#16 bit files
7 messages in this thread
Hi guys,
As I understand, 3DS internal render calculations are done at 16 bits
per channel. Then the result is dithered to 8 bits per channel. Is it possible
or are there IPAS's that would allow you to import 16bit files (backgrounds, in
VideoPost,etc) then render out as 16 bit files. Or maybe as 10 bit Cineon
files.
If that is possible, it would certainly help push 3DS into feature film work.
Any info would be greatly appreaciated and if this topic has been covered
before, sorry, it didn't see it.
Martin Kunert
Kunert/Manes Ent.
Martin:
>> Is it possible or are there IPAS's that would allow you to import 16bit
files (backgrounds, in VideoPost,etc) then render out as 16 bit files. Or
maybe as 10 bit Cineon files. <<
Absolutely. The most direct method is to use an IXP. I'll add that you can very
easily create custom file export filters with The Script Extension,
particularly the uncompressed varieties. All you need is the file format
specifications from whoever invented the format. We're working on a scripted
one in-house right now for a film recording project.
Martin Doudoroff/KUB
Martin Doudoroff hit the nail on the head… it's very straightforward to write
out any type of image file with The Script Extensions (or, if you want to
program in C, the IPAS r4 SDK). The Sony Pictures Imageworks have developed a
Cineon writer for 3DS, and I've heard of a few other outfits that have,
likewise, written proprietary hi-res-color image writers. Perhaps you can get
together with MartinD on the script that he's currently working on to do this.
– G
Since I'm not involved in the feature film work I was intrigued by your request
to access RGB Channels at 16 bits. Is that the channel bit resolution needed by
films?
DaviD "Intrigued…" GouID
<<
Since I'm not involved in the feature film work I was intrigued by your request
to access RGB Channels at 16 bits. Is that the channel bit resolution needed by
films?
>>
Most films are done in 8 bit, 10, or 12. Rarely is 16 bit used due to the
tremendous computational and storage requirements it uses. "True Lies", for
example, was all 8 bit.
<<Most films are done in 8 bit, 10, or 12. Rarely is 16 bit used due to the
tremendous computational and storage requirements it uses.>>
I understand that the 16 million possible colours available with 8 bit RGB
channels is more than the human eye can differentiate, so why use 10,12, or 16
bit channels. Is additional information stored in these "extra" bits?
Thanks.
DaviD "Cones and rods" GouID
<<
I understand that the 16 million possible colours available with 8 bit RGB
channels is more than the human eye can differentiate,
>>
Whether that is true or not, I don't know.
<<
so why use 10,12, or 16 bit channels. Is additional information stored in these
"extra" bits?
>>
Yes, just like 8 bits can discribe more shades of a color channel that 6 bits,
10 bit describes more than 8.
35mm and 16mm has more possible colors than 16 million. 24bit image files can
cause color banding in wide stretches of film color, like blue skies.
Another way to look at it is that 8 bits gives 8 levels of brightness wherein
each level is twice as bright as the previous. This is sort of like 8 f-stops
of latitude on film. Film, as you may know, can hold more than 8 stops of
latitude.Hence an image which has 10 stops of latitude will be squeezed or
currupted when transfered into an 8 bit file. Those feature films that intend
to do CGI work in 8 bits plan their shots in such a way that the images fall
into 8 stops.
8 bit film images tend to cause the most problems in compositing and, not
suprisingly, color correction. This is mainly due to rounding errors in the
calculations used to accomplish these effects. There is a simple mathamatical
example of this but I can't recall it off the top of my head.
There are also 8 bit and 10 bit Log files, but that's a whole different
subject.
Martin Kunert
KMe