16 bit uncompressed codec - lossless

[Christopher C. Murphy]

Hey, I just saw this on a video news website. Anyone know if 16-bit is to much? Is it useful?? I'd think that G5 owners would benefit from the 64-bit?

Free 16 bit None codec - 16 bit uncompressed

From Jim Tierney of Digital Anarchy: An offshoot of developing Microcosm was a 16-bit version of the None codec. We figured some of you might have a use for it, so it's available for free on our web site:

Free version:
http://www.digitalanarchy.com/micro/micro_none16.html

Pay version:
http://www.digitalanarchy.com/micro/micro_loss.html

Chris

[Rob Lohman]

The files aren't 64 bit, the pixels are. 16 bit for each RGBA channel
results in 64 bits for each pixel.

I would say this is only useful for people doing lots of compositing
work and working with CGI and things. Our DV camcorders don't
even have 8 bits per channel (due to sampling ratio etc.).

[Christopher C. Murphy]

Oh, ok. But, we should still leave it here for people that could benefit from it. I'm not doing that type of work, but I'm sure some of us are or will.

Chris

[David Newman]

16bit is little overkill for the HD10U (an 8bit source.) Although there are valid reasons for the upconvert, probably 99% of editing operations will not need it. The problem with a 16bit uncompressed RGBA format (64bits per pixel) is the file size. A single frame of 720p is over 7MB, which puts single stream playback requirements at 211MBytes per second.

[Christopher C. Murphy]

On their site it says, "With 64-bit, you can maintain smoother gradients and more accurate color. Unfortunately, deeper color usually means greater file sizes."

Does the codec actually make footage look better? What if we render out to DC30 codec - then convert to the free Microcosm? Will we get better "gradients and more accurate color" by converting at the final stage?

Maybe I'm totally lost with this point, but it seems like the final render would benefit from the 64-bit codec.

Chris

[David Newman]

Unfortunately the smooth gradients are lost in the 8bit acquistion they can't be restored. So an 8bit per channel file convert to 16bits per channel still has the quality on the original 8bit data.

[Christopher C. Murphy]

Hmm, is there anyway to mess with the HD10 and make it 16 or 64 bit? I'm just trying to figure out if there is a way to bypass and get the aquisition at a higher rate. I know, we all want that..

What if we "composited" the same footage over itself before doing anything else. So, in essence you shoot 8-bit, composite itself using the 64-bit codec, but adding deeper blacks and other small tweaks during the conversion.

I bet the new footage would get something out of the process?

[Rob Lohman]

You are not even utilizing the 8 bits fully. In an 8 bit system you
can get 16 million colors. You can't increase your camera and
almost no-one who shoots with such equipment would benefit
from a 16 bit file format.

Just stick with the regular format. That's enough data to handle
and fill with colors.

[Christopher C. Murphy]

Ok, thanks for the info!

Chris

[David Newman]

You can tweek the gamma curves is a 16 bit space better, with greater accuracy, a small advantage even with 8 bit source data. Unfortunately there is no partical way to achieve what you are asking.