logo Sign In

Color matching and prediction: color correction tool v1.3 released! — Page 28

Author
Time

Not to my knowledge. I’ve never been able to export a LUT from the Color Correct tool and open it directly in After Effects. I’ve always needed to convert them in Resolve first.

Author
Time

STENDEC said:

Not to my knowledge. I’ve never been able to export a LUT from the Color Correct tool and open it directly in After Effects. I’ve always needed to convert them in Resolve first.

Could one of you share a LUT that you created with the tool and one you transformed in Resolve, such that I can figure out where the difference is?

Author
Time

Thanks! I think I see the difference. I will add the option to choose between a Resolve LUT and After Effects LUT. The update should be available by early next week.

Author
Time

There’s been a bit of a delay in the update, but I will finish it as soon as possible. 😃

Author
Time

There appears to be a problem with the After Effects LUT. Williarob tested it, and while it loads, it gives the screwy results. So, I’ll have to check what’s wrong.

Author
Time
 (Edited)

Here’s a debugged version of the ColorCorrect tool v2.2. The exported LUT should now work in both Davinci Resolve and Adobe After Effects. I’ve tested the LUT in Davinci Resolve, and it works fine. After Effects uses a specific format for LUTs, which luckily also still works in Resolve, so it’s a single LUT for both programs. Here’s the link:

https://drive.google.com/file/d/0B8_LYKyZDiajS3ZLb090U3pqSGM/view?usp=sharing

Edit: for some reason the old version was uploaded before. This should be the correct version.

Author
Time
 (Edited)

Williarob has tested a LUT in AE, and it works. Thanks Williarob!

Author
Time

Thank you so much for this great tool. I’ve been looking for such a functionality for a long time.

Author
Time

You’re welcome! Version 3.0 should be out in a while, and will contain the automated faded print correction algorithm. Piota’s scan of the Spanish LPP of Star Wars should provide a good basis for testing the merits of this methodology for correcting an entire print, but the results have been very promising so far…

Author
Time

Is anyone working on AviSynth compatibility? I’m not even sure how CLUT’s are handled in AviSynth anyway, just curious.

Project Threepio (Star Wars OOT subtitles)

Author
Time

If you encode your avisynth script with ffmpeg, you can apply the lut by adding the parameter -vf lut3d=“filename.cube” (including the quotes). I don’t know if it’s possible to use ffmpeg filters with avisynth directly.

Author
Time

D’oh! Never tried using ffmpeg before, I just assumed it was a totally different encoder.

Project Threepio (Star Wars OOT subtitles)

Author
Time
 (Edited)

To get serious again, I actually don’t know how well x264 encoding works when using ffmpeg. But if you’ve got enogh hdd space, you can make an intermediate lossless encoding e.g. with the ffv1 codec (ffmpeg parameter -c:v ffv1) and feed the result to x264 later. It might be a good idea anyway, because the encoding is relatively fast and you can check the intermediate video if the corrected colours are ok.

Author
Time

Just an idea, since you seem to enjoy finding all these awesome ways to do color correction. Would you be able to implement a function in your color correct tool that scans a video file for the frame with the most visible colors (of course including qualifiers that require a certain amount of those colors have to be within a select range of each primary color so that you don’t just get a frame with 100,000+ shades of red.) If I’m correct, given that single frame, you’d be provided with the best option for a single frame color model full film correction.

Preferred Saga:
1/2: Hal9000
3: L8wrtr
4/5: Adywan
6-9: Hal9000

Author
Time
 (Edited)

Sorry for the late response. I’m working on something similar for faded film prints. You can find it in my other thread. Is this what you mean, or do you have something else in mind?

Author
Time
 (Edited)

Sorry, Just realized you responded -

I’m not sure if your other thread takes it into account or not. I was meaning a video scanning technique - here’s a simple example to try and explain:

Frame 1: 7 shades of blue, 6 shades of green, 3 shades of red
Frame 2: 3 shades of blue, 3 shades of green, 1 shade of red
Frame 3: 4 shades of blue, 5 shades of green, 4 shades of red
Frame 4: 4 shades of blue, 4 shades of green, 4 shades of red
Frame 5: 5 shades of blue, 3 shades of green, 5 shades of red

My idea is to scan the file for the frame with the most colors (with a small bias of evenly dispersed colors).
In this case the scan would get a result of “Frame 3”, since it has the highest minimum color, as well as an extra shade of green. Even though “Frame 1” technically has the highest color variance, it is not chosen because of the lower minimum for shades of red - this avoids radically over-defining some colors at the expense of others. However, “Frame 4” is also passed by because after the minimum has been established, the additional shade of green in “Frame 3” is still helpful in defining the color shift.

Since the Color Correction model is based on a single frame matched between two sources, wouldn’t this help identify the frame most useful for the process?

Preferred Saga:
1/2: Hal9000
3: L8wrtr
4/5: Adywan
6-9: Hal9000