Another ACES/Color Management question...or ramblings

I’ve been flirting with the “new” color management in the system for years…and every time i revisit, makes my head hurt…as it did again today. This will probably ramble…and be way longer than it needs to be but one way or the other i need to put this to bed and either figure it out or move on. I watch the AD tutorials, hear how it should make my workflow much easier…but i’m not seeing it. And it could very well be that i’m having a hard time wrapping my head around it. I work on a lot of long form projects…run time could be 2 minutes, could be 2+ hours. Thousands and thousands of clips…a mixture of Slog2, Slog3, Cannon Log, rec 709 ect. For the most part i know what to expect in terms of media…but there are always surprises. I’m a legacy user so i use the legacy color policy. I have an established work flow, I bring all my selects in native and rely on numerous user nodes i’ve created that contain the various LUTs/CC tools designed to achieve certain looks…and/or trouble shoot rec709 sources. I’m looking at this issue (legacy vs ACES) right now purely from a grading/CC work flow. I’m not concerned (yet) with things like action and what does or what does not play nice within it. These longer form projects i’m doing pretty much end with me so I’m handling the grade…which is what always brings me back to the ACES work flow and is it a better mouce trap?. I spend a fair amount of time matching all the various sources in terms of color and getting them somewhat cohesive. Its not difficult…but can be tedious and time consuming. I’ve been told ACES can put me in a better position to accomplish that…but again, not seeing it (or i’m not getting it). It seems to me like it could be just as time consuming writing various “rules” for various types of footage…or having to manually tag footage. How does one write “rules” for footage shot off a single camera that is both Slog and Rec709…is it looking at metadata or is it based off the file extension? How does one use a variety of LUTS (creating different looks) within the context of viewing transforms? Its all very confusing to me and the fear is getting in to deep and realizing i just f’d myself. So I guess what i’m asking in a very long and very round about way…is there anyone that has experience finishing longer form work, with a mixture of log flavors and rec 709 sources that is using ACES that could shed some light? or do i need to just put my big boy pants on and figure this shit out…or if it ain’t broke should i just stay the course. Any insight would be helpful…i’m very tired.

1 Like

I totally see where you’re coming from. I do a lot of Color on Resolve and have the same feeling about using LUTs vs correcting from scratch. I’ve found myself wasting a lot of time trying to find the right LUT (unless I’m given a specific one) as footage isn’t really ever consistent and needs to be graded shot by shot anyway. Granted, I’m talking about commercials, so it doesn’t have such a big overall impact. For long-form, in your case, it might be worth the trouble of finding the right LUT. But there are so freaking many of them, I just want to get on with it

My thoughts exactly too!

The input rules largely exist to help you in tagging your footage. I find them to be more trouble than they’re worth due to the randomness of media sources, and thus I tag every clip manually on import. I’ve never tried using the clip metadata. I don’t trust that crap. Haha.

In your example, I’d tag the Slog clips as Slog, and the rec709 clips as rec709. If I then create a single “ACES to SDR Video” viewing rule that will apply to “Any Scene Linear or Log” and that’ll make all the log footage from various sources look similar (in theory) without altering the color.

Now if I then had to GRADE everything, I’d apply a color management node (or a green CM timeline effect) set to Input Transform “from source” to “ACEScct”. On an Slog clip the image won’t look different, but it will be converted from Slog to ACEScct. If I had a LogC clip it would also be converted from it’s source to ACEScct. You could also do this on import. Again, you have to manually tag every clip, but once that’s done the rest falls into place nicely.

2 Likes

I know how you feel, brother. The most discouraging is when you see that there is no solid and clear way to do colour management. Now I’m wondering why there are three different f**d ways to convert alexa footage.

Despite all, I think ACES will bring you a rather safer way. Keep in mind that the IDTs (input transforms) are provided by the manufacturers themselves and here we have only one way to do the transforms.

The downside of ACES is that certain behaviors are a bit complicated. Grading, even using CCT, sometimes unexpected things happen. Dealing with negative values in comp, etc. It needs a kind of re-learnig and re-reformat mental process. I’m working in my first job in ACES, and I’m taking it as my first day in school. But I think that any workflow out out Aces, to mix different footage is just insane.

The big drawback is the transform between ACES and rec709. It seems not work perfectly, specially with graphic supers, so is needed split the workflow, and this sometimes becomes in a big monster.

When you create a new view transfrom rule in preferences>colour management you can select “add new” in the contextual menu from the view transform column, and select your personal lut.

Nope. Input rules are not effective since a file format is not specific for any footage. Even EXR can be sometimes linear, or sometimes rec709. It’s something that vfx industry has to fix using specific metadata standard. Input rules only read file names (file extensions or whatever part you can set). One solution could be use specific name suffixes.(ie: “_sLog3.mov”) but it’s simply impossible ask such thing to the rest of the team, or infinite rename bach process of source footage… . But there is a useful option in flame, “edit color space” (a contextual selecting clips in desktop). It allows multi-select. You can import your footage first as “unkonw” , sort out in different reels by footage, and select them to tag in only one step.

I’ve been using ACES over the last year and do a lot of multi camera editing/grading as well…I do think it makes matching the cameras easier once you get it all sorted out.

As Andy mentioned, input rules are there to help automate the tagging process so don’t sweat it if it’s not sorted out right away. You can batch tag footage thru the media hub, so manually doing it really isn’t that time consuming. I usually manually link shots during conform anyway (we shoot red, so there’s always some settings that need tweaking) so it’s easy to tag by footage type at that time.

But now that I just said tag, I actually usually use autoconvert and convert footage to ACEScct since I’m mainly grading and cleaning up vs vfx.

The real time saver is once you get all the varying footage into ACES, you can then create a Source Sequence using connected conform and grade there…I find it easier to match when all the clips are grouped together. Grade, sync back to the edit timeline…boom, much quicker than shot by shot.

Thanks for taking the time folks…appreciate it.

As i continue to dip my toes in this…let me throw this out.

:30 second spot…final deliverable is Rec709 for broadcast. Spots consists of 10 shots, all shot Slog3.
I create the project using the ACES policy. I have it set to auto convert and I created a new viewing rule using the Sony LC 709Type A transform as opposed to the default Log transform as its more of an accurate representation of the camera sensor.

Bring my 10 shots in, cut them into a sequence. Footage was exposed properly so all looks good (thank me, i’m welcome) however, that’s not the look i want…first 5 shots need a desaturated look and the last 5 need a warm look.

I take the first shot into batch and since its tagged w the TypeA transform that’s not the starting point i want, i need to un- tag it, then start over w a variety of nodes/LUT (which i have prebuilt) to get the look i want. And at that point i would normally drag and drop that look across the other 4 shots and be done. However, since the other 4 shots are tagged I know drag and drop isn’t going to work so i would then need to un tag each of those shot before dropping the new look on them.

Now using the above workflow…it seems to me i’ve essentially by passed the ACES aspect of the work flow and essentially ended up back to my legacy work flow.

And even if i didn’t use auto convert and manually tagged the various shots with the various custom LUTS…still seems to be about the same amount of work as my established work flow.

Now it might seems like i’m trying to talk my out of an ACES workflow, i’m not…but my little pea brain still can’t resolve what the advantages might be using ACES over what i’m doing. Am i nuts?

to be clear, you’re using luts to do your grading?

Because the way I’d approach that spot via ACES is:

–import all footage with an auto-convert from Slog3/SGamut to ACEScct
–conform the timeline
–grade the shots using Image nodes as timeline FX.
–apply the “ACES to SDR Video” viewing lut as a gap effect over the whole spot to convert it to Rec709.

ACES is only better insofar as all the camera folks have agreed “this is how our footage looks correct under ACES.” and that can be handy, though I find most of the time in commercials It’s just yet-another-colorspace.

I’d say it’s worth understanding, but not worth dying for. My own limited understanding of all this has taken about a decade to feel like I have any sort of grasp of it.

My question is: Is there a better result or experience grading in this example working in the ACES world for rec709 jobs. I have yet to grade with that workflow but am interested in the benefits.

My short answer is “it depends”

but before I get into a longer answer, I’d like to state generally for the thread and not to anyone specifically:

ACES is not a color space. It’s a bunch of them:

"Big ACES" aka ACES 1.0 and 1.1. The color warehouse. It’s Scene Linear (aka photometrically correct) and has a giant gamut. Use only to convert or store media, not for working because you can get negative gamut values.

ACEScg - Scene Linear with a smaller (but still giant) gamut. This is the one you render CG into and do comps with.

ACEScc - A logarithmic version of ACEScg for color grading and any other log-type uses.

ACEScct - A slightly different logarithmic version of ACEScg for color grading and any other log-type uses.

I’d say that grading in ACEScc and ACEScct is as nice as grading in any other log format is.

Similarly, grading in ACEScg feels just as bad as grading with any other linear format.

I’m a huuuuuuge fan of everything logarithmic, so I’m biased towards grading that way, but all roads lead to the same end (except grading in linear–that road leads to hell). If rec709 is all the job needs, why complicate things?

3 Likes

Even though I am the biggest fanboy of ACES and it sounds like it would be great for your array of different camera sources in a editorial context, Its certainly not a must have for all jobs, it can make your life easy, or painful or both.

Where it really shines is mixing stuff in a compositing and CG rendering context, mattepaintings and CG renders just fit way better right off the bat because they are beign merged together in the same colorspace. So for any VFX work I think there isnt really another choice nowadays.

Same goes for having the same view transforms across all the aces supporting apps, aces rendering might not be everyones cup of tea but you can flip things like vray VFB to ACES and see the same thing as in flame, thats where it really shines, its a standard.

The other big thing is that you can just switch the display transform to whatever you want and get a pretty good match, going from a rec709 display to a P3 projector etc.

but then having weird unsupported camera footage like random rec709 footage or unknown log formats can make your life horrible and you need to know how to work “around” aces while beign in a aces managed project it can become clumsy.

another fun part is negative values and neon blues clipping …
Flame sadly is not really liking anything linear too much,

All in all ACES is awesome and I use it for everything as I deal with VFX heavy thints… but dont know if its ideal for “simple” finishing jobs, probably not.

2 Likes

Partially yes. I have a number of “node trees” in my user node bin that i use. The “tree” consist of a Master Grade > Color Warper > Color Managment (preloaded w a particular LUT). The LUT gets me close and the MG & CW upstream allow me to dial it to taste if need be. I will try your example next week…because that was my next questions…“how do i spit it out?” Thanks for your replies…it is appreciated.

Thank you…that’s helpful. I don’t deal w/ vfx (at least hardcore vfx) that much. My scenario would be mixing and matching a variety of LOG and Rec709 sources…ala long form doc. And at its core, is ACES a better solution to help match (or at the very least) take a bunch of random sources and get them in the same area code. I’m having a hard time wrapping my head around the process, and all the ins and outs, so its hard for me to determine whether its a road to continue to go down or not.

Yea aces should be fine for you, where are you stuck?

what you did manually kinda sounds like your own color management system , so thats totally valid, I think the biggest issue for you will be the difference in rendering look and the difference in how your tools will feel and you arent raking in too many of the benefits of aces from what it sounds like.

You dont even have to use all of aces, maybe you just want to use the IDT part and auto-convert everything to acesCCT (or arriLogC or whatever you feel most comfortable with, maybe slog3) and then use a common viewer lut across all clips , wether thats the aces rec-709 or something custom or the sony one you can build your own stuff on top of the foundations of aces or colormanagement in generel, I feel like for you the IDTs are the most valuable thing.

And thats how I would do editorial and finishing for mixed camera sources for long form docs as well, transfor whatever input to LogC/AlexaWidegamut and then use the arri rec709 LUT as my viewer and then jusr burn that viewing lut in for final output. but thats just because I am so used to grading and working on LogC/alexaWide footage.

it really help to lay out all the steps of aces in manual nodes in batch to visualize whats happening where and why. I will go over this in a logic live on 3/14

1 Like