4 Things To Know About Working With LUTs

Ah, I feel better now. In truth, I have used LUTs going back to 2002 — the only colorist I know who’s done it longer is Jill Bogdonovich — and technical LUTs have their use, particularly in going from digital back out to film. But I get weary of people saying, “just use this magic LUT and Camera X will look just like Kodak blah-blah negative stock.” Just ridiculous. I agree 100% with Robbie’s conclusions above.

It would be so cool if resolve could take a lut applied to a node and translate it to adjustments in the various panes.

I know that this insight is listed as ‘beginner’ level of difficulty, but it is still frustrating to be presented with yet another discussion on LUTs that doesn’t really explain much. It’s like an attempt to explain them to people who are allergic to formulas and math. After spending the best part of a year trying to understand the core functionality and make-up of LUTs (with particular regards to Resolve) I decided to have a go and build them from the ground up, thereby gaining a truer understanding of the actual math.

I posted my findings here:

http://www.magiclantern.fm/forum/index.php?topic=15801.msg158237#msg158237

If you’re going to discuss LUTs you should always prioritise mentioning the specific input and output colour space (gamma) and primaries (gamut). Anything that doesn’t do this (including Denver Riddles’s ‘Look’ LUTs) should not be considered professional. With enough experience a colourist can learn to tell what a LUT will do to footage just by looking at it in a text editor, and that would represent actual progress in demystifying what is an essential (whether you like it or not) component in the modern grading process.

Even if Resolve could, it would be using a variety of math algorithms that all have very different looking final results, but technically reverse the look into curves or other controls. That said, manually doing this is a fantastic exercise that I try and do with any LUT presented to me.

Apps like 3D LUT Creator can do this to an extent, but you’ll discover its very much a artist based decision on which method to use and to what extent. Its not something I feel would translate well to an “auto” function ….yet.

While everything you say is valid, i feel you should remember that everyone has a different background before coming to color. While technically oriented colorists (especially with deep VFX backgrounds) can understand LUTs via opening the text file, its another language to those less familiar with it. Apps like Lattice and Lightspace CMS allow those less technically inclined to still do professional LUT manipulation using more visual methods without digging into the “numbers”.

One of the best points Robbie made was you CAN grade without any LogC – Rec709 transform, and while the primaries may not be converted from Arri Wide Gamut to Rec709, we can easily make it work and it can still look good, sometimes faster than using any LUT.

“If you’re going to discuss LUTs you should always prioritise mentioning the specific input and output colour space (gamma) and primaries (gamut). Anything that doesn’t do this . . . should not be considered professional.”

I couldn’t disagree more… including proclaiming what is professional and what isn’t.

If a LUT designed for DCI P3 gives me the look I want in Rec. 709… then it’s perfectly acceptable to use. Creative LUTs / Look LUTs are just what the name says. They don’t need to be technically accurate—not any more than a preset in Magic Bullet Looks needs to be technically accurate (whatever that would mean).

It’s only when dealing with Technical LUTs that they’re either (mostly) right or (mostly) wrong. And I say ‘mostly’ because in the Rec.709 > DCI P3 translation major post houses often have their own proprietary ‘blend’ when doing this technical translation. Are they un-professional if they vary from each other? Or is there much broader latitude in the science than meets the eye? I say, it’s the latter.

That said, I agree—there is nothing ‘beginner’ about any discussion of LUTs. I’ve modified this post to ‘Intermediate’. I also recommend, as you walk down the road of understanding LUTs, that you check out the other Insights Robbie links to. Team Mixing Light has a fairly consistent approach to working with LUTs even though all three of us have different client bases.

Also – I’ve bookmarked your post for closer reading. It looks interesting.

Although I still would have issue with your specific example of using a LUT for a purpose other than its stipulated intention, with regards to the quoted statement I would say that although I press for what I feel is proper labelling I’m not saying you can’t use LUTs for unoriginally intended purposes, it’s just that those original purposes should always be know.

There should be no scope for ‘mostly’ when you actually know the math involved. If there are two ‘different’ Rec.709 > DCI P3 LUTs, then that should be traceable back to the gamma transform involved, by way of an expression applied to a known set of linearly spaced input values. The primaries (gamut) transform is always based on a fixed 3×3 matrix. The word ‘blend’ shouldn’t come into it, unless you don’t understand the basic math and prefer to comprehend things in the form of ‘feels’. I had a quite sobering discussion with a very renowned VFX compositor recently where the general state of colourist capability was lamented, and regarded with a mixture of disdain and frustration. I’m afraid I had little reason to disagree with him.

I’ve been a member of Mixing Light for around 6 months now, and have viewed pretty much every video (including the ones concerning LUTs). Thank you for the recommendations, and I hope you find something of informative value in the link I posted.

RE: “The primaries (gamut) transform is always based on a fixed 3×3 matrix. The word ‘blend’ shouldn’t come into it, unless you don’t understand the basic math and prefer to comprehend things in the form of ‘feels’.”

The one thing we shouldn’t lose in this discussion that while it’s absolutely correct to say ‘the math is either right or it is wrong’, we shouldn’t leave out of this discussion… ‘does the image still look right?’

The math may be precise but often the image loses some of its original feeling. It’s like the difference grading on a 15″ display and then viewing it on a 75″ display… the math is precisely the same but the feel is completely different and you’ll feel the urge to do a trim pass. That’s what happens in many Rec 709 > DCI P3 trims… a precise math transform is only the beginning.

But the end is all about the human eye and how it interprets the image. If my eye tells me I need to adjust the image after applying a technical LUT that is 100% accurate… I’m adjusting the image.

Hi Paul –

Sorry for chiming in late – I was on a flight to Dubai for a project, and the plane didn’t have working Wifi (the horror!)

You bring up a lot of great points let me address some of them:

1. Yes, my intention with this article/video was not to cater to the more advanced user how wants to dive to the particulars of the matrices, rounding errors or other math particulars, but my goal was to do two things – recap some previous thoughts on LUTs by Dan/Pat and to address general questions we’ve been getting recently about lookup tables in a conceptual/workflow sense. What I published was never meant to be a mathematical explanation of everything LUTs (obviously).

So I do understand your comment about lack of specificity on the math in this and other articles/videos for this site and elsewhere, I really do. But I also think there are two levels of learning involved here: concept/general implementation & workflow and then as you’ve pointed out the mathematical details and processes/thinking that go in to building a capable LUT

I can totally see a series on the math side of LUTs that might appeal a bit more to what you bring up here and maybe to other readers.

Interested in contributing? (totally serious)

2. I do agree that the lack of input/output information in naming related to gamut/gamma or even ETOF PQ info as is the case with HDR LUTs these days is frustrating. And I mentioned that in the article I see this as a big problem – but at the same time that technical information goes well over the head of the purchasing public that many LUT developers are trying to target – i.e. editors, enthusiasts etc. ‘Awesome Thriller Look’ is more marketable and more approachable for that audience.

And that audience is clearly the target of many a LUT producer (because as I mentioned they’re looking for easy ‘fixes’ and ‘looks’) – not a colorist willing to breakdown the math of a particular LUT – in those cases, as you pointed out, that colorist probably makes their own LUTs for their particular needs anyway as you did! Or has a team of color scientists on staff to create their in house LUTs as groups like CO3 and other large outfits do.

But I don’t think its fair to call a LUT or a LUT producer unprofessional for naming a LUT in the ways that we can agree they probably should or saying (I don’t think you meant it this way) that using them is unprofessional.

3. I have to agree with Pat that what matters most to me is what ends up on screen – that’s what clients are paying for. That is of course not to discount a level of knowledge that should be required by a ‘colorist’ to be conversant in the LUTs or other parts of modern color grading.

With that said,I know many very very successful colorists working on huge projects that make beautiful images that couldn’t tell you anything about the particulars of a LUT, a knob or slider in their grading tool other then they like what it does to the footage they’re working with.

And while I 100% see the logic of understanding why a LUT wasn’t working in a particular situation and breaking down the math to figure out the ‘why’, I don’t think its the only thing that matters. Personally, I’ve never been a grade by numbers type guy – I’ve used creative LUTs over the years that SUCKED and did all sorts of horrible things to images on close inspection – noise, banding etc, but the I and most importantly my clients LOVED the way the looked. I don’t think that makes me less of a colorist, or my clients opinions any less valid. Yes the technical side is vitally important in our field but so is look and feel.

4. I’m eager to go through your post/article on LUTs when I’ve had a bit of sleep – I’ve been in the air for 15 hours and I’m writing this at baggage claim waiting for my luggage!

Using a LUT or Color Management helps keep a level of consistency that I find difficult if not impossible when just twisting the knobs to go from LogC to Rec.709. Yes you can make a good looking image, but shot matching will likely be much more difficult. My 2cents.

Anything I’ve posted in that thread is available for public consumption, so everyone is welcome to incorporate it into their own interpretations. Of course the civil thing would be to mention the source, but whether or not that occurs is of no great concern to me. I would, however, like there to be more of a conversation with regards to the subject matter so anything that might facilitate that would have my support, plus I am interested in what people think of it once they’ve had a chance to give it a read. Hopefully it didn’t prolong the effects of your jet-lag 🙂

Hi Robbie,

100% agree with you. No matter what the numbers say, if if looks good it looks good and visa versa. The end goal is getting the image to look a certain way, not getting the math right. I found myself (as i love the technical aspect and use Nuke for example to create all kinds of technical transformation luts) way too much focused on the technicality, until i let go of that completely.
The only thing you need to know is the pitfalls of a lut like you mentioned in this insight.
The most important revelation for me some time ago was the destructive clipping compared to resolves non destructive clipping of blacks and whites.
Similarly to , i think it was in a previous insight , to take care of your cache format setting, which if not high bit range , can cause similar hidden clipping. (not lut specific but same sort of data loss)
Other then that its open season and I just use them creatively, with pre and post adjustments where needed. I think the only time i am still using technical luts is for xyz conversions etc.

It should probably be reiterated that it is not simply a case of either/or, rather it should be the combination of both sets of skills. At the end of the day it is the final image that counts, absolutely, but in the long run it is surely better to have a healthy level of technical knowledge to go along with an artistic vision, if only for the reason that the ‘competition’ probably will and therefore have a likely advantage in securing work. I believe there is a considerable anxiety amongst many colourists at the prospect of learning something new and seemingly complicated, but this fear can be dispelled though education. It’s nowhere near as terrifying as some people fear, and should be viewed as a means of empowerment by those eager to advance their creative ability. These are just my opinions though, so each to their own.

Having said all that, it is often those who have the least understanding of math and colour science that are the loudest in dismissing their importance to the colour grading process, and so I suggest caution from taking advice from those who are perhaps more interested in preserving their own sense of significance than progressing the community as a whole. As always, a balanced view should come from a near equal comprehension of both sides of the argument.

Will not go into the suggestive remarks (intended or by accident) towards people devoted to actualy sharing their knowledge on this site rather then (as often customary) hiding their tricks for their own convenience, but rather give my personal example of how ignoring one side can greatly benefit you sometimes.
Myself i grew up as a scientist (aerospace engineer) and later focused heavily on computer science (where at that time there was more future for me) which in the end made my career and business. But have always been into some form of artistic endevor. Starting as an audio engineer in the start of electronic music era (in the pioneering time when working with modular synthesis) , and sound designer , later diving more into video and more recently into compositing and vfx, just out of pure interest and assisting others with their projects. Math always had a huge importance and i tried and still do, to understand them to the deepest part (specialy working with stuff like nuke and houdini are extremely demanding).
My wife, a visual artist understands zero to nothing from all that science and we started working together where i took care of her technology (creating video installation , programmed / build her software tools , compositing and grading etc etc.
In lots of our initial discussion on her projects , we started arguing , mainly me explaining her why her project would not work because the math/science was incorrect, the colors where wrong,etc etc.
In the end she won the arguments (women always win arguments) and just ignored my science babble and did her thing and i just worked on her direction doing my tricks without having any faith in the end product. In all these cases she ended up winning awards, reaching huge audiences and laughing at me in my face.
Over the years i started adopting more her view of the art and split the science of as follows.
I use the science to create and understand tools . Then i switch off science and rationale and use the tools like a painter would who is not worried how his brusher where fabricated , if they are composed of exactly so many threads in the right angle if his canvas is exactly mathematicaly a correct rectangle bla bla bla.
VFX artist work in a similar way ofter, where they get tools (for example a complete houdini digital asset rigged and ready to animate) from the “techies” and then they use their artistic feeling to create the magic.

So you are right in saying that there is place for both science and pure artistic feeling, but very wrong in the statement that you need to be equally skilled in both to succeed.
In fact your mathematical or scientific view more then often (as is no mystery) blocks your creativity which is not composed of logic and often counter intuitive. Ask mister Spock ……
I dont think that any director would give a damn if you are a wizzard in your tools and can explain every mathematical translation that would result in the shift of mangenta visible and that you user a technicaly correct lut , if you can not deliver the artistic end result he/she is looking for, but would embrace you if the reverse is true.
Tricks can be learned, any trick , any software anything given enough time. If people say that is not the case i will prove them wrong every time ( i learned fluent Portuguese in 6 weeks while failing for years at school to learn any language. Now i speak 6 languages)
Artistic feelings can not. You have them or not (in varying degrees). Period.

I love eat breath science in all my cells, but any day any time any second i would welcome advise from a true artist in the field that works by his guts and gladly ignore a science blabermouths advise on art. (including myself)

The best tools for artists let you forget the underlying math, do not “force” you to learn it but enable you to create. I had that after struggling for years in dozens of linear track based sound/music recording/editing tools and did not understand why i would never finish tracks or always got stuck in the technical details of setting up connections etc. Thats because scientists build the tools and not artists. Until Ableton Live was released (build by an artist).
Within one week i finished 10 tracks and the software became the de-facto standard for live sound/music creation due to its intuitive and creative nature.

Resolve is well on its way (although definately has its challenges) being such a tool where (with a proper control surface, which in my case i just created myself) you forget about color science and just move knobs , wheels and balls to get the look you want. I point my wife to the knobs (she is also a profession photographer) and she now hates photoshop for obvious reasons. We actualy do some photo work in resolve now in 10% of the time.

To end my blabering, Robby takes in this post the right approach in giving the artist enough info to not distract from the art , but not to get trapped by the science of using luts. There are enough posts (and linked by him) to dive deeper if one wants.
Thats also what i loved from Patricks Flight School. Yes he talk some base techie stuff (basicaly what is a pencil and how best to use which pencil for what), but focus on the creative part , more on the why , when then the how (as here are always dozens of ways to achieve the same) , a welcome exception to all the dozens of other resolve courses on the market that just learn you how to use resolve.
Thumbs up for this approach in a sea of technology focused resources. (unless you give me now the lut to do shotmatching, and aother one to create the perfect look on any footange then i give up)