Quick question as I’m not a grader. I’ve got a 4 min film, all source material is ARRI log. I’m in aces cg overall. flame has applied all the correct luts and all looks pretty good. if I want to grade it properly is it better to get rid of all the luts run the timeline as source ARRI and grade via the image node or is it better grading after the lut ? Thanks
Heya Jon,
In my opinion the most important thing is that you understand the implications of either workflow. I know of very successful colorists working either way you described.
When you say Flame has applied all the correct LUTs, I’m assuming you mean that you’ve tagged the footage as ARRI Log C V3 or V4 and then since you said your project is ACES CG then you’re probably using one of the ACES SDR View Transforms which makes it look more normal and less in a Log state - aka color management.
The long and short of it is:
If you grade color managed, then when it comes time to exporting this to send elsewhere, you will have to apply a View Transform either in the timeline or on export before it will look the way you’re seeing it in Flame. In this way you are leaning on the established color management of ARRI & ACES working together to get you to the baseline of what they believe should be the starting point. If you Bypass the View Transform in the viewport you’ll see that you are still working on the log data, it’s just displaying a more “normal” display of what that footage “should” look like. You are not necessarily baking the LUT in, unless you go out of your way to do so.
If you remove the tagging and work with no color management, you will need to work “harder” to get it back to that state that the ACES color management gave you for free, but you also won’t have to worry about applying a View Transform before getting it to look good, because when you aren’t working color managed, you’re working in a world where what you see is what you get.
If with the color management it looks pretty good, I’d say use that as a starting point and work on top of that. But just remember you need to do that View Transform on export or you won’t get what you want. Let me know if I’ve got this wrong and happy to chat more about directions and such.
About the first part, I’m not sure there’s a technically correct answer to that. If we’re talking about wide color spaces, ARRI, ACES, it might just come down to personal preference. Colorists usually have their reasons; sometimes it’s due to very technical curve’s behavior, or it’s simply based on their own preferences.
But I only wanted to point out is you decide grade “before the lut”, you don’t need to remove any color management node of your segments or change anything. You just need to use the green version of the “image”, and it will be applied before of the color managment within flame’s internal workfllow. Arri log for this case.
Thanks everyone. I’m going to try both routes and see what gives me the best results. I think keeping everything as ARRI log is the purest I reckon
As mentioned, there is no ‘correct’ answer. These workflows have evolved over time.
What is important is that you have a set of color tools that matches the material, so that a move of the dial has the right coarseness of impact. Log vs. linear makes tools feel different. Which is why in MasterGrade, in the control section, you can pick the type of material you’re grading (LogC = log, ACEScg = linear), and the feel of the controls will adjust.
The ACES workflow’s primary advantage is in mixed material timelines or if you have to deliver for multiple output formats. Barring that, the advantage diminishes.
It’s worth noting that in most color pipelines, the preferred ACES color space for grading is actually ACEScc or ACEScct, which is not linear. Those exist to make ACES color feel more familiar to traditional workflows.
When you say ‘grade after the LUT’ there are two traps - in an ACES workflow, you actually have multiple transforms - one from source to working space, and one from working space to viewing space. Generally, you want grade between them in an ACES workflow.
The second trap is that Flame runs its color pipeline differently than all the other apps. When you tag a clip, it actually doesn’t get transformed until the end, unless you tell it otherwise. You may not actually be working in the space you think you are in. All other apps do actual color transforms at each stage.
Probably too much detail for your simple question. Try it. If it looks right, it works in your simple use case.
Yes in your case I’d stick to ARRI log (viewed via ARRI trim/view lut: Alexa Rendering)
And if there was any comp work required on any shots then (depending on the nature of the required comp work) I’d convert ARRI log to ACEScg for the comp work (using the input-transform node) and then input-transform back to orig ARRI log space.
So the workflow would be:
For non-comp shots:
Source/ARRI log > grading (green image node) > ARRI trim/view lut (Alexa Rendering) for both viewing while grading and SDR outputs.
For the comp shots:
Source/ARRI log > ACEScg > Comp work (while being viewed via ACES SDR trim/view lut > comped shot (still in ACEScg) back to ARRI log > grading (green image node) > ARRI trim/view lut (Alexa Rendering) for both viewing while grading and SDR outputs.
The above workflow allows for you to use either standalone batch or BFX for comp work while keeping your main level timeline at source space/ARRI log with the ability to use green image nodes for grading for both non-comped and comped shots.
Feel free to DM if you have any questions, be more than happy to go over anything. Much simpler/easier to chat/demo than what it appears to be while writing it out.
As a personal choice I prefer staying away from automated color management (included automated tagging) for various reasons but there is no harm in using automation if you are more comfortable with it, and the above mentioned workflow should work the same with both automated and non-automated color management.
Side note: In case you have/had another camera space in the mix (Sony, canon, etc) then right before grading I’d convert both/all camera colorspaces to ACEScct (cct flavor designed specifically for grading) so matching the cameras on your grading timeline is easier and mathematically correct and the grading controls/sliders are responding uniformly and yielding predictable results on the varying cameras’ colorspaces on your timeline.
Good insights here! Thanks for that @AamirQ
I’m curious though- while we’re talking about this, are you (or anyone else reading) grading Alexa footage in Flame 2026+ and using Alexa Rendering? If so, how? I think I’ve managed to figure it out by loading a Legacy OCIO preset during project creation, but the default ACES 2.0 preset doesn’t come with Alexa Rendering anymore. Curious if I’m missing something or if loading a legacy preset is the way. For the last few grading projects I’ve done I’ve been conveniently cursed with mostly Rec.709 footage so it hasn’t actually come up but figuring that out has been on my mind since the 2026 release.
great stuff - thats pretty much what im doing - so good to know its a valid workflow.
Very welcome @Jeff. I’ve been holding off on installing 26 and currently on 25.
Not to discourage its usage as some R&D $$ have been invested in automation and I am sure it works great for many of us here, but overall I am not a proponent of using automated/preset color management for a few key reasons. I do however sometimes run preset col mgmt but only for quick R&D runs to save time and its convenient for that, but never for establishing actual delivering workflows.
And while working with manual or user color management (“Legacy” is not the correct term for it and is misleading), what I like to do for ARRI log spaces is create/load/add Alex Rendering trim/view lut in the view defaults list. This way I am bypassing ACES altogether if there is no need for me to go thru it.
But the other more important reason for me to bypass ACES view trim is the fact that for ARRI spaces I very much prefer the Alexa Rendering due to its much creamier/richer rendering and smoother rolloff at both ends compared to ACES view trim. Maybe subjective and/or by design but I find ACES SDR trims to be bit jarring sometimes. Even if the creative required for both or either end to be crushed, I prefer the smoother rolloff right up to the tail end and post the trim pass for having control and optionality right up to the end.
In fact for times when I need to convert to/use/deliver ACEScg/cct + ACES SDR/Rec709 view trim, to mimic the Alexa Rendering’‘ gamma characteristics, I have created a mild custom gamma curve lut/node which I can drop right before the ACES SDR/Rec709 view trim to mimic the Alexa Rendering’ gamma. Works great if needed, and the following is the pipeline:
ARRI Log > ACEScg / cct > comp / grade > Alexa Render mimic gamma curve node > ACES Rec709 view trim.
I can post the method with screenshots how I created this Alexa Render gamma mimic curve if anyone interested.