Have any been conducted recently? I'm trying to figure out which to use for divx encoding purposes, I'm aware that abr is ostensibly better, but I may be sacrificing compatibility on hardware (i.e., non-computer) playback.

My thinking is that the bit reservior inherent in CBR mp3 can partially make up for the lack of flexibility it has vs. ABR, but I'm not sure how far it goes in amelioration.

For compatibility I'd go with CBR, because VBR/ABR in an AVI is just a dirty hack and can cause playback problems.
128kbps (with either --alt-preset cbr 128 or -h --nspsytune --athtype 2 --lowpass 16 --ns-bass -8 --scale 0.93 as suggested here) give a pretty good quality for its bitratre (I use ff123's cmdline in my EAC LoFi Profile).

dev0

Is it possible to play divx on anything else than a PC? If there are standalone DVD players that can play divx it would be really cool, but I've never seen any.

Just go to DivX.com and find out, it's that simple. It can be played on various devices.
Anyway, XviD has better quality (my opinion yeah..).

Divx5 and XviD (and maybe Divx4?) can all create MPEG-4 compatible video streams, so when MPEG-4 video players come to market, these movies should be playable. At least, that's what I've heard. You might have to extract the data from the AVI and put it in a different package, though.

Because of MP3's bit reservoir, even CBR MP3 isn't perfectly CBR, so it might also be kind of a "hack" to put that in an AVI. I believe I read something along those lines at Avery Lee's VirtualDub webpage. But even then, the bitrate is almost constant enough that without any padding (the hack used to mux VBR MP3), the audio would still line up pretty well with the video, wouldn't it?

But my opinion about portable DivX is that it's not worth it. I play the DivX on a PC and play the DVD everywhere else. That way, I can use whatever best DivX/XviD technology is available right now. If/when standalone MPEG-4 players become common, encoding technology will probably be so much better that re-encoding my movies (with compatibility in mind) will be worth it anyway.

"If/when standalone MPEG-4 players become common, encoding technology will probably be so much better that re-encoding my movies (with compatibility in mind) will be worth it anyway."


You are so wrong. First, yes, XviD and DivX 4 and 5 are all compatible.

But, encoding technology will be so much better? It doesn't get any better than XviD right now. The technology is always advancing, always being worked on. There's no less work because standalone MPEG-4 players aren't here yet. What is good is good because it's good, not because hardware support drives people to improve a technology. DivX and XviD are two of the most popular video codecs in the world along with MPEG-2, MS MPEG-4 v1,2,3 and MPEG-1. And there's a reason, those two are very good. MPEG-1 is still popular because people either want VCD's or either want files to be compatible with "everyone's" system or because they simply know nothing about video compression... And MPEG-2 is still popular because of course DVD's use it.. XviD and DivX are the best quality ones.

(ok that's the last time this post is being edited )

Do you see the contradiction you just made? Perhaps I'm taking you out of context, but the way I've quoted you is the way I understood your post.

My point is that video encoding is always improving, and that in 1-2 years it will be better than it is now (higher quality, smaller filesize). I'm sure you agree with me on this point.

Perhaps I didn't explain my logic very well. Since video quality will be better in 1-2 years (or when hardware MPEG-4 players are more common), it will be worthwhile for me to re-encode my movies. That's when I'll worry about hardware compatibility and making MPEG-4-compliant streams, not now. Right now I just make movies with Vorbis audio and shove them in a .ogm wrapper, because that plays well on a PC, and that's the only playback device worth considering.

(the rest of my post is a bit off-topic)
Video and audio encoding aren't on the same level. We have transparent audio encoding, but not video (my benchmark for transparent video would be movietheater-quality, which even DVD's can't obtain). So while the only way to improve audio encoding is to make the file smaller, video encoding still has a long way to go. I can encode my music collection in MPC and never gain an improvement in quality from re-encoding in the future, but my video collection will always improve if I re-encode with a newer codec.

That's a different story, you didn't explain well. OBVIOUSLY in 2 years the technology will be better than now.. That wasn't what you meant or at least seemed like you didn't with the previous post.

And what do you mean you can't reach "transparent" level with video encoding??
You most certainly can, with a high bitrate. Since video (i believe) is more complex than audio it requiers a higher bitrate than audio to achieve "transparent" quality.
Try encoding a DVD (part of it) with the latest XviD with using vbr quanitizer (Koepi's build), tune some settings, and i assure you that you'll have a hard time finding a difference from the original. It is not as easy as encoding with LAME for example, with lame you have prefered presets, with video it's more complicated.

Code advances, high video quality is hard to achieve with the low avarage bitrate you believe it should be achieved in (whatever it is). So does audio quality. Encoders improve, there is a bright future.

Video is just more complex, it's not fair to compare the two directly like this, and not reasonable either.

Sorry, but I'd be inclined to disagree with this. I've never seen a Dvix or Xvid encode that was transparent, and I have spent a good amount of time experimenting with this in the past. FWIW, DVD isn't even "transparent" a lot of times. It's not really that difficult to spot compression artifacts if you know what to look for and are watching on a relatively good display. In fact, I'm actually quite disappointed in the video quality of a lot of DVD's being released these days..

Granted, at similar bitrates to what most DVDs are encoded with, Dvix or Xvid would probably look better, but where's the practicality in that? That's only going to become functional when we start getting High Definition movies, and even then I'll be quite skeptical as to whether it will be "transparent" there either.



He wasn't comparing audio and video, he was comparing the level of transparency of the perceptual encoders for their respective fields. This is pretty much an across-the-level comparison if you ask me.

At any rate, I do agree that video is very complex to encode. I also believe (and I'm not certain of this, but it is my theory) that we are more sensitive to visual artifacts than aural artifacts. That coupled with the fact that uncompressed video (especially high res, high frame rate) requires absolutely enormous amounts of bandwidth, and that it will likely continue to increase by orders of magnitude (unlike audio... which does increase in bandwidth demands, i.e. 96khz/24bit, but certainly not to the same degree as video), makes it very difficult to reach a true and relatively universal level of transparency in video compression. It might happen eventually, but I don't suspect it will be for quite awhile.

OK i have nothing to say really .. Except i know that DVD isn't transparent either and that artifacts can be noticed.
And it's "DivX".. how long ago did you check them? ( kidding)
(if you tested XviD like 6 months ago, there have been giant leaps since then)

Hasn't been all that long, maybe 3 months or so. I usually go back and retest all the latest video compression technology every 3-4 months just to see how things are progressing

just some rumble

from my latest tests with some of my animations (which can be considered as perfect input since there is no analog or any other noise, also they were rendered in non-interlaced way) the transparency was possible with divx502 (quality mode) and also with few of the mpeg2 encoders.

the 2nd tests was hi-quality mjpeg video file -> divx502 , but since there was deinterlacing involved the results were not transparent, but comparing 'uncompresed deinterlaced/scaled-down avi' with a 'high-quality divx502' was pretty surprising, knowing that input material was captured from analog source (analog betacam) and also very noisy.

about dvd, the compression ratio is even smaller than 1:25, iam not aware of any audio encoder which is absolutely transparent at that rate.

apples? peas? anyone?

just by the way, SK1, xvid might be iso-mpeg4 (even though the iso-mpeg4 decoder of some player choked on my xvids last time i checked, couple months ago), but its interlaced source encoding functionality blows arse, to say it in a relatively friendly way. it's really SO not an option to mpeg2 right now.

oh well, seeing how fast xvid development is advancing i guess i don't have to worry too much about this. xvid is not the be all end all right now, but i sure hope it'll eventually be.

last time i checked xvid (XviD-23092002-1 _ALPHA_ Release, Based on CVS from 23.09.2002 10:50h MET) there was this slight greenish 'bug' (encoder/decoder ?, dont know), so i did a test by encoding the colorbars, cyan comes out as rgb=0,253,251 ... (iam sure there is a highly complex mpeg reason for that )

xvid isn't exactly stable right now.. alot of features in flux and almost ready for prime time, but not there yet. Alotta bugs, and much tuning to be done to find optimal settings. So I believe it will get "better than xvid right now"... AFAIK xvid is still considered an alpha codec by the developers.

As for transparency, its plain to see there is a long way to go for all mpeg-4 codecs. Blockiness in dark, foggy, or watery areas is very distracting, even viewed on my svideo tv-out. Using 2cds helps alot, but if you want to keep the ac3 track then the video isn't much better than a 1cd/mp3 rip.

One funny thing I noticed is that sometimes a video clip looks good on my monitor but not on TV out, and sometimes its vice-versa.

do some tests where the input is NOT mpeg2, you will be surprised (at least i was), and of course dont expect it in the 500-600 kbit/s area. B)

Does anyone have any links to any short (very short in fact, considering download times) of uncompressed/losslessly compressed or at least vv high bitrate M-JPEG or somesuch video (non copyrighted of course) of either DVD-res (720*576 PAL or 720*480 NTSC) or even better of HDTV-res (1920*1080). Preferably progressive. Could anyone host any such footage? A few seconds would do (yes, I do realise that uncompressed HDTV is about 150 Megabytes per second of video!).

I would like to see what DivX/XviD can do for transparency on material like this.

And yes, I do realise my computer wouldn't be able to play the HD uncompressed file due to Hard Disk read speed being too low, or the compressed HD file due to CPU usage being too high...oh well!