Cineform Qualities Explained

[Paul Curtis]

I've been directed here by Jake at cineform, hopefully to clear up some issues.

I've actually been a long time user of cineform but stopped a few years ago but recently upgraded everything up to date because i'm doing post on a heavy VFX feature and want the benefits of lightly compressed intermediate files.

My post workflows are based around AE CS6 and Nuke X on windows 7 x64.

Nuke X doesn't work with AVI but can use (read/write) QT

There seems to be quite a bit of confusion between AVI and QT implementations.

My tests are all in CS6 to keep things simple. 32 bit float, HD with an alpha.

I can write from AE in AVI no problem, the AVI options have different descriptive settings. I can export a 4444 film scan keying version no problem with the alpha in. As i understand it the cineform alpha implementation keeps blacks and white uncompressed to preserve keying quality. At the highest setting my test file generates a 4.2GB file

The problem is QT which just seems broken. Firstly the options are very limited but i've tried a combination of settings both in the QT dialog and in the AE dialogue as follows:

1) Cineform set to 444 and Force 16bit. AE Settings RGB, Trillions of Colours WORKS (2.441 GB)
2) Cineform set to 4444 and Force 16bit. AE Settings RGB+A, Trillions+ FAILS frame 31
3) Cineform set to 4444, AE Settings RGB+A, Millions+ FAILS frame 4
4) Cineform set to 444, AE RGB Millions WORKS (3.130 GB)
5) Cineform set to 4444, AE Settings RGB+, Millions+ FAILED frame 56
6) Cineform set to 4444, AE Settings RGB, Millions FAILED frame 56
7) Cineform set to 444 Force 16bit. AE settings RGB Millions FAILED on 3 [Expected to fail should be trillions]
8) Cineform set to 444 Force 16bit. AE settings RGB Trillions WORKS (2.441 GB as above)
9) Cineform set to 444. AE settings RGB Trillions WORKS (3.336 GB) [Expected to fail but it worked]

In other words i cannot get an alpha out via QT like i can with AVI. Also Force 16/trillions seems to produce a SMALLER file than millions (4 and 9) which is not 16bit and just millions.

I have not been able to match the filesizes between AVI or QT, AVI is producing larger files

So, you can see how confusing it all is. There's no consistency between AVI and QT and even some odd results within QT. However it appears i cannot get an Alpha channel via QT no matter what i do.

Please can someone explain to me what these settings are actually doing behind the scenes and what i can do to help troubleshoot the alpha fail on QT.

Also what is the colour depth cineform is working in with all the scenarios?
What does force 16 bit actually do - it appears to make smaller files!

Any help would be really appreciated as i'm about to set down a complex path so i need to understand fully what cineform is doing if i'm going to put it at the core of my pipeline...

cheers
paul

[Paul Curtis]

Not having too much luck here :)

After some more investigation.

Cineform is 12 bit internally, the Force 16 bit refers to the QuickTime UI so that it displays trillions, it's not a codec setting AFAIK.

So my three issues are:

Seems impossible to generate an Alpha channel via After Effects and QuickTime, can anyone else try this and perhaps confirm that it's a bug? My setup is pretty freshly installed, so i think it's the software combination.

Secondly i seem unable to achieve the compression levels that AVI offers using QuickTime.

Finally, setting QT to Millions produces larger files than Trillions, this seems at odds with what is supposed to be going on...

anyone?

thanks!
paul

[Giroud Francois]

i think you make a mix between color resolution (8, 10 or 12 bits) and color encoding (16, 24 or 32 bits)
regular video is 24 bits (8+8+8 for RGB) so there is no alpha.
As soon there is an alpha layer, video is coded with 32 bit.
I am not aware of of video format using more than 32bit video, so if you need alpha, you are stuck with 8bit RGB.

[Paul Curtis]

Giroud thanks for replying.

I'm not sure i fully understand what you're saying. I'm running AE is 32 bit float mode, not 8 bit, AFAIK that means internally each RGB pixel is a 32 bit float number.

When i'm done processing i want to output at 12 bit 444 (or 4444). In this case 444 is just an hangover to video terms but i'm expecting a colour value at each pixel. Also the 12 bit version stores as RGB, not YUV.

That becomes my master intermedia file, i have 4 hours of footage like this.

I did manage to get AE to output a 4444 QT through some help from Jake at cineform.

However it seems that cineform <> nuke is very very difficult, in fact seemingly not possible at all. Which is such a shame. Neither QT will import and AVI is truncated to 8bpp via what i assume is the avi framework.

Unless i find otherwise i have no choice but to settle on OpenEXR as my interchange, this has been made easier by the discovering that there's a lossless wavelet compression option inside there.

cheers
paul

[David Newman]

Nuke does need to switch using the CineForm SDK, as QuickTime is so 8-bit centric, it is not a good solution to interface to compression technologies.

[Paul Curtis]

Originally Posted by David Newman

Nuke does need to switch using the CineForm SDK, as QuickTime is so 8-bit centric, it is not a good solution to interface to compression technologies.

That's certainly true also, i'm sending a request into the foundry as well. Although i did hear a rumour of a 3rd party plug in for Nuke/Cineform but it seems to have disappeared.

QT in general seems really bad at these kinds of things, in terms of speed and features.

It all seems a bit broken which i guess i why facilities are all file based. There really are no workable options out there that are truly cross platform and work with all apps. Neither QT nor AVI seem to offer the structural bit depth for this work.

Is there an opportunity to add cineform as a compression option under OpenEXR?

thanks
paul