How to deal with HDR+SDR content blending?

Discussions - Using ACES
1

sRGB and Display P3 specifications both requires the reference display to be calibrated to power 2.2 TRC curve while using piecewise sRGB TRC curve for content encoding. This would be fine when dealing with SDR content + factory-calibrated monitors. However, things would get worse when using user-calibrated ICC or HDR+SDR mix, since they have the TRC data accurately consistent with monitor calibration. sRGB and Display P3 content should be converted to linear blending space using power 2.2 curve in this case or they would be displayed washed-out.

So the problem is: How to deal with HDR+SDR mix content in ACES? I saw sRGB content is using piecewise TRC to convert into ACEScg space.

1 Like
2

Hello, anyone could help?

3

@sdyer Could you help on this?

4

We encode the image with a pure 2.2 gamma function when viewing for sRGB. We skip the mismatch basically.
Then everything is just fine.

1 Like
5

Thanks for reply. So when using ACEScg for blending space, you are just using gamma 2.2 to convert sRGB content to ACEScg? Does it follow the specification?

6

Sorry I misunderstood your meaning of the word blending , you mean literally blending content?
I think all sorts of issues arise if you mix scene- and display-referred image states when blending images, or if you start to blend utilising inverser DRTs.
I would try to avoid that.

1 Like
7

But I think there should be some cases that need this? Such as applying a HUD on HDR scene?

8

At the point when we have a picture, it’s a linear wattage signal. There’s no need to complicate things. It is a linear interpolation across two wattages triplets, by degree of occlusion.

It is composited as two pictorial depictions, along the proportionate scale of the display medium. The two pictures are formed, and should be composited as is. If one wanted to make a HUD lower emission, it is no different to the basics of composition that have existed since the advent of the alpha over.