Test request : Ideal XviD settings for anamorphic 16:9 encodings on 2 CDs ...

[ChristianHJW]

Hi,

with the upcoming support for anamorphic encodings without having to use a resizing filter for 2 CD rips in 'certain containers' and 'certain players' ( ;) ), i'd ask you guys to make a few tests for us :

- Please try to find out how low you can go with actual XviD dev builds with respect to bitrate for a full anamorphic encoding, until things start to get blocky. Enable all the good stuff like b-frames, Q-PEL and GMC, even if you dont use that for 2 CD rips normally, the codec may need the extra savings.

- Please make the above test for both MPEG and h.263 quants. While first is normally preferred solution for 2 CD rips, this may change for anamorphic encodings as the codec has to deal with about +30% more pixels for the same bitrate, but of course with the advantage of a much higher vertical resolution.

Sorry for bothering you guys, but we are still crazy busy to release the new DShow parser filter, coming with seeking and a few other Goodies ;) ...

[symonjfox]

Ok mister, I'll try ASAP.

I don't thik that will be a problem IMO, because we can just reduce HORIZZONTAL resolution. For example, we could do a 1 CD rip using a SVCD res (480*576 for pal) instead of a full D1 res (720*576).

In special cases also a Half D1 res should be good enough (352*576).

I use/create SVCDs and CVDs from long time and I think that their quality is very good, so I think that using an MPEG 4 codec (that should compress better and better) the quality will be amazing.

[ChristianHJW]

Originally posted by symonjfox For example, we could do a 1 CD rip using a SVCD res (480*576 for pal) instead of a full D1 res (720*576). .. yes, this could be a nice option for making a 1 CD rip that's going to be watched on TV mainly .... havent thought of that yet :) !

BTW : cropping the black bars off is fine, just dont use resizing, or only vertical as you were suggesting ....

[crusty]

GMC isn't considered part of 'the good stuff'. But I'll do some tests to see if end credits will be better with either VHQ=4 or GMC, because they don't work together.
I encode end credits at the lowest possible bitrate without blockiness, which with heavy filtering and decreasing the saturation usually lies beneath 500 kbps.

[Emp3r0r]

@Christian: for a 2 CD anamorphic encode try the following for good results

- Load Defaults
- Enable ChromaMotion
- Enable VHQ 1
- Encode @ quant 2

If the result is way oversized, try
- 2 B-frames

If the result is still way oversized, try
- Lower resolution

If the result is slightly oversized (50-500 megs)
- Run a second pass with linear scaling

[Teegedeck]

Hm, I usually have MPEG-quantizers for a full res encoding, and if it doesn't look good for one CD, I use two. 3 B-frames and VHQ=4, quarterpel to make up for not using the nice and sharp Lanczos resizer. 3/4 - 1 1/2 hours on one CD look better to my eye that way than encoding a resized picture. Blocks (if there are any) get nicely hidden by anamorphic resizing while decoding.

[edit: If necessary I preprocess with MPEG2DEC3 and Fluxsmooth.]

BTW, you guys are great - go, matroska, go! :D

[ChristianHJW]

I am sorry in advance to dare to bring this topic up again ( yes, i know, my blatant matroska advertising sucks :( ), but has anybody tried it yet ? Just try to forget its mkv related for a moment and concentrate on the original subject, after all the MPEG4 norm allows it also.

Just finished my first anamorphic 2 CD encoding with pretty nice results ( i think ) at 1250 kbps at a 720 * 432 resolution, vhq=4, Q-Pel, and max. 5 b-frames ( dont expect that encoder really makes use of that ). Both of my machines ( PIII 1000 MHz and SMP PIII 800, normal 32 GB video cards ) can play this nicely in real time using ffdshow and without postprocessing ( you wont need any ;) ), and its looking really neat IMHO.

If anybody is interested to check i would happily upload a small sniplet to one of my public webspaces ( max 10 MB ), just gimme a shout .....

[Teegedeck]

Hi Christian,

well, I'm all for it! :D I've really been wondering today why so few people encode at anamorphic resolutions. From my experience, there is a gain in perceived visual quality in most cases (I know that's a daring statement).

I'd go for full resolution encodings for 2 CDs. Playback isn't a problem on a 1.7 Ghz Athlon (it was on my old Celeron 466...). Also if you go for one CD I find a 512x368 encode, resized in BSPlayer to 512x272, far more pleasing to the eye than a 608x320 encode (both videos having about the same bits/pixel*frame ratio) although I have to filter a bit more . Or the other way round, n encode at quant=3 is likely to look worse than the same clip at quant=4 that has about the same filesize and same horizontal resolution but has been encoded as anamorphic.

Come on guys, give it a try! ;)

[Defiler]

OK. I'll give this a shot tonight, with some kind of high-quality DVD as input.

[symonjfox]

I made many tests and yes, quality is very good.

I used to create CVDs and SVCDs that have 2/3 and Half D1 res (so anamorphic).

I also often capture from DVB that has such resolutions, so my goal is NOT to use resize (or resizing just horizzontally).

To tell the truth I don't understand what exactly you want to know.
I think that quality depends a lot on settings used in the encoder, so bad settings in anamorphic or not, will result in a bad file.

PS: remember the INTERLACE setting ... very important in PAL countries :)

[Defiler]

What he wants is a set of test results that gives guidelines for how many bits are needed to make various types of footage look good without any resizing, so he can place that information in some kind of readme file.
Right?

[Koepi]

Wrong.

As you should already know there aren't "ideal" or "best" settings codec-wise for a general output (i.e. 640x272 or something like anamorphic encoding).

This is just one more advertisement-thread for using matroska (relate that somehow with xvid so it suites into the xvid forum).

Remember that MPEG-4 allows for bitstream embedded aspect ratios like MPEG-2 and thus is container-independent. We just didn't came across to finally use them as athos (developer of ffdshow) didn't find a solution to the problem of resizing the image after directshow graph building time.

Regards
Koepi

[Defiler]

Regardless of whether it is folly/advertisement or not, I think I correctly interpreted what the original poster was asking for. Heh.

[Teegedeck]

Hm, well, perhaps Christian doesn't always have sinister motivations...;)

Just joking; of course he has! :D But still, I for one enjoy another discussion about anamorphic encoding, regardless who brought it up. After all, we don't need a special container for resizing videos in BSPlayer.

[Selur]

but it would be nice if the player would do it on its own thx to some info contained in the audio&video container ;)

[Koepi]

So how does a player resizing video according to data in the container relate to xvid?

[spyder]

Well, the original topic of this thread (XviD settings for anamorphic encoding) has NOT been discussed on this board. Calm down. We are all supposed to be a "community".

[ChristianHJW]

Originally posted by Teegedeck Hm, well, perhaps Christian doesn't always have sinister motivations...;)
According to some people here i obviously do, making their life unbearable. Dont know what makes them react like this, they will have their reasons i guess and i decided to not care too much.
After all, the purpose of this board is to make progress. This was and still is the case, at least i hope it is. Anamorphic encoding is such a progress IMHO, and i started this thread primarily to find out if other people think the same way as i do and can confirm what i found. And yes, i am quite happy that matroska is currently the only sensible solution to use it, at least on Windows and with DirectShow based players. Its a proof that its paying out to invest time into a concept that lasts instead of quickly hacking some working stuff together. But dont think i am naive to believe it will stay like that, MP4 is standing outside the frontdoor and will support it nicely also, no doubt here.

Just joking; of course he has! :D But still, I for one enjoy another discussion about anamorphic encoding, regardless who brought it up. After all, we don't need a special container for resizing videos in BSPlayer.
Thats the right spirit to have here on this board mate, again IMHO ;) ....

[Koepi]

3) Keep the focus: Questions outside the scope of a certain forum will either be moved, locked or simply be deleted.

4) Be nice to each other and respect the moderator. Profanity and insults will not be tolerated. If you have a problem with another member turn to the respective moderator and if the moderator can't help you send a private message to Doom9.

5) Do not spam. No blatant advertising.

That about the rules concerned.

As for the number 4: nice to know that I'm not helping progress and that the permanent rule 5 breaking of yours doesn't concern you too much.

I'm closing this thread now and hope that you start rethinking your position. It#s nice to have a new file format, but advertising it like this (we could call it SPAMing) isn't apropiate and IMO doesn't help anything but is poisening the general mood.

I know that you like exactly that to happen (me and some other developers giving up on OGM) so you have the monopoly on your side. But "the market" (I hope you understand if I get it out of the technical shape and put it to marketing which you always do) can only be interesting and crowded as long as there are alternatives.

I feel sorry for this development, but you didn't react on a single post which asked you to do a little less advertisement.

I'm also confident that you'll put my words upside down and whirl them like hell to make me look bad again, but I'm just asking for less blatant advertising on every possible occasion.

Koepi