Colormanagement: Skip view transform for byte images
AbandonedPublic
Actions

Authored by Richard Antalik (ISS) on May 3 2021, 11:33 PM.

Details

Reviewers

Brecht Van Lommel (brecht)
Jeroen Bakker (jbakker)

Maniphest Tasks

T86622: View Transform -> Filmic adds 333% to VSE render time and miscolors the result.

Summary

In some cases, users start video editing work from General template,
where Filmic view transform is chosen by default. Currently, this causes
issue when working with sRGB footage, which are not properly decoded
into linear working space and final render looks miscolored.

This commit disables any view transform on byte renders, as it is not
expected to be performed on such images. View transform will work as
expected on float images. This change should affect only VSE.

This solution is not really correct way to resolve this problem and
should be considered as temporary.

Diff Detail

Repository

rB Blender

Branch

T86622 (branched from master)

Build Status

Buildable 14351
Build 14351: arc lint + arc unit

Event Timeline

Richard Antalik (ISS) requested review of this revision.May 3 2021, 11:33 PM

Richard Antalik (ISS) created this revision.

Harbormaster completed remote builds in B14351: Diff 36742.May 3 2021, 11:33 PM

I thought we agreed to change the view transform to the default when adding the first strip to the sequencer in some cases, similar to the way render resolution is changed to match the footage. Did you intentionally go with a different solution, or was there a misunderstanding?

This patch I think is too unpredictable, it means the view transform settings have no impact depending on the strips in the sequencer, without any communication in the UI. And worse, this can actually change from frame to frame.

This revision now requires changes to proceed.May 4 2021, 1:16 AM

Also, IMB_colormanagement_display_get_default_view_transform_name() should be used instead of "Standard".

In D11149#284857, @Brecht Van Lommel (brecht) wrote:

I thought we agreed to change the view transform to the default when adding the first strip to the sequencer in some cases, similar to the way render resolution is changed to match the footage. Did you intentionally go with a different solution, or was there a misunderstanding?

From discussion in chat I thought this would be more propper way of "So then don't do filmic transform on 8-bit image would be my solution if that would be acceptable"

Though I would agree, that this could lead to seemingly unpredictable results. Especially when part of timeline can be float and part can be byte, I haven't realized that.

Ok so I will do view transform setup based on added strip.

Closing this patch, as the solution is too bad. D11151 is followup