F35: ACES full circle

F

Film Scientist

Active member
After my recent deep dive with LUTs & F35/ACES/Resolve, I feel like I am finally free to shoot/light scenes by what my eye sees in each environment.
With 12bit uncompressed files and an SLOG/Sgamut to ACES/Linear transform,
I've actually moved back to traditional filmmaking with a simple light meter, exposing for middle grey.
If middle grey properly is exposed, the only parameter I really need to adjust on camera is tungsten vs. daylight.

Below is the DSC One Shot frame, the original SLog Sgamut file (left), and the color corrected shot (right) to give you a baseline for the camera's great range and color in 12bit,
TLIP0000594_0086484 corrected.jpgTLIP0000594_0086484.jpg
TLIP0000600_0086696 corrected.jpgTLIP0000600_0086696.jpg


The stills are graded to the colors I remember seeing that day.
With ACES I am starting to see how sensitive the camera is to the light hitting an object in a scene,
while I might be tempted to color correct for coolness so whites on the waveform are 'right',
the color of the paint and white chips were slightly warmer based on the light hitting it.


Original DPXs (12bit unpacked to 16bit) for Davinci
https://www.dropbox.com/s/jo1w3jloc3nf6uh/DPX.zip?dl=0

8 Bit 1080P TIFs
https://www.dropbox.com/sh/0yc34efr7lhklgi/AAAgb5U_7UajZTLOXy_DzsBUa?dl=0
 
Last edited:
paulcurtis said:
Great work, Very interesting but your special sauce is the LUT to convert into ACES space. Normally these are mathematical transforms supplied by manufacturers of cameras. So how did you generate that LUT and how accurate is it do you think?


cheers
Paul
Click to expand...

There's an earlier thread where he explains how he generated the LUT. It's based on Sony's own IDT, which is available here on the forum.
 
I made this new transform LUT using Lattice, the math within the program is 64bit, and the resulting output LUT is in 16bit. It's very accurate.

FYI, I compared this new LUT to one that can be implemented through the original After Effects OCIO process, and they yield almost identical results, (maybe a .5 percent difference between the two), so I know it's definitely in the right ballpark.
 
Last edited:
Hello everyone,

Looking for an ACES workflow for the F35 I stumbled across this Google Groups forum page where Thomas Mansencal developed an ACES workflow for the powerful compositing and effects program Nuke by The Foundry. You can find the post here:

https://groups.google.com/forum/#!topic/academyaces/zhmR8fPUZIQ

I wrote to Thomas and asked if he had completed and perfected his workflow, as I too use Nuke for certain effects. He very kindly sent me links to his files and to a program that he used for the various colourspace conversion matrices.

Here are all the links he sent me:


Here is a working version of the Nuke file of the ACES Mailing List thread (just copy paste in Nuke):
https://gist.github.com/KelSolaar/620e1cadf1505e248601

You can change the various colourspace conversion matrices using this online tool: http://colour-science.org/cgi-bin/rgb_colourspace_models_transformation_matrices.cgi

The input files (for the Nuke tree) are now respectively here:
https://www.dropbox.com/sh/lj9x6cyd5t56tha/AACv5bjilPq05nRkM9G7wNyKa/ACES?dl=0
https://www.dropbox.com/sh/lj9x6cyd5t56tha/AADEJDQeXVwJU9w8PrJ4CW3Ua/camera?dl=0


I have very limited knowledge in ACES and am trying to learn as much about it as I can and I thought others here might find this information interesting and useful.

Thomas was also so kind in offering any help, should we have questions on his work above.

He also an interesting website with amazing 3D work. Should anyone be interested I suggest taking a look at it at:
http://thomasmansencal.com

Through his website I also found this other website which I think some of you will find very interesting as well:
http://colour-science.org/


Cheers,
Igor
 
You must log in or register to reply here.
Back
Top