Color Correction Leveling Process—Question

Hello everyone,

I am a film student with dreams of being an editor. I just finished color correcting my crew’s student film, and my director and co-colorist are about to take over and start the creative grade. They are telling me however, that they are going to have to redo all of my work because I followed our coloring professor’s workflow, and our professor was “wrong.”

My professor taught us to lower the lift, raise the gain, possibly adjust gamma, etc, get the image white balanced, then match all the images within a scene to look alike. Then move on to secondaries.

My co-colorist and director say that by attempting to lower the lift and raise the gain consistently across the film, I have created more work for them because they want to go for a low-contrast look in the creative grade (they are only telling me that they wish I had done things differently after the correction is finished, I might add).

My professor insists that coloring is non-destructive, so if I did my job correctly and the clips are well matched, it should only be a matter of changing the contrast, getting the look they want, and applying it across a scene.

I am caught in the middle of all this and don’t know who is right. Can you all help?

Thank you so much,
Molly

[Peter Chamberlain]

Moved to Resolve forum

Did your co-colorist and director not tell you the look that they were going for? Did you not ask?
I'm not a professional colorist but personally I disagree with your professor's approach of this is what you do with everything before anything else. Also, did your professor not explain the use of scopes?
Hopefully some professional colorists like Marc Wielage and Walter Volpatto will chip in.

Well, it's obviously something that should have been discussed before.
But your professor is right (so we are already two): before going for a creative look one should make the clips neutral and match those which are from one location.
But if they want a low-con look, nothing is wrong with leaving Lift and Gain where they are out of the camera, but still match them based on a 'hero' shot.
That's the beauty of a node based system: you can separate the steps.
And, finally, the process is non-destructive, since DR works in a 32 bit float space internally. You could, for example, completely overexpose a shot in one node and pull it down without any quality loss in the next one.
No damage done!
Prof. Uli Plank

And, finally, the process is non-destructive, since DR works in a 32 bit float space internally. You could, for example, completely overexpose a shot in one node and pull it down without any quality loss in the next one.

Ehhhhh... that's not quite true. If you crush a clip with the lift control in one node (say, -30), and then try to bring it back up and normalize it in the next mode, some information could be destroyed in the Order of Node Operation. So there are limits.

But I agree with much of what you said. It's not impossible for a "temp grade" to be thrown out and restarted, and that's just the facts of life (and post). Next time, the creative team has to set looks with the original colorist so they don't stray from the intended look.

Me personally, I think a low contrast look is wimpy and bland, and it's not going to have the impact of a higher-contrast "normal" look. But... I concede it's a creative choice, and there are good reasons to use it in some cases.

Of course there are limits, even in 32 bit float, I just tried to illustrate.

Actually, it's a typical case of not communicating. If they are aiming for a specific look, colorist and those setting the light for a scene should already talk before shooting. Some things are achieved more easily on set and others can be left for post. That's what we teach here.
Well, this wouldn't apply to all parts of a documentary, but I hope they are not going for a bland look in a doc.

...some information could be destroyed in the Order of Node Operation. So there are limits.

Sure there can be certain limits with 32bit math but if we are purely discussing balancing shots with lift/gamma/gain I don't think you would ever run into those extremes.

...our professor was “wrong.”

My professor taught us to lower the lift, raise the gain, possibly adjust gamma, etc, get the image white balanced, then match all the images within a scene to look alike. Then move on to secondaries.

Semi non-sense. Lift/Gamma/Gain is one of the tools at your disposal to modify an image whether it's for balancing or a creative look but always lowering lift and raising gain is rubbish and depends on what has been shot and the intended creative look.

It also depends on the colorpipeline. If you don't colormanage the project or use a conversion lut/node to go from camera to display space at the end of your chain, yes probably because it's unprocessed otherwise but grading from a log image without such a conversion is very bad practice because your decisions will be based on your output and cannot be changed after the fact for different display targets without severe extra work.

My co-colorist and director say that by attempting to lower the lift and raise the gain consistently across the film, I have created more work for them because they want to go for a low-contrast look in the creative grade (they are only telling me that they wish I had done things differently after the correction is finished, I might add).Molly

More non-sense. If you've balanced all your shots in relation to eachother there is nothing stopping you from globally changing the overall look and contrast and I find it hard to believe that requires a lot more work.

Next time, the creative team has to set looks with the original colorist so they don't stray from the intended look.

I echo this. It's important to have everyone involved aligned in such a process to prevent surprises. A few quick test looks to send of before you do the heavy work helps a lot.

More non-sense. If you've balanced all your shots in relation to eachother there is nothing stopping you from globally changing the overall look and contrast and I find it hard to believe that requires a lot more work.

I gotta say, I've inherited disastrous projects from other people where they literally "painted themselves into a corner" so badly, there was no easy way to undo the damage. In one memorable case, a well-intentioned DP tried to do his own color by using a Normalizing LUT as the first node, followed by a Look LUT as the second node, and then a series of a dozen additional LUTs that stretched and squashed the signal into something they wanted to see for dramatic purposes.

The session was brought to me by the distributor, and I sat down, kind of gasped in horror, and said (probably a little too loudly), "well, this is the Node Tree from Hell," and I blew it all out and redid it all from scratch -- no LUTs, just manufacturer's color science, a contrast curve, some Lift/Gamma/Gain, some sat, and little tweaking. I always save the original look as a version, so I'd switch back and forth and the filmmakers were impressed and not a little embarrassed by the difference. I repaired all the damage in a couple of days and sent them on their way. The distributor later told me "you took this from 'unreleasable' to 'pretty good.'" And that's not bad for a low-budget indie that wasn't that great to begin with.

CST Nodes are enormously helpful if you're working from Camera Raw files. Do that at the very beginning, make sure all the decoding and color science is correct, and you'll basically see what they saw on the set. If they had a Viewing LUT (or a Show LUT), drop that in and use that as a guide. There are pros and cons with using Show LUTs, but if it gets the filmmaker the look they want, go for it. Aim for simplicity and speed, and be ever-mindful of the Order of Node Operations and how earlier nodes can cause terrible damage -- particularly LUTs that crush or clip the signal.

I gotta say, I've inherited disastrous projects from other people where they literally "painted themselves into a corner" so badly, there was no easy way to undo the damage. In one memorable case, a well-intentioned DP tried to do his own color by using a Normalizing LUT as the first node, followed by a Look LUT as the second node, and then a series of a dozen additional LUTs that stretched and squashed the signal into something they wanted to see for dramatic purposes.

That is indeed a recipe for disaster. But yes as you described having a clear solid colorpipeline from the start and being mindful of order of operations will prevent a lot of potential issues and keeps the entire process open and flexible.

To the original poster:
If you are still fairly new to this all I can highly recommend this channel.
https://www.youtube.com/c/CullenKelly/videos
He has a lot of great content explaining ways color management and science and does live streams. Great for beginners.

To the original poster: If you are still fairly new to this all I can highly recommend this channel. https://www.youtube.com/c/CullenKelly/videos He has a lot of great content explaining ways color management and science and does live streams. Great for beginners.

Cullen is terrific and knows the subject well, and he's one of the best out there covering advanced topics. I would also recommend Darren Mostyn, who does a good job. The paid tutorials from FXPHD, Ripple Training, Mixing Light, Lowepost, and TACResolveTraining are very helpful: I learn something from all of them, and rely on them every time Blackmagic makes changes and updates the software. Learning Resolve is a process, not an ultimate goal: you just get better at it over time.

Thank you all so much for your wisdom and clarification! I am a newby, so I will definitely check out those YouTube channels. I’ll also take your advice and make sure that my team members and I communicate better in the future.

-Molly