option and thought it was cool

[poopity poop]

Let us compare nandub vs CCESP SVCD's. I am playing the devils advocate here :devil:, and trying to get a rouse out of you guys for some good conversation...be nice to me.

We shall start with 0 points for nandub, 0 points for SVCD

Fact 1:
It is possible to produce a 2ep/CD SVCD encode (~400mb's), and a 3 ep/CD nandub encode (~233mb's), and have the same comparable quality. I hope none of you will argue with me on that one; if the CCESP encode was done correctly with 3 or more passes, and the nandub was set up correctly the quality is the same...maybe even the mpg is of higher quality (debatable with no side winning: mute discussion)

1 nandub, 1 SVCD

Fact 2:
One can play an SVCD in most DVD players and basically have DVD quality, nandub you can not.

1 nandub, 2 SVCD ( I think it should be 3, because that is such an ENOURMOUS plus, but...)

Fact 3:
If one does not have a DVD player, one can download Elcard video decoder (http://www.doom9.org/Soft21/Filters/MPGDec.zip) and stream the mpg file from the CD to play on the computer, much like one starts an .avi from the CD.

2 nandub 3 SVCD

Fact 4:
Downloading 800 megs for 2 episodes is more than downloading 700megs for 3. SVCD defense (my opinion): Hey...quality is the only thing that matters, if people can't handle the bandwith get a crappy encode instead.

final score:
3 nandub 3 SVCD

If SVCD's can do everything that nandub avi's can do, PLUS play in a DVD player -- Why are people so opposed to SVCD's? Let's here some opinions


:devil:

[Inindo]

1) You can do Sub-Dub with Nandub
2) People need computers to download the stuff with, and so should be able to play them as well. Now there is a DivX player for Macs and Linuix.
3) Movies, where the movie is so long that file size becomes more of an issue (i.e. Mononoke Hime)

[ThePanda]

Although I haven't forced myself to spend the money yet, I'd say DVD-R is the best format. It's finally becoming affordable but I'm going to wait a bit longer.

[poopity poop]

Inindo
You...really didn't answer the question did you?.
1.SubDub..what does that have to do with anything
2. a DivX player for mac and linux..What? I was tlaking about being able to play it on a TV
3and your sentance fragment just came out of no where:
Movies, where the movie is so long that file size becomes more of an issue

Your entire statment was a non-sequitor.


ThePanda
While you didn't really answer the question, you bring up a good point, and is worth responding too he he:
Yeah DVD-R will be the best. When they become just a little more affordable, I will burn SVCD format onto the CD-R's with 2-3 movies on the 4.7gig and like 5-6 movies on the 9gig DVD's. That will be sweet

[ThePanda]

Are you allowed to use AC3 audio in an SVCD movie? It would be sweet if DVD players could play MPEG4 or something and you could encode it like SVCDs without having to worry about the interlacing. I think that is the next DVD technology that is coming out.

[poopity poop]

no unfortunatly SVCD's ONLY can have mp2 audio. Trust me I've tried muxing mp1, mp2, mp3, and ac3...and they accually do not work, although my DVD player can read mp3's(individually) and AC3's(DVD's).
And yes that is the next DVD technology that is coming out from the news section on doom9 a couple weeks ago...that will be cool as hell.
But most liekly it will only support 720x480 reoslution and interlacing...and whatever so like we most likely won't be able to play our old DivX3 movies in there.
There is a player coming out for dreamcast...what dreamcast...yes dreamcast. Its called dreamplayer, I had a link somewhere, but right now it only supports super small resolutions and small bitrates, its under development

[OUTPinged_]

poop, all your statements above are true only for 4:3 aspect ratio video. wide screen movies suck ass in svcd.

here are some statements from me :-PP :

SVCD strong points:

1. easy to use(no compatibility problems)
2. somewhat easyer to produce a good rip than with nandub. (debatable)
4. plays with dvd players and slow pcs
4. better quality on interlaced content


divx strong points:

1. better compression, 50% smaller files for same quality.
2. dont have to be burned to watch.
1. clearly superior for encoding of 1.85:1 and expecially 2.35:1 content
2. a small part can be downloaded and watched before getting all. (svcd usually coem in rars)
4. evolving (better and better a/v compression)


so in both cases we have 4 advantages. so just pick what you need.

and remember, divx rips are MooPolice approved!

[poopity poop]

ha ha well you need to take a writing class.
You give no arguments for your stements and are false on some of them:
the main point is that....aspect ratioed movies suck ass in SVCD format? WHAT? How can you say that? What is your proof, WHY WOULD THEY? All you have to do is get the aspect ratio correct and it encodes the entire frame..its not harder to do, or produces any crap.. It looks great! If you don't beleive me download any of the latest DVD encodes.
A small part can be downloaded and watched:
Um...no. You can't download part of a DivX avi and watch it. It just can't do that. Of course most SVCD's do come in rar's but I distribute mind as one large rar. works fine.
has to be burned to watched SVCD's.
No, extract the mpg using vcdgear/vcdgearGUI, and double clikc on the file.

Here's the things: i've done both, perfectly, I know the ins and the outs of both, and I don['t want to get on your case or anything, but it seems to me that all you know is nandub, because the statements you made just aren't true. I mean, some thigs are debatable(quality,i ssues and what not), but 16/9 and 2.35:1 AR work fine, you can sxtract the mpg from an iso(sometimes gorups distribute JUST the mpg), and you CAN'T watch part of a DivX file
(well you can recontruct the missing index block by direct stream copy audio and video, and save it, but...whatever)

[OUTPinged_]

in divx you compress wide stuff in at least 640x resolution.

most divx 2cd encodes go as 672x or even 704x. the movies i had were compressed at 480x horisontal res.

i didnt mean it gets screwed or something. it just looks more blurry on SVCD. may be taht i have badly encoded svcd, but i watched 4 movies already and all looked worse than their divx encodes.

bah, you need just to check a picture and if it is ivtc'ed, so small preview is enough. you cant preview a 15mb rar of svcd.

a couple of says ago i have watched newer hellsing ep which was 7% from completion and even w/o rekeying it played ok in vdub.


i can extract mpg from iso, that is true. dump that :-) though its not like just playing an avi :P

disclaimer: i'm by no means am svcd expert. i am judging them only from users standpoint. and i watch stuff on my PC, so SVCD are no go for me.


(and these spelling errors are MooPolice approved, so... :PPP)

[poopity poop]

I am by all means an SVCD expert, AND an SBC encoding expert (at least I'd liek to think I am). and I'm telling you from an objective viewpoint, when I encode an SVCD it looks better than SBC DivX on my TV(using s-video, or component or optical). Even with DivX on full decoding processing priority...you can still notice miscolorings in large areas of color. A properly encoded SVCD simply looks like the original movie on the original frames...is my opinion.

[OUTPinged_]

sorry, but i dont watch anime on TV (only checked how it looks like at my friend and it looked like crap)


for PC divx looks better :P

although it depends. i am pretty confident in my divx viewing setup but i can not tell the same for my svcd setup.



(divx is still 50% smaller :PPPPP)

[ThePanda]

if you use SVCD can you use Temporal Smoother or will it mess up the picture because of the interlacing lines?

[OUTPinged_]

sure you can, but you have to use a version which is adapted for interlacing.

[ThePanda]

do you know where i could find that version?

[OUTPinged_]

tmpeg's one is afaik.

[ThePanda]

oh but there's not one for avisynth or something? thinking about it, would temporal smoother smooth together the blackness from the interlace lines into each other as they jump back and forth from frame to frame, making the rest of the picture look worse, or would they not have a bad effect like that? hmm maybe i should just leave temporal smoother off if there is no easy way for me to use it together with dvd2svcd..

EDIT: I turned up TemporalSmoother in the .avs to a really high setting, then stretched it out vertically in VirtualDub and I could clearly see some interlacing in both fields where it blended the frames together, so I guess it doesn't work right for interlaced movies (as I suspected).

EDIT2: Hmm, it seems that the original frames are like that too at that spot. So I guess it wasn't because of temporal smoother. Maybe it's because you're not supposed to see one of the half-images or something?

[geoffwa]

Some DVDs, like The Castle Of Caligostro just make me want to quit ripping altogether.

For people who haven't seen it, the material appears to have undergone a conversion process like so:

Film -> NTSC -> PAL

It shimmers, it gitters, it ghosts, it does it all. There's no magic cure for crap blended fields.

[OUTPinged_]

lame.

ripped it. (r1) no problems at all with default decomb 191 setup.

if you are ripping pal dvds and they are blended, then nothing can be done. no curent tool will help you.

just give your encode more bits and watch the blended output.


(for real castle of cagliostro encode (by me of course) check SSX 2CD version =) Its quality is even a MooPolice approved!)

[triffid]

ack! these ranma season 2 dvds that i just ripped are evilly interlaced...tried a host of settings on decomb as well as auto IVTC in TMPG and i still get incredible amounts of ghosting...im no newbie and this is killing me - this is an SOS, any info not already on this thread on IVTC is requested. Especially decomb.

PS: hmm maybe i shoulda started a new thread, but this one is so nice i dont wana leave it behind, and as this is an anime encoding question...

PPS: Dont you just hate it when they stick each episode in its own set of vobs? And then encode it at a super-low bitrate on top of that? I expect better from Pioneer and Viz. For shame.

UPDATE: I'm managing to manual IVTC in tmpg but its driving me nuts with the pattern switching several times each minute of video. I dont know if i can take doing this 18 times (once for each ep in the season)...

-triffid
doing his part to keep this thread near the top of the list.

[ThePanda]

I'm not sure if I'm posting in the right place either, but as I'm talking about anime, if you are making an SVCD and leaving it as interlaced, if it is a 23.976fps framerate from the DVD, do you encode it at that and use the pulldown.exe with it, or do you just encode it at the 29.97fps? Would you lose any frames that aren't duplicates by going to the 23.976fps? And would you go from 30 to 24 by using Force Film? Sorry for all the questions, I couldn't find much info on encoding while leaving it as interlaced, although I don't know why you would do it any other way.

[OUTPinged_]

tri:

use interpolation at low thresh.
be sure field order is right. decomb 1.91 output was good already.
download random hq ep and check how does ivtc'ing look there.

the r1 dvds arent producing that good output. get your hands on r2 ones :-)

panda: if you want to leave it in interlacing mode, dont ivtc. be sure your svcd will have "interlaced" flag through the whole ep(dvd2avi).

FF drops fps to 24.

[triffid]

thanks outpinged

i think i tried interpolation but couldnt really figure out the thresh settings...im using the latest decomb, do you recommend finding the old ver?

anyway i did manual but it was mind-numbing...im really annoyed with pioneer/viz on their ranma 1/2 s1/s2 dvd releases, but oh well. Could have been worse...somehow. I had to filter the heck out of these to get them to look good, get rid of mpeg2 artifacting etc, and still my final encode doesnt look all that great.
oh well...

[OUTPinged_]

try to set thresh at 5

i checked some group's ranma eps and it looks like you cant do anything.

a hint: dont use filters on that. it will make thins look only worse.

[triffid]

outpinged: thanks for the settings, i finally got around to trying them out and they seem to be doing a much better job of pattern matching now. However, there are weird effects in certain frames (namely, the ones where the bottom half of a frame is interlaced while the rest is OK) giving some frames a squished effect...BTW i agree with you on the filtering, gona try again running just a light denoise to help with the graininess a bit.

[OUTPinged_]

>However, there are weird effects in certain frames (namely, the
>ones where the bottom half of a frame is interlaced while the rest
>is OK) giving some frames a squished effect...

You cant do anything about these. Just leave them alone. They do look a tiny bit smoother with blend postprocessing.

here is a list of settings one guy used on this:

post: on/blend
gthresh: 50 thresh 15 dthresh 5

these are confirmed to work and give a tolerable output.

[triffid]

another extremely weird circumstance: for some reason nandub's anti-shit and min quality settings hate the subtitles this anime has. When the subtitles are on (only when they are displayed in a frame, everythings fine between subs), it kicks in heavy and causes a 50-90% decrease in encode speed as well as creating macroblocks all over the frames with subtitles. Upping min quality to 20 got rid of the macroblocks but didnt help the speed at all. Running with just anti-freeze on, the encode came out just fine. Anyway, just wondering if anyone has seen this before. I find it very weird.

and this on top of all the other problems with this anime...

[OUTPinged_]

a color of subtitle is guilty.

if you are using pure white/yellow with pure black borders, then make a color a couple of notches darker and a border a couple of notches brighter.

still, wonrking antishit = slow.

the bad thing about antifreeze is that you will end with macroblocks anyway :-)

[triffid]

actually, i was surprised with the final output, no noticable macroblocks. I'll experiment with the color some more, im using the default sub colors (which happen to be semi-transparent yellow with black borders). I had already experimented with colors that i had used before successfully, so i'm thinking that the font used might be partially behind this.

btw i also tried out subrip, planning to convert to SSA and use subtitler to put the subs in (or maybe use ogg subtitle stream) but this font seems to be particularly hard on the OCR, would take at least an hour to edit them back to how they should be. Its still an option tho.

thanks again outpinged

[Dreassica]

Im new to anime ripping, but im still gonna resurrect this thread.
I read this thread with a lot of interest as i am trying to rip anime myself. I was wondering if someone has some new settings for SSm HQ and 2d cleaner? I tried tweaking its settings but i didn't get satisfactory settings for my PAL DBZ rip. Maybe someone doing DBZ as well has some settings that work for him. I know, those settings may not apply to my case, but then i have something to work off again, because reading this thread showed that one time preferred settings become obsolete and replaced by new ones and as the thread is silent for 2 months now i guess the settings have changed again.
any pointers will be appreciated!

[Hanty]

Settings becomes obsolete once you finish an encode, when you start a new one you do it all over again. One of the most basic but still most hard-to-accept fundamentals of encoding that newbies get stuck on is that what works on one encode might be a disaster for the next. There are no general settings that will always do you well.

[Dreassica]

I wasn't looking for the "perfect" combination of settings, but more of a general set that produced satisfactory results, that i can use to work off.
Reading this thread i saw that at one moment a certain combo seemed to be favourite amongst several people only to be replaced by new ones weeks later and even those changed or were adjusted.
Like the 2d cleaner which preferred settings was adjusted downwards after awhile because their effects weren't that pleasant as it was first made out to be and because the thread was silent for more than a month, i thought they had changed in the meantime and that reinitiating the conversation would shed more light on it>
Hope i clarified myself a bit now and u don't think i expect any perfect settings.
Any comments on this subject are welcome!!;) ;)

[MasterYoshidino]

hate to resurrect this thread but i am just addressing what ppl should have learned about some of the plugins mentioned
and to also bump it for the n00bs to read =)

When encoding an anime, no one setting is best. This applies especially to encoding in SBC or DIV3 codec. Some general settings apply though. Most anime is telecided from 24 to 30 fps and needs to be inverse telecided. You can either use ITVC or Decomb. They both do the job well , as long as you have the recent dll's. Decomb needs to be correctly tweaked for the source, to prevent ghosting, which occurs when incorrectly ivtc'ed, leaving behind fields that were not dumped by the plugin. (Fields are interlaced 1/2 frames that when played quickly enough, appear to produce whole frames. Two fields can reconstruct a frame)

Filtering is a touchy subject as some believe that 2d cleaner is better than sshiq (Smart Smoother HiQ). Actually it depends on your goal. 2d cleaning blurs the screen, thus fine details are lost. This is though faster than the edge preserving sshiq (which when I use in an avs script with C3D (convolution 3d, helps simplify the image for better smoothing) and DCT filter (to help smooth mpeg2 compression blocks) I encode at a fast 8~10 fps on a 2.4 GHz Pentium4 (shows how slow the sshiq is).:D

If you want to encode for details, sshiq is the way to go, otherwise a simple 2d cleaner will suffice.

Temporal smoothers can help smooth as well and can produce just as good as results as spatial smoothing. The only importance is it will _not_ smooth very wide and similar color areas as effectively as spatial smoothers. They can produce nice results but can destroy the mpeg2 look as generally compression blocks are filtered and can cause some objects to appear blurred or the background to have luma noise , which spatial smoothers better handle.

It takes several encodes and settings to see how the codec will compress various settings. Never assume that your 1st encode will do it, encode several times with various settings to see which is more pleasing/asthetic to the eye.