I said here what color tools I need for a log space.

By offset I mean adding a number to the R, G, and B channel sample values - individually for white balance, or the same number to all three for exposure compensation. White balance needs fine granularity, like 0.001. 1% is way too coarse. And yes, there's an issue with contrast. You might think contrast is a simple adjustment you can do any old which way, but I've found it's quite special just how well it works when you do it this specific way in the log space: straight-line slop pivoting at the middle grey value, which if I remember correctly is equivalent to a gamma adjustment in linear. And when you start doing curves you see how curves in log are a lot nicer to use than in a power space or in linear. Exposure compensation and white balance are nice, but it's really this contrast adjustment that brings it all together to make a complete color correction system. I tried everything to make it work in Premiere, and it just wasn't going to work, not with Premiere's built-in effects, not with any of the third-party plugins I tried, not with SpeedGrade, not with After Effects or Color Finesse. I tried log spaces, power spaces, and linear spaces. None of them gave me everything I wanted in Premiere. Premiere support is something I can revisit later, but it's going to require a custom plugin to make it work. Logarist works in DaVinci, Vegas Pro, and FCPX because the math of those color effects is doing exactly what I need. It's all about the math. Premiere just didn't have the right math.

@balazer

I still fail to grasp math and logic behind this approach.

As you are using LUT, later use editor build in transforms and later use another LUT to return to original values (not space!).

@radikalfilm, you should try to get the white balance fairly close to correct in the camera, and then make whatever corrections you want using Logarist in post. I always do the corrections by eye. It's not difficult. It's a what-you-see-is-what-you-get workflow in which the adjustments are based on your own aesthetic judgments. It doesn't need to be based on any measurement of the scene. But if you aren't sure how to get started with the adjustments then I suggest you shoot a white or grey card, bring the footage into your video app, bring up an RGB parade scope or vector scope, and then do the white balance correction in Logarist to bring that white card to white. It's the same idea for exposure compensation: you can shoot an 18% reflectance grey card and then adjust the card to 38% RGB in BT.709 (with Logarist LUTs applied). I think you'll find that after doing that a couple of times you won't need to bother with the card or the scopes again.

@Vitaliy_Kiselev, you can read this white paper: http://cinematiccolor.com/ My approach is fundamentally the same as what's in that paper, and also fundamentally the same as ACES. Compared to ACES I've just made some different design decisions for the color space and display rendering, and I've covered more camera color spaces than just the ones with ACES IDTs. There's a lot of math behind it, too. If you're really interested in the math part of it, you'll need to get a good understanding of the CIE 1931 color model. Here's a decent little paper: http://dougkerr.net/Pumpkin/articles/CIE_XYZ.pdf Another resource: http://onlinelibrary.wiley.com/doi/10.1002/9780470175637.ch2/summary I use the RIMM/ROMM gamut: http://www.photo-lovers.org/pdf/color/romm.pdf Many of the formulas I used are from http://www.brucelindbloom.com/ . For white point adaptation I use linearized Bradford as described in http://www2.cmp.uea.ac.uk/Research/compvis/Papers/FinSuss_COL00.pdf .

@eatstoomuchjam, yes, I tested Lumetri extensively. It's not the same workflow. In Lumetri, the first LUT is a camera-to-display LUT. You make all of the adjustments in the display color space. Lumetri's color adjustments assume a gamma-type display encoding. The last LUT is a look LUT. Also Lumetri doesn't provide the kind of contrast adjustment that I use.

Just recently saw this thread and the Logans'wrist stuff. I was testing/comparing a workflow between ACES and standard within resolve. Like ACES all our LUT collection is rendered pretty useless and 'cause of that I thought that was probably not worth it. But just to be sure and being free, pv user's stuff and all that saliva I gave it a light spin. Must say I'm very very impressed; both with the ease of use, the math behind the CTGs (colour transform's gremlims), the accuracy, speed and fine adjustments that allows. In resolve this was already pretty clear, but in FCPX the "difference" is astounding.

So congrats @balazer and thank you for sharing your work =)

I'll come back with questions

GH2 but it's been all -2 smooth since the beggining; actually that's one of the questions I had:

• is there anything I can change to optimize the GH2 standard 0 0 initial LUT to before mentioned settings?

• So if I understand correctly it is ok to use other LUTs after the Logarist BT.709? If so those are very good news =)

• Another thing that's not quite clear to me - and I've read your Logarist's page (which would improove in readeability if had some sort of center delimitation/format for the body of text) and watched the video several times - other than exp compensation, WB (offset) and contrast with pivot at .5, which tools are "optimized" to work with Logarist? Say inside of davinci and given the series of nodes you talk about, what can I use and what should I not?

• In one of my carnage WFs I separate chroma and luminance (which I set to use only info from the green channel, R > G , B >G), is it okay to implement this before the 709back-to-earth Logarist LUT node?

• Same scenario as before but I change the colour space of one of the nodes?

• For cafe con leche or milky blaks it seems curves or low/low soft doesn't work so well, where should I tweak the deep african no-tomorrow blacks?

 
That's enough for 1 tray, love the HL roloff high five natural looking and the math behind it +)

@balazer much appretiated. I'll mess, test, compare more and surely implement your suggestions while searching my own ways too... and change the GH to standard 0 0. Cheers

PS
Yes FCPX's colour board controls are much more finicky or abrupt if you will and no curves, arrows' use needed

No, it won't work. Lumetri can load the LUTs, but Lumetri doesn't provide the color correction tools you need for a log space.

Hi, @wslagter. I never had settings for S-Gamut3.Cine in my ACEScc config. If you were shooting with the camera set to S-Gamut3.Cine color, the colors would not have been quite right in ACEScc. Logarist does support S-Gamut3.Cine with Sony Cine1 gamma, and also Sony Cine1 gamma with the standard sRGB/BT.709 gamut. The Cine1 profile in my ACEScc config is the same as in Logarist.

On the Sony A6300, I recommend shooting in Cine1 gamma with Still color. You can also shoot in Cine1 gamma with S-Gamut3.Cine color if you really want the wider gamut. Just choose the matching Logarist input transform. (Cine1 with or without S-Gamut3.Cine)

The workflow in ACEScc is very similar to, but not identical to Logarist's. Logarist uses different primaries and a different log scale. So you can't just take your existing ACEScc project and switch it to Logarist and have everything look the same. But the color correction process is the same. Use the same filters, just with slightly different correction values, and you'll get a nearly identical result, assuming it's the same output transform. The Logarist BT.709 output transform is the same as the "Standard Rec.709" transform in my ACEScc config. It's not the same as the ACES RRT.

@wslagter, yes, disable ACEScc by setting the View Transform to Off. Once you do that, the color space settings in the media properties and in the render dialog no longer have any effect, so it doesn't matter what you set them to.

Thanks for the LUTs. I tried them in Resolve and found nothing I could not replicate easily and exactly, using the standard Resolve tools. I don't really understand the benefit of using this approach. Why use 3 nodes when one will create the exact same look? Is this for people who don't know how to grade?

I've always wanted color correction to be this simple. And exposure compenstation has to be done in the linear space... Lift gamma gain and the color wheels, though powerful. Have never felt like simple math, like lightroom does, those wheels are more of a dark art

@balazer In Resolve there is no hue shift when applying contrast, luma curves or using the primary controls in rec709 colour space. I have never experienced the issues you mention. Which software displays this problem?

I just finished a more extensive test and I do not like the shadow artefacts that your logarist process adds to the image. I was able to perfectly match the logarist result in Resolve using a single node in rec709 YRGB colour space using only primary sliders for gain, gamma and a little mid tone detail. The main visible difference was that using standard Resolve setting maintained a better and cleaner skin tone. Logarist introduces a contrast change that made the skin tones gradation from the lit side of the face to the shadowed side less smooth and changed the rendition of skin colour in those shadow areas. The skin tone in shadows was less saturated and was a bit less pleasing to the eye.

I think for FCPX it may be OK but for an experienced colourist in Resolve, I think it is unnecessary and creates more issues than it solves.

Don't get me wrong; I think it's a really good idea and I applaud your hard work and thorough approach, but I don't need it for working in Resolve. In Premiere it would be great because of how awful the grading tools are.