home contents changes options help

No studios routinely work end-to-end in 32-bit per channel color because of cost.

Animation and special effects are calculated in 32-bit floating point internally because it is easier computationally and the standard way of doing things in 3D modeling. After being calculated in 32-bit space the output may be truncated to 16-bit for film, or 8-bit for TV, or output (rarely) as lossless 32-bit.

Motion picture film is typically scanned at 10-bit log. That expands to about 12-bit linear. For CPU convenience and for extra headroom that is typically handled as 16-bit linear in studio pipelines. The ICC CMS standard only defines 8-bit and 16-bit profiles, not 32-bit.

Although the standard architecture today for building new tools is 32-bit, studios often have a variety of tools with 8-bit, 10-bit log, 16-bit, or 32-bit architectures depending on the age of the tool. Even with tools that are 32-bit internally, the output is rarely set to 32-bit because of cost. Of course, cost keeps coming down. Studios tend to use the highest quality they can afford. Would studios shoot 65mm film all the time if it became cheap? You betcha!

What Other Toolmakers Say

http://www.intraware.com.au/html/shake_FandB.htm

"The ability to set the colour bit depth (8-, 16-, or 32- bit float) of each operation independently means artists can achieve the quality they need without burdening an entire project with time-consuming calculations."

http://www.idruna.com/photogenicshdr.html

"Why 32 Bits Per Channel? Regular 8 and 16 bit per channel applications are simply not capable of realistically or accurately processing real world levels of illumination, because they clamp brightness levels to be discrete values between 0 and 255/65535."

"The latest 3D rendering packages work in floating point format internally, and support a technique known as image based lighting."

http://www.d2software.com/faq_body.html

"What colorspaces and bit depths does NUKE support? NUKE supports all industry standard colorspaces. NUKE's native colorspace is 32 bit per channel sRGB. In order to render scripts a fast a possible, NUKE automatically unifies all elements to this bit depth and colorspace."