how I handle interlaced video [Archive]

sh03z

29th April 2004, 23:36

I wanted to gather some thoughts that you all have on interlaced video.

From what I've read, there is truly interlaced, and just progressive sped up to 29.97 fps...

Is that necassarily true?

What Im really interested in is how everyone handles their interlaced video...

For example, if I was recording a special newscast off of TV, obviously the source would be interlaced to comply with my TV...
Personally I would leave it interlaced and use convolution3D on movieHQ in most cases.

I would do the same for a DVD source, as deinterlacing would dull the picture.

Only thing is, when I view an interlaced DVD "backup", on my DVD player, I see more lines than on the original...

However, I've noticed these lines are mooshed together, and are practicly invisible on my TV/DVD player.

So, I'm wondering, what causes interlaced DVD "backups" to retain those fine lines. Also, how do *.avi recorded TV programs retain the lines if the source appears flawless on my window using virtual dub as my capture program?

Does this have to do with analog/digital conversion etc.?

Thanks,

-Mike

(I hope this is the appropriate forum division =0 )

rfmmars

30th April 2004, 00:41

I don't know where you heard that deinterlacing would dull the picture.

First visit www.100fps.com for a complete coverage of deinterlacing.

I run a telecine service for very old 8mm & Super 8 Kodak home movie film.

The final product is 59.985 fps progressive to the DVD. I am going to give you two scripts, one is a AVISYNTH, and the other is a VD .VCF file.

loadplugin("c:\windows\system32\mpegdecoder.dll")
mpegsource("e:\KanasCity\xmas67.mpG")
#complementparity
separatefields
#bob
#SWAPFIELDS

This produces 59.985 fps interlaced NTSC.

Next is the .VCF file........There was a mistake, and the top source line was removed.

VirtualDub.video.DeleteComments(1);
VirtualDub.video.AdjustChapters(1);
VirtualDub.video.SetDepth(24,24);
VirtualDub.video.SetMode(3);
VirtualDub.video.SetFrameRate(0,1);
VirtualDub.video.SetIVTC(0,0,-1,0);
VirtualDub.video.SetCompression();
VirtualDub.video.filters.Clear();
VirtualDub.video.filters.Add("Auto-Levels");
VirtualDub.video.filters.Add("brightness-contrast-gamma [1.0]");
VirtualDub.video.filters.instance[1].Config(0, -27, 65, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0);
VirtualDub.video.filters.Add("Alparysoft Deinterlace");
VirtualDub.video.filters.instance[2].Config(2, 16;
VirtualDub.video.filters.Add("brightness-contrast-gamma");
VirtualDub.video.filters.instance[2].Config(0, -27, 93, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0);
VirtualDub.video.filters.Add("radial luminance [2.1]");
VirtualDub.video.filters.instance[3].Config(17, 17, 0, 0, 0, 0, 0, 0, 255, 0, 0);
VirtualDub.video.filters.Add("RGB to YUV");
VirtualDub.video.filters.Add("xsharpen (1.0 beta 2)");
VirtualDub.video.filters.instance[5].Config(255, 125);
VirtualDub.video.filters.Add("YUV to RGB");
VirtualDub.video.filters.Add("shadow smoother (0.7)");
VirtualDub.video.filters.instance[7].Config(8,1,0,1,1,1);
VirtualDub.video.filters.Add("chroma noise reduction (1.1)");
VirtualDub.video.filters.instance[8].Config(1, 192, 35, 0, 256, 47, 0, 256, 47, 0);

This is the deinterlacer that I use, and it's defult setting

[VirtualDub.video.filters.Add("Alparysoft Deinterlace");]
[VirtualDub.video.filters.instance[2].Config(2, 16;]

This VCF file will load without error in VDMOD 1.5.10.1 whether a video clip is loaded or not. Make sure the compresor is the one that you will be using or it that line. This is Leadtools new MJPEG

I don't use all the filters at one time, but this chain produces the best looking DVD from VHS & telecine sources.

It results in super smooth playing and sharp video without that 'Digital" look.

richard
photorecall.net

sh03z

30th April 2004, 01:39

anywhere I can get those extra VDF's?

I'm using CCE SP =(

EDIT
I tried using virtualdub with just the deinterlacer you suggested.
very good results

rfmmars

30th April 2004, 04:07

most of those .VDF files are on Donald graft's site. http://neuron2.net/index.html

I don't know if I have everyone out there, but I am close. If you can't find what you are looking for e-mail at rfmmars@cox.net

That deinterlacer is really nice.

richard

FulciLives

1st May 2004, 23:34

Originally posted by rfmmars

loadplugin("c:\windows\system32\mpegdecoder.dll")
mpegsource("e:\KanasCity\xmas67.mpG")
#complementparity
separatefields
#bob
#SWAPFIELDS

This produces 59.985 fps interlaced NTSC.

HUH ?!?!?

No I haven't tried it but I can't imagine for the life of me how this AviSynth AVS script would work as indicated?

What am I missing? :confused:

- John "FulciLives" Coleman

*** EDIT ***
OK I tried it and it works though I'm still not sure how hehehe but my capture went from 720x480 29.970fps to 720x240 59.985

What if I want 720x480 59.985fps ?

Also I am slightly confused as to how CCE will handle that frame rate?

stickboy

2nd May 2004, 00:05

Originally posted by FulciLives
OK I tried it and it works though I'm still not sure howExactly what do you think SeparateFields does?
What if I want 720x480 59.985fps ?Get a bob interlacer (e.g. DGBob, but there might be better ones around; I haven't been keeping track).

FulciLives

2nd May 2004, 00:28

Originally posted by stickboy
Exactly what do you think SeparateFields does?
Get a bob interlacer (e.g. DGBob, but there might be better ones around; I haven't been keeping track).
Well I thought SeparateFields just throws out one frame without changing the framerate but I guess that is when you use () after it.

Anyways I'm not overly found of this method. The video looked very "jumpy" to me doing a preview in VirtualDubMod (loading the AviSynth AVS script).

- John "FulciLives" Coleman

rfmmars

2nd May 2004, 01:23

Yes it's down sized because it split out the the two fields. Now you only have 240 line virtical. I use Magix's MEP2004 NLE program to rebuild it back to 720x480.

If you want to stay at 720x480 add this to the Aviscript

doubleweave.

Now you have 720x 480 interlace, so what I do is deinterlace in VD so then I have 59.985 deinterlaced.

richard

stickboy

2nd May 2004, 02:02

Originally posted by FulciLives
Well I thought SeparateFields just throws out one frame without changing the framerate but I guess that is when you use () after it.SeparateFields is no different from SeparateFields().

And no, that's not what SeparateFields does at all. Read the docs.
The video looked very "jumpy" to me doing a preview in VirtualDubMod (loading the AviSynth AVS script).When you call SeparateFields on a clip, the frames alternate between even scanlines and odd scanlines from the original clip, so of course it looks jumpy.

Kika

2nd May 2004, 13:24

@sh03z

First do separateFields, then all the Filterings you want, at last, you need weave. That's all... But be carefull on Resizings, all vertical resizing have to have done by mod2 or you will mess up the interlacing.

rfmmars

2nd May 2004, 16:20

The reason for using doubleweave is to double the frame rate. This is usefull to correct some of the panning studering assocated with many camcorders and some VCR's that to not output 29.97 fps or 25.00 fps PAL.

More a more complete explaintion goto www.100fps.com

richard
www.photorecall.net

sh03z

3rd May 2004, 00:37

Well,

I now am using Kerndeinterlace for all my deinterlacing having to do with my DVD backups of the DVDs that I buy.

For VHS/TV I am going to use the advice that rfmmars gave me.
But rfmmars: will this work if I want to put the footage on DVD???
That was what I was wondering...

The reason I don't like other techniques for deinterlacing (combing, inverse telecine) is that is leaves little artifacts in the picture in random places, and especially at parts of fine detail.

Kerndeinterlace just seems to be ONE of the best, in my research.

-Mike

scharfis_brain

3rd May 2004, 00:55

please do not deinterlace unnessesary.

If inverse Telecine is possible, I strongly recommend using it!

if you are not satisfied with telecide.decimate try smartdecimate.

deinterlacing (NTSC) Film will result in stairstepping and high detail shimmer. and stutter.

only deinterlace video and blended streams, or other irregular contents.

rfmmars

3rd May 2004, 03:26

My final product is always a DVD for my customer. As I stated before there is two reasons why I deinterlace, one when tranfering from analog and some DV.avi camcoder format,I don't use doubleweave because the editing program that I use[MEP2004] will rebuild the 720x240 back to 720x480, and the burn is non interlaced.

This almost always takes care of the horizontal studder when there is movmnet or panning.

I use doubleweave to improve the look of my telecine captures. I have made many A-B DVD copies for playback to evaluate the benefit of deinterlacing. Deinterlacing is always visualy better. Maybe I am splitting hairs, but quality is first and formost.

What kind of deinterlacing do I use? What I do is make the filtered window 4x, and watch the content, Most of the time it's just a simple deinterlace.

The downside is that a 59.985 fps deinterlace takes twice as long to process. The customer desides if this is the process, and it's he or she who decides if they want to pay an extra fee, and when it is I who's the customer, I choose it every time.

When I prove to myself that a process is better, I don't let go.

Other forum members have their opinons, this is just mine.

richard
www.photorecall.net

cweb

3rd May 2004, 19:25

I have read (and my encodes seem to confirm this) that keeping interlace will mean the video will compress less - more bits will be needed. I generally always deinterlace (unless the source of course is progressive). For PAL tv interlaced recordings (DVB-S) I use TomsMoComp which gives me great results. This is my experience plus opinion.. YMMV

sh03z

3rd May 2004, 19:52

@scharfis_brain:
sorry! I forgot to include in my humble paragraph that there was one or two instances where I did inverse telecine on particular DVD backups, and I noticed improved quality over deinterlacing.
It's just for truly bad interlaced DVDs (most?), that I deinterlace with kernal deinterlace.
For example, I just did a backup of a used DVD I bought, and it came out fine with kernal deinterlace. Tomsmocomp = terrible, inverse telecine = lines all over the place, mushy picture.

@rfmmars:
yep, deinterlacing is needed a lot of times

@cweb:
Could you possibly share your settings for tomsmocomp with us?
I get terrible results with tomsmocomp...flashing white/black all over the place.

scharfis_brain

3rd May 2004, 23:20

@rfmmars: I don't really get your stuff?

why don't you just cut & edit your Video on the interlaced basis, and do an interlaced encode to DVD?

why do you do all the pain of sepfields(), doing doubleframerate deinterlacing etc?

rfmmars

4th May 2004, 01:29

@scharfis_brain

Why do I do this, because the source requires it. How else are you going to eliminate the horizontal studder with movement when the camcorder doesn't produce the required number of frames for good production of DVD or Video tape media.

These tapes that look ok on preview capture, do not look that way after capture. The capture card is expecting a certain standard, and it is not being forfilled.

I was at a loss to solve this until I found (www.100fps.com) where it was all explained. That became the core of the filtering I do.

Maybe the problem isn't the same with PAL because of it's lower frame frame rate, I don't know.

On the telecine captures, there is a definete improvment in the look of the final product. I have kept DVD's of the same source of various filter chains I have used, so I am not drawing from my memories what was better, I can always review them at any time.

I also keep a copy of every project that I have done, and sometimes I go back and redo the project just to see if there was anything that I have gain or loss with newer filter chains.

You really can't move ahead if you don't know where you have been.

Why would I burn a DVD with interlaced mode. If I do that then I can't get the advantage of progressive scan and the studder is back.

richard

richard

scharfis_brain

4th May 2004, 11:12

Du you mean forward-backward-forward-backward-forward-backward stutter on pans an moves?

deinterlacing interlaced video should never be done if DVD & TV is the final media to output on.

this can easily be solved setting up the correct fieldorder!

avisource("yourvideo.avi")
assumetff() #or assumebff()

every app, that is able handling interlaced video has a Fieldorder switch, too! (Top FIeld First, or Bottom Field First)

please do not trust 100fps.com. he sometimes tells something that is more convusing than helping.

No Video Cam to me is known havin judder or stutter neither in horizontal nor in vertical direction.
I've never deinterlaced the Video.

For PAL it is simple to remind:
DV-Video (even analogue video capped through DV) is always BFF
Analogue Video capped with a TV-Card is always TFF

EDIT:
Why do I do this, because the source requires it. How else are you going to eliminate the horizontal studder with movement when the camcorder doesn't produce the required number of frames for good production of DVD or Video tape media.

These tapes that look ok on preview capture, do not look that way after capture. The capture card is expecting a certain standard, and it is not being forfilled.

Huh?

A NTSC Camcorder puts out 59.94 fields, a NTSC capping device records 59.94 fields. The a=same on PAL, with thew only difference only recording 50fps.

Du you mean Framedrops?

rfmmars

4th May 2004, 14:17

No it not dropped frames. A lot of camcoders just can not output the frame standards required by NTSC & PAL. This also is true for some VCR decks.

You need to put more faith on what is said on www.100fps.com. If I have a progresive scan DVD feeding a 480P monitor, I am not going to burn a interlaced DVD. If it is going to be played on a computer, I am not going to make a interlaced DVD.

Just finished a 22 hour rendering / resoration of a two hour special recorded on a S-video tape of a CBS special on 911. This was another A-B comparasion with a interlaced burn of the same source.

The noninterlaced video was by far the best looking at 59.985 verus 29.97 interlaced or 59.985 interlaced. The restoration would have been only 11 hours if I had kept the frame rate at 29.97. Why would I have wasted another 11 hours if there wasn't a benefit.

My playback decks are not junk JVC 9990u Prosumer, and a JVC BR-S611u broadcast.

I have been doing NLE for over 15 years, with customers like Discovery Channel, TLC, NBC, and NHK of Japan. there satified with the product.

The proof is in the pudding. Why do other members posting agree with me using the steps that I laid out.. The key is that you are not working with the same source that I am.

My second post was to address the wishes of FulciLives for a way to produce 720x480 @59.985 no-interlaced progressive, not 720x240 non-interlaced @59.985 fps. So how else could you do this? I would be more than happy to get the same results in half the time or less.

richard

scharfis_brain

4th May 2004, 14:59

A lot of camcoders just can not output the frame standards required by NTSC & PAL. This also is true for some VCR decks.

every camcorder or VCR puts out 25 or 29.97 fps.
(okay some 0.01 percents of deviation)
Do do not understand why a VCR or camcorder should output a non-compatible framerate ?!?
May I be, you have insufficent Capping devices?
I had never problems using a lot of VCRs or analogue Video-cams.

You need to put more faith on what is said on www.100fps.com. If I have a progresive scan DVD feeding a 480P monitor, I am not going to burn a interlaced DVD. If it is going to be played on a computer, I am not going to make a interlaced DVD.

then deinterlacing may be a solution, but not for feeding an interlaced TV.

My second post was to address the wishes of FulciLives for a way to produce 720x480 @59.985 no-interlaced progressive, not 720x240 non-interlaced @59.985 fps. So how else could you do this? I would be more than happy to get the same results in half the time or less.

Ever tried a bobber for AVISynth?

there are a lot of plugins and functions out there, that can do this.
dgbob, kernelbob, tomsbob
and lots more.
a fully motioncompensated deinterlacer is called HiCon32 form Fraunhofer.

most of them can work in realtime.

btw. what the hell is 59.985 fps?

if my math teacher didn't fail all the time i was in school,
2x29.97 = 59.94

Mug Funky

4th May 2004, 15:05

forgive my possible misunderstanding, but 720x480x59.94 is not kosher for DVD... am i missing something?

i haven't worked in broadcast (though maybe in a few weeks i will be), but i can't see how 720x480x59.94 is going to work, unless it gets a "separatefields().selectevery(4,1,2)" before it hits the DVD (and then it's interlaced again...)

btw, i agree with scharfi on 100fps.com... some of those samples look obviously photoshopped to me, but as you say the proof is in the pudding, and just because i'm not privy to the same deinterlacing techniques is not to say they don't exist.

[edit]
@scharfi:

most of them can work in realtime.

man... my system's out of date - tomsbob, kernelbob, dgbob, etc run at nothing like realtime here, unless i use tiny tiny frames.

scharfis_brain

4th May 2004, 15:11

oh, I should rephrase it:

some of them can work in realtime

zilog jones

4th May 2004, 17:38

Originally posted by Mug Funky
[B]forgive my possible misunderstanding, but 720x480x59.94 is not kosher for DVD... am i missing something?

i haven't worked in broadcast (though maybe in a few weeks i will be), but i can't see how 720x480x59.94 is going to work, unless it gets a "separatefields().selectevery(4,1,2)" before it hits the DVD (and then it's interlaced again...)

Grr, why do some people seem to think interlaced video is the tool of the devil or something?

Look, ANY video recorded from a normal video camera (not progressive HDTV or from film), be it a home camcorder or from a TV broadcast, will be interlaced video - interlaced at 25 FRAMES per second, meaning there's 50 FIELDS per second (29.97/59.94 for NTSC). On a standard TV set this gives the impression that the video is 50 frames per second.

Yes, I know interlaced video is all weird and liney when viewed normally on a PC, but that most certainly does not mean you must de-interlace it if it's for use on a TV set! If you de-interlace actual interlaced video (i.e. proper video, not something that was filmed), you're losing half of the fields! Movements won't look as smooth as they did when originally interlaced, and in a way you're losing half of the quality of the video in general.

However, for PC viewing, you could convert 50 field-per-second interlaced video into 50 frame-per-second progressive video. The vertical resolution will be halved as a result of this, though.

If you de-interlace PAL video, it'll look just as if it's sourced from film (because film is 25fps in PAL). Unless you want the footage to appear as if it was filmed, I see no point in doing this and see it as quite destructive of the content. I see it done for no reason so often in recent these times of easy video editing from your PC, and I find it quite annoying.

Can some people not tell the difference in motion quality with interlaced and progressive video or something? Then again, saying this, I can easily tell the difference between 50Hz PAL and 60Hz NTSC video just by looking at the motion.

100fps.com is really ambiguous, and doesn't tell the whole story.

rfmmars

5th May 2004, 01:03

Just because something might look suppect to you on www.100fps.com doesn't mean it is. Is his site 100% coreect, I don't know.

I do know by following his guide, I have solved a problem that I and some others have posted here without a responce.

I have also made a (" Proof of Performance Compatabily Test DVD") from all of his PAL video clips that I have download and converted to NTSC.

This DVD was encoded non interlaced, and the clips were all 50 fps of various Pal resolution formats, all done at one time, frame served all at once from AVISYNTH 2.54 to VIRTUALDUBMPEG2 1.5.4 and frame served from multible VIRTUALDUBMPEG2s to Magix's MEP2004, or known as Video Deluxe 2004-2005 in Europe.

This was done simutainlessly as one opertation, with multi Virtualdubmpeg2 frame servers, with no intermediate file. In other words a mix of PAL resolutions all at 50 fps rendered to DVD at 720x480 59.985 fps noninterlaced in one editing process.

Avisynth was filtering, Virtualdubmpeg2s were filtering, and MEP2004 was putting the project together & filtering and buring the DVD, all at the same time.

There is no problem playing these NON interlaced DVDs in any DVD player.

The purpose of forums are to help people and put forth new ideas. Nothing ticks me off more than somebody that has an opinon not back by doing the same process or for some reason thinks a website is suspect without proving so.

As far as www.100fps.com is concerned, I have been able to repeat and duplacate what he has posted.

I will always turn out the best quality videos possible with the tools that I have to work with, and my German boodline line says it's never good enough, try to do better.

Am I nuts? Look at this thread. http://support.magix.net/boards/magix/index.php?showtopic=12182

richard

scharfis_brain

5th May 2004, 07:30

oh-oh....

I do not think the final output on a DVD-compliant DVD (I don't have better words) will be progressive 50 or 59.94 fps
(AGAIN: what is 94.985 fps ?!?!?)

I hardly can believe this.
I think, the progressive fullframerate video has been reinterlaced by one of your wild mixed appz.
from 59.94p to 59.94i
the same is achieved by the avisynth command sequence
separatefields().selectevery(4,0,3).weave()

a Fullframerate progressive DVD won't play in any Standard DVD player. how should that work?
- a dvdplayer has only memory to hold interlaced frames (half height frames), there is no space for progressive ones.
- a (normal) dvd-player only has interlaced output
- for fullframrate progressive, IMO even 9.8 mbps (max. DVD rate) is too low.
and lots more.

rfmmars

5th May 2004, 08:02

As far I can tell by reading the object's properties in MEP2004, it agrees with with the frame rate in VDMOD except VDMOD doesn't say whether it's interlaced or not. Since it is being deinterlaced at that point in VDMOD, I would say it is. Also if you 4X the preview window, you can see that it'e deinterlaced. MEP2004 says it is noninterlaced at 59.985 fps. when fed to the DVD burning section of the program. It doubles the rendering time.

That program is told that it is a non interlaced file which it is. If I burn interlaced, the video has the same flaw that I had taken the time to eliminate. There is no doubt that it is what I say it is going to the final rendering, but a great looking DVD comes out.

I have burn the DVD in another program, and the results are still the same. What I will do is pull the .VOB file out and look at it and maybe it will shed some light on it.

When all said and done, It is not me that has the problem, it's you accepting that it solves a problem, and varified by other users.

richard

scharfis_brain

5th May 2004, 08:16

again: what is 59.985 fps?

provide a VOB snipplet of one of your everywhere running progressive fullframrate DVD's and we will see.

zilog jones

5th May 2004, 12:02

You sure you don't mean fields per second, not frames? Or is it really saying 59.94Hz? That would make sense, then, as vertical frequencies of TV video always measures assuming each field is a cycle - i.e. intelaced TV is always said to be either 50 or 60Hz.

I think there's some misunderstanding going on here - standard DVDs are either 50/60Hz interlaced, or 25/30fps progressive.

And again, 100fps.com is designed mainly for people wanting to de-interlaced stuff for viewing on computer monitors and converting to formats like DivX properly - you don't need to de-interlace and then re-interlace when making DVDs. Oh, I don't even know what's going on now - what are you trying to do exactly?

rfmmars

5th May 2004, 19:01

You may be correct on fiels verus frames. Give me a couple of days to look at the .VOB files.

I will do two burns, one interlaced and one non interlaced and compare them.

richard

The Test

I had a chance to look at a .VOB file. The properties were 720x480 non interlaced 29.97. So we can see that in the final rendering the frame rate is converted from 59.94 noninterlaced to 29.97.

I compared the .VOB with the master clip which has horinzontal studdering and is interlaced bottom top. I redid the entire process and burn the DVD top first and another bottom first, and again non interlaced.

Only using non-interlaced solved the problem. Using the master clip directly, there was no solution at all using interlaced or non interlaced encoding.

richard
www.photorecall.net

sh03z

8th May 2004, 00:48

I found a new toy, Decomb511

It has the feature telecide, which is an alternative to deinterlacing, as noted in the beginning of this thread.

What grabbed my attention about this avisynth filter was it's ability to remove the interlaced frames without producing a jittery studder.

I don't know what this thread has accomplished, and I'm sure that some people are sick of it, but for what it's worth:

I have this DVD I bought, it's a pulldown I'm pretty sure,
from a TV source,

However, nothing would work. I tried telecide, version 4 or something from the old decomb. Then I tried greedyHMA, then I tried actually deinterlacing it with KernDeint, and also fielddeinterlace()

What I finally came up with is:
Telecide(order=1,guide=1,post=4,vthresh=40)
from decomb version 511
Post=4 gets rid of the jitters I found in post=2 or post=3

It works really great.

Also, you HAVE to deinterlace/telecide some sources.
I tried watching this DVD I bought, on my TV, but all the interlaced lines showed up as blurrs, where as they showed up as lines on my computer screen. Therefore I must conclude that telecide, or deinterlacing, is needed to produce a proper picture at certain times.

This confuses me, because I thought the TV would deinterlace EVERYTHING...

rfmmars

8th May 2004, 03:51

sh03z.............thank you

richard

FulciLives

8th May 2004, 04:16

Originally posted by sh03z
Also, you HAVE to deinterlace/telecide some sources.
I tried watching this DVD I bought, on my TV, but all the interlaced lines showed up as blurrs, where as they showed up as lines on my computer screen. Therefore I must conclude that telecide, or deinterlacing, is needed to produce a proper picture at certain times.

This confuses me, because I thought the TV would deinterlace EVERYTHING...
I think seeing INTERLACED artifacts on a normal TV is the result of one of two things (or perhaps a combination).

1.) The way it was encoded including the way the video was "flagged"
2.) The way the DVD player read the video and/or the "flags"

Let me give you an example.

I have a Cyberhome CH-DVD 500 because of it's very nice capability to convert PAL to NTSC since I live in the USA and have an NTSC only TV.

However I have a USA made and marketed NTSC region one DVD of a Japanese anime called NOIR (Volume One)

Now this NTSC DVD exhibits INTERLACED artifacts on my standard NTSC INTERLACED TV when I play it back on my Cyberhome CH-DVD 500.

Now this particular DVD player has 3 "filter" settings. AUTO, INTERLACED and PROGRESSIVE and this has nothing to do with PROGRESSIVE output these are "filters" that work on normal interlaced composite and S-video outputs. There is a totally different option for PROGRESSIVE COMPONENT VIDEO output.

Anyways on AUTO just about everything looks normal but not with NOIR

When I change the "filter" setting to INTERLACED it THEN looks normal without INTERLACING artifacts.

Now I don't know if the problem is with the DVD itself but consider this ... most (nearly all) of my DVD discs be they NTSC or PAL ... progressive video or interlaced video ... play fine on the AUTO setting with is the default setting. There are only a few where I need to change the setting.

I'm not an expert on these thing but I do seem to recall reading that a DVD video stream can be "flagged" right or wrong as to it's progressive or interlaced nature etc. and that some DVD players only read the flags and some only look at the video and ignore the flags ... or something like that.

So I think that explains the problem. The video really is A-OK but there is a communication problem perhaps because of the video being improperly flagged along with the inability of the DVD player to "pick up" on this error.

However I don't think most DVD players have an option to change things ala my Cyberhome CH-DVD 500 ... and I hear that newer firmware upgrades even lock the "filter" to AUTO which is why I have never upgraded (firmware upgrade is mostly ofr SVCD problems so who cares).

I know some other brands of DVD players have similiar so-called filters but it seems to be something mostly regulated to those models that can do PAL to NTSC conversion ... for instance I know the JVC and Malata players have similiar settings ... though they have different names.

- John "FulciLives" Coleman

sh03z

8th May 2004, 04:21

alright, that sounds logical...

I'll try to find out how I would flag material as "interlaced" with CCE then...

thank you

Mug Funky

8th May 2004, 08:28

interlace artifacts on a tv? i've only seen them on progressive monitors (panasonic plasma screens have the crappyest bobbers ever - static scenes are okay, but motion is semi-combed somehow. this might be what you're seeing, not sure).

in CCE, just select "alternate scan order" instead of zigzag and you'll be right. also make sure your field-order is correct, or there'll be severe ugliness on the TV (fields playing back in the wrong order is more distracting than combing any day).

oh, and usually to avoid confusion, terminology for interlaced and progressive is usually a "p" or "i" after the frame rate, or vertical resolution.

like PAL is either 25p or 25i, but not 50p (unless you're going into HDTV... i'm not sure about that area).

scharfis_brain

8th May 2004, 10:10

rfmmars, please provde a short MJPEG Sample (50 frames) of your stuttering interlaced Video.

I doubt, that only deinterlacing is the way to fix that.

rfmmars

8th May 2004, 17:40

scharfis_brain

NO I am not going to post a clip.........my problem has been solved, my me with no help from this board or others, I reeived no responce when asking how to fix my problem.

After posting the solution to my problem, your's is the only negitive one, while others confirming that deinterlacing is nessesary for some materials.

You SIR do a injustisc to the other board members. If you keep your "know it all attitude", people will stop posting anything new of value. We don't come here to be put down with our ideas.

If my solution didn't solve sh03z problem, at least it got him thinking, and he found something that did work for him. We all received new information about decomb511, something I had not herard about. I who gave out something of value, has received somthing in return.

richard........who doesn't know it all

Mug Funky

8th May 2004, 17:52

rfmmars: you misunderstand scharfi's intentions.

i suspect he's on a crusade against the need to reduce vertical resolution via deinterlacing. i think he wants a sample clip to see what he can make of it. rest assured that if he finds a solution that improves on current solutions (in efficiency or quality or both), he will share it with the community here. this will potentially improve your workflow and client satisfaction, should you choose to accept it.

not posting a clip does the board a disservice in this case :)

btw, if you read the boards a bit more, you'll see that scharfi may not know it all, but he's helped us all out immensely with what he does know.

no need to get all hot and bothered over a forum... and apologies if we missed your original post on this... remember we're all learning here, and it's possible when you posted people didn't actually know how to help that situation (again, a clip speaks 1000*framecount words ;))

rfmmars

8th May 2004, 18:41

I have never posted a clip on a forum, plus none of work stations are online.

If you could give me a liitle help on posting a clip, I in return will do so. This is not a current problem, so I will have to go back though my backup to find one with this problem.

Just because scharfis_brain has a crusade againist deinterlacing, that's not problem. If I am making a DVD forviewing on my computer, why would not deinterlace. Is there a problem with PAL that we do not see in NTSC?

If scharfis_brain has a position paper on this, where is it so I can see all of this it from his point of view.

If you can help me in posting that clip I will do so. I have a website that is actived, but my web computer has bite the dust, so I can't ftp there.

richard

Mug Funky

8th May 2004, 19:05

yeah... webspace is the key really :) otherwise, simply placing the link in a post will do. if you're worried about security you can send the link by PM if you like.

hmm... i think scharfi has some stuff online about video. i can't vouch for it's quality as i don't speak german (and babelfish stops translating halfway through this page, so you get half babel-ish and half german).

there's no probs with PAL that NTSC doesn't have (having a framerate so close to film, but not exactly that of film means actors have higher voices in PAL land, and ac3 quality is quite poor from transcoding though).

hmm. i should update my bobber script. it's semi-smart - it doesn't touch moving areas, but it doesn't perform matching, so there's potentially more detail to be got out of it. for pure video this doesn't matter, but for standards-converted material, it could be better.

i might resurrect an old thread and post that script.

sh03z

9th May 2004, 02:54

Okay,

thank you all for your help.

I have concluded the following from our discussion:

First off, rfmmars you may be right about the way you deinterlace, but it isn't neccassary under one condition, that being you flag the video stream so that the TV recognizes it as interlaced.
This is possible in CCE via the "Upper Field First" option used in conjunction with the "alternate scan" option.

Furthermore, deinterlacing does dilute/degrade the picture if it is not done right. It also produces studders, etc, if not done right.

I therefore conclude that deinterlacing is not necassary if the final product is to be played on a TV system.

Again, thanks for everyone's help. I learned something =D

rfmmars

9th May 2004, 22:52

I really believe it depends on the source problem that your are working with. Going back to some of the problem clips, I can show you the section of the clip that studders, but since the only sources that I have kept are .VOB files.

Once that .VOB file has been made, I find no way to undo it at this time. However the solution that I have posted works everytime for captured clips with this problem. I remember this really bad clip, and at that time I did not have a fix. I was playing back the 8mm analog tape in my Canon Hi8 camcorder. I thought this might be the problem.

So I went back to my customer and borrowed his Sony 8 to see if the problem went away, but in fact is was worst. I even installed my Miro DC30 card, but its capture had the same problem, but when capturing, you do not see this. So this capture was repeated on three workstations with 4 capture boards, with always the same problem

I never burn a final DVD without choosing non-interlacing, and they always look better using this basic solution. It may be a part of my software chain, but it works correctly on all my workstations. I keep a log on everything that I do, so I can go back and repeat the process again.

richard

scharfis_brain

9th May 2004, 23:26

hm. sounds strange.

Once I've seen a field problem using an Bt848 based card with the manufacturers driver. (Using Btwincap driver & Bt-Tweaker solved that problem completely)

everytime something was out of sync, the card didn't drop a whole frame. instead of doing so, it replaced the missing field (not frame!) with a field of the past (sometimes even a field that has been captured 20 frames before)

this maked the video almost unwatchable because of the erxtreme flickering and stuttering.

Do you mean this behaviour?

I would be really nice, if you could provide a non-deinterlaced sample.
*.VOB would be okay, too, if the video itself is unprocessed.