No announcement yet.

Colour Grading order workflow

  • Filter
  • Time
  • Show
Clear All
new posts

    Colour Grading order workflow

    I was wondering,

    So you have a sequence of a scene you want to grade, but first you need to get the shots all look consistant as the reverses were done at a different time of day and on another lens.

    What is your bullet-pointed order of things to do once you've put on your onlining gloves?

    Do you match blacks first, then whites? Whether in AE, Premiere, Resolve, using Colorista or FCP's colour tools... Would love to see the steps you tend to take to get from offline to graded master sequence.




      Ok I'm no expert but this is what I do, lots of it i find with time has to take place during shooting.

      Assuming you're shooting long gop h264 (dslrs and the indy thang)

      - Always expose the brightest image you can. Stop downdrastically to see where your shadows are and keep track of them. I usually do this to see the chiaroscuro and keep lifting my shadows (lighting them) till the shadows are almost gone then i pull back a bit. Let's say if I'm exposing at f4, I'll preview my shadows at f8 or 16. That way I know I can slam them down in post and get my clean blacks. I always want to maximize my shadow range on the color wheel. The only way I manage is for my darkest shadows to register not so dark. That move alone more than doubles the possibilities I have in grading.

      - Also: Do not reframe in post. It's a pain, but you wanna drop that file in there and be sure it's composed the way you want. When I start resizing, I may notice a quality difference, the nosie is larger when I'm scaled differently between file to file, it just makes them look way more different. Like taking a nicely downscaled file and cutting it with something using ETC mode. That's the feeling. It ain't fun to fix. Also be stable. Stabilizing is not too friendly, takes care of the movement but the microjitters make everything so obvious.

      That's for shooting. With those two taken care of...

      - I conform all my files to the same format, preferably cineform, coz it's ten bit. This way if I have to send the files somewhere else they don't kick my ass.

      - I work in 16 bit linear workspace, and simulate REC 709 viewing for my final grade. Any other outputs I can open new comps and grade to another simulated output.

      - White balance. I use colorista, and i balance all the footage. If you have no white elements in your frame, have the same card, or balance of the white on the slate. Please be slating. It's so useful past syncing.

      - Set black point and white point. You want to read for this one. It can look blown out, but mouseover and read the RGB. At 8 bit R255 G255 B255 is clipped white. R0G0B0 is clipped black. You don't want that. You want a few digits shy of each end. You want a bell-shaped histogram when you shoot. This is making sure of that using your RGB levels.

      - Adjust primary gamma (midtone) to match all shots, before you get into molding your midtones into a look.

      - My footage should be balanced, exposed as intended, white and black points set, no odd movement, with all the noise the same size (no scaling)

      - Some ppl like to denoise. I haven't ever had success. What I do is simply a very subtle gaussian blur, then an unsharp mask, then some grain on top. This has taken care of some instances of aliasing but is definitely not a cure at all. The cure for aliasing is art direction.

      - Stick that nice comp into your preferred output, simulate your target view (REC 709, Fuji eterna, DCDM) and grade till it looks like you want it to look. Export, and set output module to the view simulation you graded to.

      That's my deal. That's how I do it. I would not be surprised if nore than half of this is not the best way, but I've been in this for about 5 years, it is working for me.

      I cannot stress enough that the secret to good post is shooting for it. If you are shooting for options in post, it won't be easy to sit somewhere. If you have a look in mind, and shoot for it, shoot for every move you make in post, you will do well.

      Let me know if this work.

      P.S. Also, make sure your exposure is the same for each shot before you grade. You do not want to be equalizing with post exposure. It will make your noise levels inconsistent.

      Bring lots of light to a shoot and post should be cake.


        Thanks for replying. Just to say, I wasn't necessarily looking for advice but rather interested in how other people work. That's a lovely breakdown you did there of your processes.

        I use Cineform 10-bit too as it happens and agree about denoising, certainly until closer to the end of the process. And obviously I couldn't agree more about trying to get as close to your intended final look in-camera.

        But in terms of post workflow for example I have alway set black, white and gamma points first before playing with colour. Interested in why you might do this the other way around. To my mind perception of colour can be influenced by it's brightness so I sort those levels first.


          Hey man. It's all good. The reason I do all that first is to be able to place an adjustment layer over several output comps and grade to match. This could mean if I ahve exposure misses that I need to mask, secondary masks, for exposure not color, I do that before. This way I know the grade will take to the output comp and every shot in it evenly.

          As for brightness, this is why my meter and histogram have to be the same across every shot coming into the grade. If exposure is same, your look will take to each shot evenly.

          I guess that's the reason for my order. to create several output comps and leave the two tone look to the end. I believe the tedious work should come first as a set up for that wham bam final look. Those have to come fast. First gut type thing.

          Then I look at it for two weeks over and over ad nauseam and slightly tweak. If I like it the first night, I'm sure in two days I'll look and go 'what?' and make things subtler. It's all prep for that really meditative final look.



            haha I now make a point of saving a still of the first 'final' grade because I'm fairly sure 9 times out of 10 now I then spend 2 weeks tinkering with it only to come full circle and end up with what I started with!

            So with talk of Adjustment layers I assume you're doing your grading in AE? I used to grade in Avid and then AE and I'm about to do my first project with the Da Vinci Resolve workflow which should be fun.


              Originally posted by greymog View Post
              - I conform all my files to the same format
              - I work in 16 bit linear workspace, and simulate REC 709 viewing for my final grade.
              - White balance.
              - Stick that nice comp into your preferred output, simulate your target view (REC 709, Fuji eterna, DCDM) and grade till it looks like you want it to look. Export, and set output module to the view simulation you graded to.
              1) I usually do that, for editing: I color correct my videos in After Effects, render to intermediate codec and then edit from that, color corrected.
              2) Could I ask why not 32bits, as there's more values for whites using it?
              3) Done when recording. And I usually use the presets in-camera, as I mostly use halogen and CFLs lights combined. I like the look it gives, creating a specific mood from colder to warmer colors.
              4) I'll definitively look into that for my next shoot!


                I tend to work 32-bit but there's a pretty high performance hit for relatively little gain over 16-bit.


                  Honestly, sometime when i'm in float, things happen. There's also 8-bit and 16-bit plugins that don't take to 32 bit, i still save those for the end coz there's a conversion, but still I get these hot pixels al over the place at random, total random, can't find a single pattern so far. I stick to 16 and it's bug free. That is the only reason I don't set to 32. I'm dying to figure that out and be done with it. But scared to ask cause I'll probably b asked to google it. I've been getting that a lot.


                    And there's a perverted love i have for 16 bit is that you can't stack endlessly like you do in float. Makes me conservative. I'll get the most out of one instance of a plugin instead of stacking 9 on top. I don't like too much ever. A little bit less always makes me work nicer. It is not a logic I'd recommend it's just how I do. I need a couple of limits to keep me on my toes.