Video workflow, when is the best time to color grade

I know this might be a bit off topic, but here are so many talented videographers around that maybe my question is easy to answer.

I did a huge mistake recently by agreeing to film my son’s kindergarden’s winter musical. I did little movie projects before, and back in the 90’s even a bigger project. And if I do not do it, nobody will. So please do not discuss my stupidity :wink:.

I want to use kdenlive for cutting and grading, but in the projects I did before, only little grading was done. There are different options for colour grading:

  1. Directly grading the clips before adding them to the time line
  2. Grading of the clips on the timeline when adding them
  3. Grading after the final cut is done, at least an initial version of the final cut

Maybe there are other approaches. However, whenever I playback graded clips, the preview becomes extremely slow. So what would be the best option to do a hassle-free cut but have reasonably graded output? The final length will be between 30 and 45 minutes.

It this because the grading is happening in real time? If so, I suggest you export the highest quality version of each clip and then reinsert them. Just a guess. Last time I made a video was 10 years ago. :slight_smile: So, to answer your question, a combination of 2 and 3.

//// Disclaimer I am NOT a professional colourist, I just like colour ////

There are many ways of doing anything; for such a “complex” task as editing a video/film the possibilities greatly multiply 'cause there are more steps and each of them can vary, adapted to and for the person doing it. The deadline, the length of the finished project and moreover the amount of material one is dealing with “should be” the cornerstone to figuring out a sensible workflow.

That said, and if you’re following a more classic (“logic”, optimised and tested) approach you’ll always grade towards the very end - when you start hearing the archangels’ trumpets - normally when the edit’s locked, some do it before the sound work.

{ manual entry: here a loose silly yet interestingly truth-full phrase }
Planing is like super OMG, OMG!!! The backup of editing ,-)


PRE >>
More often than not the editor has to deal with “issues” even before starting the process.
This is what I call BAcTH time.

  • INGEST (Asses, organize, reference <- catalogue) = it’s about the content

  • EDIT (Layout, build, trim <- to tell) = it’s about the flow

  • GRADE (Colour correct, VFX, grade <- look) = it’s about coherence with the whole & subtle harmonies

More often than not the editor has to go back and change things, a “healthy” process would have implemented feedback sessions along the way, besides its normal conversation enriching function this helps to avoid nasty surprises at the end.

A personal note on speeding the (grading) process.

The video editing suites have evolved a lot, probably today one can get away colour correcting and grading in some of them. For the non linux (and excluding multiplatform DR and Lightworks), I’ve used almost all of the big boys but Avid.
For what it is worth I can tell you that the single most important thing that improved quality and time spent doin’ color was when I started using the proper tool. For me was jumping from Final Cut Studio to Apple’s Color. Today that tool is without a doubt Davinci Resolve and it is fucking free!!! For all the steep that the learning curve might look (there are plenty good quality, serious tutorials available) the payoff is worth it, specially if you’re expect to do more such a work. Nodes are just way faster, more sophisticated and useful to grade with than layers IMO

There are several other ways to cut corners like creating an overall LUT (look) and apply it once the main colour correction is done; but if you’re using resolve you won’t even need that, just group your clips under the same tree and be done =)

Here a video example of someone who (despite annoying voice and some generalisation; mostly due to time constrains and target audience) knows what’s he is doing

I would grade the clips on the TimeLine to average them out with the other clips. For playback purpose, just disable the filters. You should do all edits before color grading.

If the lighting is constant through out all clips, then you could export via lossless and do color grading after. I would do it in the Timeline though.

My experience is with home-made software, not with current “real” software.

Personally, I regarded grading as needing two steps. The first was to make each shot camera-neutral and lighting-neutral, so I could easily edit segments of shots together without jarring. I’ve seen too many two-camera videos where the same person looks ruddily pink from one camera and sickly green from the other. So I do that grading at the “import shots” stage.

The second step is at the clips stage, where I want a particular look for that shot.

A bad movie, but it emphasises the distinction: CantFeet.

With hindsight, I think this is in accord with an aspect of the ACES workflow: regularize the camera shots so they match each other (though I didn’t use linear encoding), and apply the “look” later.

1 Like

I have done several projects in Kdenlive. Butt…I never color graded them, because I sort of “learned as I went” and I did not get to color grading yet.

But I will say that in general, if you’re having problems with slowdown while editing your project together, read and use “proxy clips.” This helped me tremendously when working with clips that had a lot of effects attached.

With Proxy Clips, Kdenlive will basically duplicate all your source material at a lower resolution so it’s less CPU-intensive to work on. Then when you do your final render, it will use the full-rez version of the source videos.

I haven’t done any video editing for over a year now, and at the time, the developers were just starting to offload some of the computation onto the GPU for many effects, so maybe check if there’s a way to do that, also. I’m not sure of the state of the software at this point and how far that work has progressed.

Hope this helps.

Before importing the footage in Kdenlive I suggest to check the YUV range, in theory there are only two standard, YUV limited (16-235) range and YUV full range (0-255) but for unknown reasons some manufacturer like sony and panasonic encode videos in the 16-255 range :rage:

My camera works in this way, it encodes h264 video in the 16-255 range, it looks overexposed with clipped details in the highlights but that’s not true.

Straight out of camera, histogram on the right

After re-encoding ( I like vapoursynth for this kind of jobs)