How to get correct gamma from linear EXR sequence?

I'm trying to learn compositing just in Fusion Studio besides Resolve. But when I bring in an EXR sequence from Blender, which is rendered in linear space, not a single one of the presets is linear, unlike Resolve where linear is an option in the VFX LUTs, and I have an additional LUT that converts linear to Rec.709 from a tutorial I saw.

So the only option left is the custom one, but that's eyeballing it. Which is OK I guess, but I can't believe that a program that is used mostly for compositing for major movies doesn't have an easy option to load EXR, which if I read correctly, it's a format created by ILM for the purpose of rendering from 3D software.

There's something I'm missing, right? I mean, sure I can add a color space transform to it, but the loader has a long list of gamma options, so it has to have something to load linear EXRs, right?

So the only option left is the custom one, but that's eyeballing it. Which is OK I guess, but I can't believe that a program that is used mostly for compositing for major movies doesn't have an easy option to load EXR, which if I read correctly, it's a format created by ILM for the purpose of rendering from 3D software.

There's something I'm missing, right? I mean, sure I can add a color space transform to it, but the loader has a long list of gamma options, so it has to have something to load linear EXRs, right?

Fusion is natively linear and most tools in it, expect linear to work the best. Linear EXR is essentially native to Fusion, so its assumed its what you already want to work with. If you want to convert it to something else, there are nodes like gamut node and others for that.

The Gamut node has controls to transform one color space to another and remove/add gamma curves. This node, along with the Cineon Log node, is primarily used to linearize incoming images and then reapply the applicable output gamma curve at the end of a node tree.

A Gamut node is most often placed directly after the MediaIn node in DaVinci Resolve or a Loader node in Fusion Studio. Another Gamut node is usually placed at the end of a node tree before a MediaOut node in DaVinci Resolve or a Saver node in Fusion Studio.

Output Space

Output Space converts the gamut to the desired color space. For instance, when working with linearized images in a composite, you place the Gamut node just before the Saver node and use the Output Space to convert to the gamut of your final output file. You leave this setting at No Change when you want to remove gamma using the Source Space control.

NOTE When outputting to HD specification Rec. 709, Fusion uses the term Scene to refer to a gamma of 2.4 and the term Display for a gamma of 2.2.

Remove/Add Gamma

Select these checkboxes to do the gamut conversion in a linear or nonlinear gamma, or simply remove or add the applicable gamma values without changing the color space.

Pre-Divide/Post-Multiply

Selecting this checkbox causes the image’s pixel values to be divided by the Alpha values prior to the color correction, and then re-multiplied by the Alpha value after the correction. This helps to avoid the creation of illegally additive images, particularly around the edges of a blue/green key or when working with 3D-rendered objects.

When the gamma is linear usually metadata info graph shows nothing since its the native space for Fusion. When there is something else, usually it will show up in metadata. Here is easy way to work with linear (EXR) only one I had handy was a HDRI image converted to linear EXR. And a parrot in Rec709.

sshot-3834.jpg (136.14 KiB) Viewed 4529 times

sshot-3835.jpg (175.04 KiB) Viewed 4529 times

sshot-3836.jpg (163.13 KiB) Viewed 4529 times

Of course when you work with Linear in your node tree , the viewers will need to use viewer lut or buffer lut, if you are in 3D workspace. As a rule of thumb you want linear which is natively expeted in your node tree and you want to also be seeing what it is output color space going to be, in your viewers. and you do that by using viewer luts. If you don't know what it is, reference the manual.

Thanks for all the info! It just seems weird to me that you can select so many different gamma presets but not linear. Because I understand that Fusion uses linear natively, but you can't work in linear because it's extremely dark. But well, at least you can choose the Color Space Transform preset for the LUT menu, then edit to choose Rec.709 and you get the same result. Just seems a bit convoluted, but it works, same as the Gamut node.

Because I understand that Fusion uses linear natively, but you can't work in linear because it's extremely dark."

No, no you can and should work in linear space, and while the image is dark, assuming you want to see it as gamma 2.2 or 2.4, that is why we have viewer luts.

I suggest you learn how Fusion is built to work and use it for its advantages, rather than trying to make it work like you want it to work based on some personal preference, and you will avoid frustrations. You said you wanted to learn compositing in fusion. Here is your chance. I see a lot of people who don't even understand the fundamentals of some applications, like fusion in this case and yet are full of suggestions how to change everything because they think they know better, or have personal preference that ignores reasons why things are done the way they are, nevermind the 25+ years of people working and tweaking the application to be good for what it supposed to do.

Linear workflow set up in Fusion is well though out and very flexible. Once you understand how it works and how to use it, it makes working with all sorts of footage or images predictable and flexible. The way how independently you can have two viewers to preview what you are doing in all kinds of ways is super handy for many compositing tasks that would be handy to have in many other applications. Including Color Page of Resolve.

gamma transform is not enough to display your renders from Blender.
what is your view transform in Blender? filmic? agx? you should simply bring what you are using in Blender to Fusion via OCIO.

Fusion is natively linear

In what way would you think it is natively linear? What does it mean? Last time I checked it is fully up to the user to decide what to do with any data inside Fusion.

Fusion is a giant graphing calculator. It works with raw numbers. On a graph from 0.0 to 1.0, half way across is 0.5 or in a channel of color 0.0 is black, 1.0 is white and 0.5 is mid gray. That's a linear interpretation of color. It's a straight line.

Where things get real funky is when you use a gamut like sRGB (aka viewable computer screen). Now mid gray is no longer at 0.5, it's at something like, 0.24 or something. I'm not in front of my computer. So all the math that you would use to add to make a number higher or lower either cranks way faster in certain ranges or has to be aware of the color gamut and compensate for the weird curvature of something like sRGB.

A linear image is one where the highlights and shadows fall across the whole range from 0.0 to 1.0. it gives equal amounts of information to each pixel? When you use an sRGB. It shoves all the highlights and shadows into a really small area on the edges and makes adjusting those hard because there's not enough pixels to cover the breadth of the image.

To make it more confusing, the whole idea is intertwined with bit depth. A sRGB jpeg in 8bit only has 256 colors per channel to pull from. It's a glorified gif to a compositing program like Fusion. You work linear so that all your numbers are easier and that really only works when you have billions of colors so that you math can get really subtle. That's why compositing works with floats instead of integers if you can help it.

It's the same idea as analogue audio vs digital. You want as smooth a wave from as possible so any stepping from the number of bits just makes for gaps in information. Same with color.

So at the end of the day, you would work linear and then add in the special sauce curvature at the end to make it an sRGB image. When somebody brings up something like ACES or Filmic, that's just another special sauce to pull the colors around to give it a look that starts at linear but gets tweaky. So you have to know what the exr's special sauce is and remove it to make it linear so you can work with it. That's a simplistic take on it because something like ACEScg is close enough to linear that all the tools are indecernable in their use but you could linearize of you want to.

Fusion is a giant graphing calculator. It works with raw numbers. On a graph from 0.0 to 1.0, half way across is 0.5 or in a channel of color 0.0 is black, 1.0 is white and 0.5 is mid gray. That's a linear interpretation of color. It's a straight line.

0.5 is mid gray in nonlinear interpretation, unless you mean just middle value, which has little to do with mid gray in color perception sense. In linear domain, mid gray (perceptibly half way between black and 1.0 white) is around 0.18. In gamma-corrected images, like sRGB, correction lifts that value to somewhere around 0.45. What the display does with the values is another story, but in the end what must be achieved is light emission somewhere roughly around 18% of maximum luminance to be perceived as middle gray.

Where things get real funky is when you use a gamut like sRGB (aka viewable computer screen). Now mid gray is no longer at 0.5, it's at something like, 0.24 or something. I'm not in front of my computer. So all the math that you would use to add to make a number higher or lower either cranks way faster in certain ranges or has to be aware of the color gamut and compensate for the weird curvature of something like sRGB.

sRGB gamut has no correlation to nonlinarity, it is gamma that affects it. sRGB encoding lifts the middle value due to applied gamma correction. Math is math, question is rather, whether it is easy or not to model some physical behavior using simple operations like addition or multiplication. Linear-light domain allows handling values as if they are quantities of light (within limitations) where most effects are linear by nature.

A linear image is one where the highlights and shadows fall across the whole range from 0.0 to 1.0. it gives equal amounts of information to each pixel? When you use an sRGB. It shoves all the highlights and shadows into a really small area on the edges and makes adjusting those hard because there's not enough pixels to cover the breadth of the image.

There is equal amount of information in each pixel whichever gamma you use, question is what that information means. Linearized representation, not sRGB, bunches a lot of data into low end because each stop up has twice the value range: 0.25-0.5 to 0.5-1.0 to 1.0-2.0 etc. Gamma-encoded imagery (and log encoding) alleviate that by bending the encoding curve so that more code values are used for storing the low end, while compressing the highlights. I won't go into detail about why it is useful, but that's what happens.

To make it more confusing, the whole idea is intertwined with bit depth. A sRGB jpeg in 8bit only has 256 colors per channel to pull from. It's a glorified gif to a compositing program like Fusion. You work linear so that all your numbers are easier and that really only works when you have billions of colors so that you math can get really subtle. That's why compositing works with floats instead of integers if you can help it.

Bit depth and linearity are not correlated, bit depth only affects precision. 1bit image can be perfectly linear and cover humongous dynamic range, it just has no granularity inbetween its range.

That's a simplistic take on it because something like ACEScg is close enough to linear that all the tools are indecernable in their use but you could linearize of you want to.

ACEScg is literally linear, not close enough. Only difference from sRGB is the gamut (color primaries and whitepoint).

Long story short, your description is somewhat backwards. I blame the hideous misuse of terminology in Resolve for that.

Fusion is natively linear

In what way would you think it is natively linear? What does it mean? Last time I checked it is fully up to the user to decide what to do with any data inside Fusion.

This ^ but given that loaders by default don't linearize and there is no default viewer LUT and color palette is pure system sRGB one could even say Fusion is natively sRGB. Obviously that sounds wrong, but that's how it is. Compare it to Nuke for example, which by default linearizes all reads, applies a display lut and has linear-value color palette...

Thanks Ben for your effort to explain and thanks Hendrik for the necessary corrections

The reason I questioned the statement is because I believe there is a common misconception or misuse of the idea of working in linear. Software only crunches numbers. It has no concept of linear versus anything else. When you tell an operator to multiply one value with another, you get the same value as you'd get from a normal calculator. It is totally irrelevant if the state of the incoming image as a whole is considered linear or not.

Fusion is natively linear and most tools in it, expect linear to work the best. Linear EXR is essentially native to Fusion, so its assumed its what you already want to work with.

So Fusion is not natively linear and neither do the tools expect it.
The only difference is that it is practical for several reasons to crunch the numbers in the linear domain because it allows for behavior that is relatively the same as the real world, allowing for consistent and predictable results.

I'm trying to learn compositing just in Fusion Studio besides Resolve. But when I bring in an EXR sequence from Blender, which is rendered in linear space, not a single one of the presets is linear, unlike Resolve where linear is an option in the VFX LUTs, and I have an additional LUT that converts linear to Rec.709 from a tutorial I saw.

So the only option left is the custom one, but that's eyeballing it. Which is OK I guess, but I can't believe that a program that is used mostly for compositing for major movies doesn't have an easy option to load EXR, which if I read correctly, it's a format created by ILM for the purpose of rendering from 3D software.

There's something I'm missing, right? I mean, sure I can add a color space transform to it, but the loader has a long list of gamma options, so it has to have something to load linear EXRs, right?

ok.
so basically what you are looking for is Gamma space: None .
or any other "linear gamma" or "no-gamma" (gamma often called transfer function nowadays) like: ACES, ACEScg, RED wide gamut RGB etc.

but you do not need that. your exrs (if saved properly, without any transforms from Blender's OCIO) are already linear (exr format is not made for storing gamma encoded (or display referred) image data (in half float exr that is most common in cg you are left with precision comparable to 10bit integer - rarely enough for post production).

edit: Gamma space in Loader node is useful for loading log encoded footage from camera for example and linearizing it for compositing.

also "Rec.709 lut from a tutorial" does not sound like something you are using in Blender to view your renders - you should stick with OCIO in both Blender and Fusion for simplicity.

Thanks Ben for your effort to explain and thanks Hendrik for the necessary corrections

The reason I questioned the statement is because I believe there is a common misconception or misuse of the idea of working in linear. Software only crunches numbers. It has no concept of linear versus anything else. When you tell an operator to multiply one value with another, you get the same value as you'd get from a normal calculator. It is totally irrelevant if the state of the incoming image as a whole is considered linear or not.

Fusion is natively linear and most tools in it, expect linear to work the best. Linear EXR is essentially native to Fusion, so its assumed its what you already want to work with.

So Fusion is not natively linear and neither do the tools expect it.
The only difference is that it is practical for several reasons to crunch the numbers in the linear domain because it allows for behavior that is relatively the same as the real world, allowing for consistent and predictable results.

true.
the only thing you can say about Fusion in this regard is that it interprets numbers in 32bit float and that is it.

Well, every day is a good day to learn something. Of coarse I explained the entire thing backwards.
Thanks Hendrik, for putting in the time to correct all that.
The only thing that you said that I question is the ACEScg being linear. If you add a Gamut to an ACEScg file and set it to sRGB, it won't come out correctly. There's a lot of tweaking in the highlights that make it just different enough to warrant doing it correctly with an OCIO or aces transform. As I understand, AP1, which ACEScg falls under is non linear. AP0 is the raw linear format? It's been a minute since I looked at the definitions though.

Well, every day is a good day to learn something. Thanks Hendrik, for putting in the time to correct all that.

No problem, I hope it made sense too.

The only thing that you said that I question is the ACEScg being linear. If you add a Gamut to an ACEScg file and set it to sRGB, it won't come out correctly. There's a lot of tweaking in the highlights that make it just different enough to warrant doing it correctly with an OCIO or aces transform. As I understand, AP1, which ACEScg falls under is non linear. AP0 is the raw linear format? It's been a minute since I looked at the definitions though, so I could be wrong.

ACEScg is a combination of ACES AP1 gamut and linear transfer function. Gamut can't be linear or nonlinear, gamut is gamut. AP1 and AP0 can both be linear or in case of AP1 it is also used in colorspaces with log transfer function. AP0 linear is ACES2065-1 colorspace, AP1 linear is ACEScg. Here we are again bumping against the stupid semantics of Resolve and Fusion (Nuke suffers from it too) where "linear" is handled as if it were a colorspace, but it isn't, it only describes the transfer function part of a colorspace.

ACEScg linear and sRGB linear do not match due to gamut difference so to make them match you need a gamut conversion. If that doesn't work, what specifically are you doing and how? One common trip-up is expecting ACES color management to produce sRGB images through ODT output that are matching pure technical transform from ACEScg to SRGB. That is not the case due to render transforms built into ACES, it is not a bug, it is expected.

Well, every day is a good day to learn something. Thanks Hendrik, for putting in the time to correct all that.
The only thing that you said that I question is the ACEScg being linear. If you add a Gamut to an ACEScg file and set it to sRGB, it won't come out correctly. There's a lot of tweaking in the highlights that make it just different enough to warrant doing it correctly with an OCIO or aces transform. As I understand, AP1, which ACEScg falls under is non linear. AP0 is the raw linear format? It's been a minute since I looked at the definitions though, so I could be wrong.

Gamut node is not a display transform.
ACES2065-1 is linear
ACEScg is linear
sRGB encoded AP1 texture is ACEScg space texture with sRGB transfer function.

Awesome, thanks for the info guys. I piece mealed my understanding of linear work a long time ago and some bad info stuck in my head. As soon as I read the exposure stops part, my brain kicked into gear.

Thanks, Marcin for what you added as well. There's a distinction between Out-sRGB and sRGB texture in ACES. I was referring to the Out-sRGB. Unless I'm in a 3D program, I use an OCIO set to Out-sRGB to move traditional images in sRGB back and forth to Aces. Otherwise, client colors go out the door.

Usually, I will just throw a Gamut or OCIO at the end of the flow after the MediaOut or Saver and use it like a display transform but as Chad said, it's 1000 times slower so maybe it's time to do it the right way.

also "Rec.709 lut from a tutorial" does not sound like something you are using in Blender to view your renders - you should stick with OCIO in both Blender and Fusion for simplicity.

Thanks for your detailed response, I need to read it more carefully to get a better idea of what you're talking about, but regarding this, I know that "Rec.709 lut from a tutorial" sounds like a joke and I see your point about Blender.

But here's the thing. For me to see in either Resolve or Fusion Studio my Blender render as I saw it in Blender itself with the render preview, I need to apply that LUT. Otherwise, it looks darker and the colors are overly saturated.

Now, I'm not sure what OCIO is, it definitely rings a bell as a very new color technology, but I just checked and do not have that in Blender. Just to clarify, I'm on Blender 4.1.1 on macOS, and I have some options regarding color management, but none of them is called OCIO. But I will definitely look into that as soon as I can.

Now, I'm not sure what OCIO is, it definitely rings a bell as a very new color technology, but I just checked and do not have that in Blender. Just to clarify, I'm on Blender 4.1.1 on macOS, and I have some options regarding color management, but none of them is called OCIO. But I will definitely look into that as soon as I can.

OCIO can be seen as sort of the backbone for how the colormanagement in Blender or any other supported app works. It's a mechanism to define, manage, convert color spaces via a system that is open to the user and can be highly customized to someone's needs. Blender has been using this system for their color management for a long time already, and they don't offer any other mechanisms, hence you won't see the name itself mentioned.

Here is a video I just looked up that covers using the blender config in Fusion/Resolve, assuming you used AgX for rendering this should match.

The guy missed quite important bit of info about AgX:
the three versions (ocio and dctl from Troy Sobotka and ocio from Blender 4) are different.

Lets see what display transform OP is using with Blender and we can take it from there. The config might need some modification for Resolve (if I recall this correctly ocio looks are used).

The DCTL offers customization of the AgX rendering and may not default to the same as Blender yea. For true parity it would be best to load the Blender OCIO config from your install instead.

I think the rest of the video does provide enough information on how to handle it.

The only 'problem' with Resolve is that it doesn't allow OCIO to be used as management outside of Fusion. If the desire is to process the image through grading on the Color Page the only alternative would be some custom setups and baking out the AgX DRT as LUT which is too technical for the OP to get in to...

Life would be a lot easier if OCIO would be supported on project level in Resolve.

Not different defaults, completely separate transforms - the idea is the same but all three operate in different gamuts and have clearly different looks . Blender version is also developed by other people.

Ah, I see. So Blender’s AgX was further developed to suit their needs probably to stay consistent with the established filmic config since both are availabe within a single config. Troy’s AgX is the original version and thus neither the DCTL or OCIO from his GitHub can be used. Only the Blender OCIO config.

Ah, I see. So Blender’s AgX was further developed to suit their needs probably to stay consistent with the established filmic config since both are availabe within a single config. Troy’s AgX is the original version and thus neither the DCTL or OCIO from his GitHub can be used. Only the Blender OCIO config.

Oh, every time I have to deal with color management it makes me so so so happy I can't even begin to tell you. Paradise for me would be eternity having to deal with color profiles and everything else to color management. Who wants simplicity? What am I, a simpleton that cannot deal with some challenges of the colorific kind? No sir, I love wasting time I could be otherwise using learning the stuff that is really worth learning.

So, working on a passion project in Blender that I'm bringing into Resolve to edit, and I realized that I need to adjust something in Blender because the render looks really different in Resolve, even after applying the linear EXR LUT. If I don't apply the LUT, it's even worse. But with the LUT, it's still very different, with the reds being overly saturated, so much that I can't even see the texture I added to it, or the borders of the red cubes. So basically, it sucks.

So question for those of you who use Blender and composite in either Resolve or Fusion, what settings do you use in Blender, and then in Resolve or Fusion?

Color management may be discouraging to some because of the initial complexity. But it's not a good mindset to think that it is not important. The image doesn't 'magically' look pretty. Knowing why that is, how it works and how to manage it is crucial in any digital artist's skillset.

I don't know what your intent is in Resolve but if you really desire a "I don't want to deal with it" workflow and just edit the shots into a sequence, consider not rendering EXRs but PNG/JPGs that aren't linear but rather the image as you see it in the Blender picture viewer.

If you want the full control method, the only way is to start learning these mechanics and principles. At least enough so you can know how to manipulate and treat the image for final quality.
What you need to setup was already mentioned. You need to use the Blender OCIO config inside Fusion to convert from linear to whatever you had set up in the color management settings in Blender.

Color management may be discouraging to some because of the initial complexity. But it's not a good mindset to think that it is not important. The image doesn't 'magically' look pretty. Knowing why that is, how it works and how to manage it is crucial in any digital artist's skillset.

If you want the full control method, the only way is to start learning these mechanics and principles. At least enough so you can know how to manipulate and treat the image for final quality.
What you need to setup was already mentioned. You need to use the Blender OCIO config inside Fusion to convert from linear to whatever you had set up in the color management settings in Blender.

I was only half joking, I know that stuff is important, it's just frustrating, especially when you deal with printers and printing photos on paper that costs money and ink that also costs money.

But regarding Blender and Resolve/Fusion, I realized that setting Blender to AgX was causing me to see a render preview in Blender that was very different from the actual rendered file. After doing a few tests, I set it to Display Device sRGB, View Transform to Filmic, Look to Medium Contrast, Exposure and Gamma defaults, and Sequencer to Rec.709. That renders a file that to my eyes is identical in Resolve to what I see in a Blender preview, and then in the render window.

Well, identical only in the main Resolve window in the program monitor, because for some reason we're on the fourth beta of Resolve and Blackmagic still hasn't addressed the problem of the clean feed displaying a different gamma than what should be. At least when the monitor is just another computer monitor, I don't know if that's the case with the expensive Ultrastudio 4K and Decklink, etc.

But I have two identical monitors connected to my Mac Studio, both set to exactly the same picture settings (I just checked again), the gamma is off in the clean feed, displaying everything slightly washed out. That is the case in every page, except the Fusion page. You go page by page and it displays washed out, you get to the Fusion page and it displays perfect, just like the program monitor in the main window.

I just don't get it. How hard can it be to correct the clean feed in all the other pages to display the correct gamma? Now, if you have a monitor with plenty of controls, you might be able to change the gamma in the clean feed monitor for when you work in Resolve, but then you'd have to change it back when you're done if you want to use it for something else. And that, depending on which monitor you have, might be a pain, like these Samsung LF32TU87 monitors I have which have one joystick and nothing else to control everything.

After doing a few tests, I set it to Display Device sRGB, View Transform to Filmic, Look to Medium Contrast, Exposure and Gamma defaults, and Sequencer to Rec.709.

for what it's worth - if you need to match that transform in Fusion with Blender's OCIO - use those settings, no need to modify the config:

So question for those of you who use Blender and composite in either Resolve or Fusion, what settings do you use in Blender, and then in Resolve or Fusion?

I have OCIO setup in Blender and use ACES in Color management, rendering out in ACEScg with a Rec709 view transform in Blender. Works well in Fusion as I work in ACEScg there so no need to do any transform on the Blender Renders. I use a Rec709 viewer in Fusion and usually render out ACEScg for delivery.