Aspect V5 codec improvements?

[William Gardner]

Hi Everybody,

One of the advertised benefits of Aspect V5 over Aspect V4 is encoder codec quality improvements.

I was wondering if anyone at Cineform (David?) or if any new users of Aspect V5 can provide some specific details on what these improvements are as this may make or break my decision to upgrade. In particular:

1. Will the peak quality be better, or is it just that the compression is more efficient (i.e., same very good quality with smaller file sizes)?

2. Are there specific conditions where the quality improvements are most noticeable? I have used Aspect V3/4 for over a year and in general am very happy with the quality. The only slight issue I occasionally see is banding/color rainbowing/quantization in relatively dark regions with a mild color gradient across a sizeable region of the frame. Is this something that would be improved? Are there other specific issues/ problems that are addressed with the new algorithms?

3. Are there any frame grabs or, better yet, small video clips that could be posted that demonstrate the improvements of the V5 codec over the V4 codec?

4. And, in general, have folks noticed a difference in quality and/or file sizes? Anything else jump out between the V5 and V4 version?

Thanks in advance for any specific data that Cineform folks and/or end users can provide. Cheers,
Bill

[David Newman]

We haven't had time to produce samples to demo the enhancements, but basically Aspect HD used to have its own internal structure that was getting old, so we updated to use the Prospect HD codec core. The primary advantage is further reduced artifacting for a similar data rates. Also banding/contouring of 8-bit processing is significantly reduced.

[William Gardner]

Thanks for the info, David. I know you guys are busy, but if and when you have the time to demonstrate the enhancements (particularly the banding improvements) I'd love to see them with my own eyes to get a feel for how dramatic the improvements are.

Has anybody else done any comparisons and does anybody have any opinions?

Thanks,
Bill

[David Newman]

I know we will have no time with CS3 and Vista upgrades in the works. Why not use the 15-day trail of ASpect HD v5 to test it yourself. You can return to v4 whenever you need -- just uninstall, not deactivate v4.

[William Gardner]

I decided to do what David recommended: namely, try encoding files with AspectV5 and compare them with some AspectV4 encoded files. I focussed on a particularly troublesome file that had noticeable color banding on a dark green wall in the background when using AspectV4.

My conclusion: AspectV5 does substantially reduce the color banding. If I had to make a completely qualitative assessment, I would say that AspectV5 reduces the banding by about 75% (meaning that it looks closer to the original m2t file than to the V4 version with banding, though there is still some ever-so-slight banding if you know where to look). I'm not sure that I would have noticed the color banding degradation in the first generation V5 file had I not known where to look for the problem.

Also, as an interesting data point, here are the file sizes for various encoded options:

Original M2T File: 38 Mbytes

AspectV4 Medium: 42 Mbytes
AspectV4 Large: 56 Mbytes

AspectV5 Low: 42 Mbytes
AspectV5 Medium: 52 Mbytes
AspectV5 High: 72 Mbytes
AspectV5 FilmScan: 101 Mbytes

So it also looks like what was "Medium" in V4 corresponds to "Low" in V5, filesize-wise, and "Large" in V4 is a little bigger than "Medium" in V5.

Cheers,
Bill

PS: Has anyone noticed the following anomaly?

If I play multiple (say, 3) different CineformAVI files with Windows Media Player (select 3 files, right click, and select "Play" or "Add to Now Playing List"), when they play back the brightness alternates during playback.

The first file plays a tad dark, the 2nd file plays a tad bright, the 3rd file plays a tad dark, then the first file plays a tad bright (even though it was dark the first time it played!), etc. I've tried this on 2 machines and seen the same effect...

[Kris Bird]

great study, thanks a lot for the info!

which quality settings produced the banding results (v4 and v5) you talked about?

[William Gardner]

Originally Posted by Kris Bird

which quality settings produced the banding results (v4 and v5) you talked about?

Both the V4 versions had noticeably more banding than all of the V5 versions. I would say that I saw a little more color banding/quantization in the Low setting in V5 than in the Medium and up settings, and that I didn't notice a whole lot of difference between the V5 Medium and up settings.

I believe that DavidN has previously recommended the higher quality settings particularly if you'll be doing effects work, bringing stuff into and out of After Effects multiple times, etc., but that was a while ago and I don't know how much of that advice is still accurate.

Cheers,
Bill

[Craig Irving]

Has the multi-cam feature been accelerated by any chance?

I use multi-cam editing all the time, and find I need to use native-support for HDV in Adobe to get it to function without glitches. It's too slow though.

[David Newman]

Craig,

No multicam upgrades yet.

William,

The data increase is likely for that scene in particular, overall the average data rate changes are minor from Aspect HD V4 to V5. Low is still Low -- the quantization used is the same, have the ability to reconstruct shallow gradients is significantly improved to the point the only main difference between the M2T and the AVI results (other than 4:2:2 upsampling) is the AVI has a little less noise.