Autodesk Wiretap

[#1]

SGI Flame 9.5.13
take a 12U black frame into Color Corrector.
Make duration 4095
frame 1 set Master RGB Offset to 1
frame 4095 set Master RGB Offset to 4095
goto Animation View and set interpolation to Linear.
now using the color picker, any frame you goto, the image will have the corresponding RGB value.
Render that and then analyze the pixel values of the result clip with the color picker in the Player...they dont match.
Take that clip into Color Corrector, Reset, then use its color picker, same thing, dont match.

Also, the Burnin module messes with color integrity....take the clip from above, run it thru BurnIn then do a difference with 0 tolerance. Not Good.

I have included a File Archive of the above mentioned 12U RGB Ramp.

[#2]

Hi Alan,

Are you sure this is posted in the right forum?

Dan

[#3]

Is there another forum more suited to this? I felt it was more back-end and wasnt appropriate for flame-news. Is there an infrastructure forum?

[#4]

No, this is definitely a flame-news question. I'll bet this has already been answered once before too.

[#5]

No, this is definitely a flame-news question. I'll bet this has already been answered once before too.

[#6]

Can't repro Alan. same steps, my results are good.

I'm going to guess your project is set to 8 or 10-bit intermediate graphics processing. Izzat the case? Because this exhibits the problems you describe.

[#7]

With my project set to 12 Bit graphics, the difference is small, only 2 units off, but it is a difference, which was throwing off my "debugging" tests. With 8 Bit graphics it is widly in-accurate, as excpected.

[#8]

Something tricky here is that, if memory serves me well, the color picked in the Color Corrector module is "special", in that it picks values directly from the image in memory, and so should display the exact value of every pixel, whereas the color pickers in other parts of the application pick values out of the frame buffer, and can thus be thrown off by not having the graphics framebuffer set to 12 bits (and on a Linux box with a floating point frame buffer, I'm not even sure how accurate picking from the frame buffer can be).

JF