Lossless Compression

[Michael Duff]

how much footage are we talking about? I use QT animation codec…files are still pretty big….but then again, I’m backing up 30second TVCs.

[Angus Mackay]

Hi,

If your using QT then the Annimation Codec is probably the default choice. However the best choice is probably Digital Anarchy’s Microcosm which has 16bit colour depth and incredibly small file sizes. To find out more about Microcosm check out

http://www.digitalanarchy.com

HTH

Angus

[Chris Zwar]

Hi,

The animation codec at 100% is lossless, and supports an alpha channel. But the compression is fairly limited and so the file sizes are really big, depending on your image content.

The JPEG codecs- Photo JPEG & Motion JPEG, are practically lossless at 100% and will be about 1/3 the size of the same files compressed with animation. There will be some very minor rounding errors but no actual compression artefacts. Certainly nothing visible. However they do not support an alpha channel.

If you are totally utterly and completely anal about preserving image quality then the animation codec is your best bet. Otherwise, if you don’t need an alpha channel, Photo JPEG or Motion JPEG will save you about 3 times the disk space. In all cases, you need to have the quality set to 100%.

Personally, I use Photo JPEG at 100% whenever possible, otherwise it’s animation for anything with an alpha.

These codecs are Quicktime based and cross platform.

-Chris Zwar

[Bill Clotz]

If you use avi, then I would suggest the lagarith codec: https://lags.leetcode.net/codec.html

It gets excellent compression and even supports an alpha channel. It supports 8-bit RGB32 colorspace (the standard colorspace used by After Effects), although it does not support the 16-bit RGB colorspace.

[Ryan Wieber]

Another option is to encode in QuickTime with the PNG codec, which was made speficially to be a widely-compatible, fully lossless compression format.

[Stefan Rochfort]

Excellent, looks like photojpg is ideal for my needs. Many thanks for your help. Just one question on photojpg – I’m assuming that if I’ve got multiple generations of compression (i.e. I use one compressed mov in a composition which is again compressed using photojpg for the output and so on) I presume I’ll be compounding compression/rounding errors. Any idea how many generations at a 100% before you get visible degradation (to the very anal)?

Thanks again.

[Steve Roberts]

You can test this by making multiple renders, recompressing, then placing the original render and the last render into a comp and comparing. You could mask half of one render or you could set the transfer mode of the upper layer to “difference”. In the latter setup, identical pixels would be black, non-identical pixels would show as a colour. The more black, the less degradation.

Steve