Hi Patrick,
Daniel did a great roundup of supported codecs here:
viewtopic.php?f=2&t=3730PNG is lovely because the image compression is lossless 8-bit per channel, but the performance penalty makes it an unwieldy beast. Especially since it is effectively a VBR encode for your videos.
the coolux QT plugin recently added YCoCg. This combined with the Google Snappy compression is basically exactly what HapQ is, but I am given to understand that the container is formatted in a way that makes it play nicer with Pandora's Decoder stack.
I have done some tests with HapQ and found it works really well for camera footage. Others have reported issues with gradients in CG content, though I think this is actually a problem with the colorspaces that most content producers work in these days (in other words: if you produce digital media in a floating point color space, you NEED TO ADD NOISE/DITHERING or your gradients will get rounded to the nearest neighbor, which results in banding).
I think coolux DDS is actually pretty good, but it lacks the extra Luma data that you can pack into YCoCg, so it would probably show more banding in gradients from darker to lighter colors.
The biggest issue I have with the coolux DDS/YCoCg codecs at this point is that the compression is slooooow. I'm in a theater as I am writing this, and we actually just do draft renders for testing content in Pandora using M2V CBR because it takes about 1/4th the time. For all the high fidelity of our color/image encoding, MPEG-2 still manages to fit a workflow performance sweet spot sometimes.