[Sidefx-houdini-list] Film at 16 bit
antoine at floqfx.com
Fri Mar 7 17:02:17 EST 2008
Part of the trick if you're working in 16-bit integer space is to
*not* have your whitepoint at 65535. If you do that, what happens is
that anything that you rendered that has a value greater than 1.0
gets clipped. The downside is that all your downstream software has
to know how to deal with the whitepoint setting.
As Ivan wrote, it's probably easiest to work in 32-bit float. That
way, 1.0 is your whitepoint. At some point, e.g. when converting to
Cineon or Dpx files, the floating point 1.0 will translate to the
film's 90% white, which in Cineon land is at code value 685.
Likewise, the blackpoint, while in floating point is at 0.0, is at 95
in Cineon land.
On Mar 7, 2008, at 1:55 PM, Ivan DeWolf wrote:
> the internal format you work in is often different than the final
> delivered for scannig/layoff. Usually, footage is scanned in to a
> 10-bit log
> file (often a cineon format file) and delivered to be put to film
> in that
> format; 10 bit log makes sense for recording to/from film, but it
> doesn't make
> sense to work with when compositing. To convert a 10-bit log file
> to a linear
> file without losing data, you need 16 bits of data; hence the
> "standard" of 16
> bit linear output. converts neatly to 10-bit logarithmic color space.
> Often, it is useful to work internally in floating point space,
> when dealing with odd lighting effects; floating point doesn't clip
> values or posterize dim regions.
> I would suggest if you have the diskspace and can tolerate the
> network traffic,
> work in 32-bit floating point colorspace. Render 32-bit float,
> composite in
> 32-bit float, and at the very end convert it all into your 10-bit
> log file.
> That way, if you need to push your colors around, you have ample
> Quoting Larry Giunta <larry at gcreativestudios.com>:
>> I was hoping to hit you film guys and gals up for just a bit more
>> advice about rendering for 2k film.
>> Just wondering, of the available quantization option on the mantra
>> output drive,
>> if there is any one that is considered a "standard"?? ..... or if it
>> varies widely depending on the shot, production. etc.
>> I re-read Jeff's old school blog on the subject which makes it pretty
>> clear that it needs
>> to be at least 16 bit. I'm just wondering what might be the best
>> choice to consider.
>> -16 bit integer
>> -16 bit float
>> - 32 bit float
>> Background on this particular shot we're producing is
>> 1. It will be rendered in mantra, composited and output from COPs
>> 2. There is no live action, not many textures.
>> 3. It's very heavy on volumetric light ( bursts and light beams, that
>> sort of thing) lot's of blurring etc at the COP level.
>> While I'm at it, I guess I'll ask about preferred image format. I was
>> assuming 16 bit tiff, but not sure if that's best
>> Anyway, sorry if these are somewhat basic questions but would very
>> much appreciate any thought from those who have
>> more experience with film.
>> Thanks as always,
>> Larry Giunta
>> Creative Director
>> G Creative Studios
>> larry at gcreativestudios.com
>> Sidefx-houdini-list mailing list
>> Sidefx-houdini-list at sidefx.com
> "Perfection is achieved when nothing else can be removed"
> -Yvon Chouinard
> Sidefx-houdini-list mailing list
> Sidefx-houdini-list at sidefx.com
Floq FX Inc.
10659 Cranks Rd.
Culver City, CA 90230
More information about the Sidefx-houdini-list