Page 1 of 1

EXR compression

Posted: Thu Jul 24, 2014 7:49 am
by Wenneker
Is there a way to save out compressed exr files? Uncompressed exr comes out at 14mb per frame (720p) or 31 mb/frame (1080p) and are really slowing down my composites.

Re: EXR compression

Posted: Thu Jul 24, 2014 7:55 am
by glimpse
just curious, isn't PNG is goint to be the same as compressed EXR?..

Re: EXR compression

Posted: Thu Jul 24, 2014 8:02 am
by TBFX
glimpse wrote:just curious, isn't PNG is goint to be the same as compressed EXR?..
Does png have a 32bit floating point option?

Anyway if saving compressed EXR is possible can we have the single scanline zip method as one of the options as this is best for Nuke.

T.

Re: EXR compression

Posted: Thu Jul 24, 2014 8:27 am
by prodviz
Or possibly a half float .exr option, to maintain a lot of the highlight detail etc., but bring the file size down.

I guess you could take the full .exr set in to Nuke and then output a half float .exr set to speed up your comps.

Re: EXR compression

Posted: Thu Jul 24, 2014 9:39 am
by Wenneker
Yes I'm converting my renders to compressed EXR before comp now but it would be nice to cut out this step.

Re: EXR compression

Posted: Thu Jul 24, 2014 10:09 am
by Wenneker
just curious, isn't PNG is goint to be the same as compressed EXR?..
No it isn't the same. Ping is an integer format. EXR is floating point.
PNG will have 8 or 16 bits of precision giving it just over 65.5k color values at 16bit. EXR is 16bit (half) or 32 bit(full) but being a floating point format not all values are equally spread over the bits. For example a 16 bit EXR uses only 10 bits for precision but uses 5 for exponent and one as a sign bit. This gives it the the abillity to represent a much larger range albeit at a bit lower precision.

Re: EXR compression

Posted: Thu Jul 24, 2014 11:02 am
by glimpse
Wenneker wrote:
No it isn't the same. Ping is an integer format. EXR is floating point.
PNG will have 8 or 16 bits of precision giving it just over 65.5k color values at 16bit. EXR is 16bit (half) or 32 bit(full) but being a floating point format not all values are equally spread over the bits. For example a 16 bit EXR uses only 10 bits for precision but uses 5 for exponent and one as a sign bit. This gives it the the abillity to represent a much larger range albeit at a bit lower precision.
thanks for info, Wenneker!

Re: EXR compression

Posted: Thu Jul 24, 2014 8:46 pm
by TBFX
prodviz wrote:Or possibly a half float .exr option, to maintain a lot of the highlight detail etc., but bring the file size down.

I guess you could take the full .exr set in to Nuke and then output a half float .exr set to speed up your comps.
+1 for half float, it usually all that's needed for renders. The benefits of getting EXR's out of your renderer exactly like you want them without having to convert them are huge, particularly in a studio environment where you have many renders from many shots constantly coming out.

T.