[HDR] Use ExtendedDisplayP3 as the HDR backing store colorSpace on iOS #47313

shallawa · 2025-06-27T18:19:35Z

`c52a198`

[HDR] Use ExtendedDisplayP3 as the HDR backing store colorSpace on iOS
https://bugs.webkit.org/show_bug.cgi?id=295118
rdar://153919243

Reviewed by Mike Wyrzykowski and Simon Fraser.

For compositing HDR drawing on macOS, the DisplayColorSpace should be used to
avoid extra colorSpace conversion.

For compositing HDR drawing on iOS, using ExtendedDisplayP3 ensures any blending
operations happen in the same gamma space as the compositor.

For non compositing HDR drawing, ExtendedDisplayP3 will be used for everything
else on macOS and iOS.

* Source/WebCore/platform/graphics/ContentsFormat.cpp:
(WebCore::contentsFormatExtendedColorSpace):

Canonical link: https://commits.webkit.org/296814@main

2364e58

Misc	iOS, visionOS, tvOS & watchOS	macOS	Linux	Windows
✅ 🧪 style	✅ 🛠 ios	✅ 🛠 mac	✅ 🛠 wpe	✅ 🛠 win
✅ 🧪 bindings	✅ 🛠 ios-sim	✅ 🛠 mac-AS-debug	✅ 🧪 wpe-wk2	⏳ 🧪 win-tests
✅ 🧪 webkitperl	✅ 🧪 ios-wk2	✅ 🧪 api-mac	✅ 🧪 api-wpe
	✅ 🧪 ios-wk2-wpt	✅ 🧪 mac-wk1	✅ 🛠 wpe-cairo
	✅ 🧪 api-ios	✅ 🧪 mac-wk2	✅ 🛠 gtk
	✅ 🛠 vision	✅ 🧪 mac-AS-debug-wk2	✅ 🧪 gtk-wk2
	✅ 🛠 vision-sim	✅ 🧪 mac-wk2-stress	✅ 🧪 api-gtk
✅ 🛠 🧪 merge	✅ 🧪 vision-wk2	✅ 🧪 mac-intel-wk2	✅ 🛠 playstation
	✅ 🛠 tv	✅ 🛠 mac-safer-cpp
	✅ 🛠 tv-sim
	✅ 🛠 watch
	✅ 🛠 watch-sim

webkit-early-warning-system · 2025-06-27T18:19:42Z

EWS run on previous version of this PR (hash 340e1ff)

Misc	iOS, visionOS, tvOS & watchOS	macOS	Linux	Windows
✅ 🧪 style	✅ 🛠 ios	✅ 🛠 mac	✅ 🛠 wpe	⏳ 🛠 win
✅ 🧪 bindings	✅ 🛠 ios-sim	✅ 🛠 mac-AS-debug	~~🧪 wpe-wk2~~	⏳ 🧪 win-tests
✅ 🧪 webkitperl	~~🧪 ios-wk2~~	~~🧪 api-mac~~	~~🧪 api-wpe~~
	⏳ 🧪 ios-wk2-wpt	~~🧪 mac-wk1~~	✅ 🛠 wpe-cairo
	⏳ 🧪 api-ios	~~🧪 mac-wk2~~	⏳ 🛠 gtk
	✅ 🛠 vision	⏳ 🧪 mac-AS-debug-wk2	⏳ 🧪 gtk-wk2
	✅ 🛠 vision-sim	~~🧪 mac-wk2-stress~~	⏳ 🧪 api-gtk
	~~🧪 vision-wk2~~	~~🧪 mac-intel-wk2~~	⏳ 🛠 playstation
	✅ 🛠 tv	⏳ 🛠 mac-safer-cpp
	⏳ 🛠 tv-sim
	⏳ 🛠 watch
	✅ 🛠 watch-sim

webkit-early-warning-system · 2025-06-27T18:37:02Z

EWS run on previous version of this PR (hash 0125679)

Misc	iOS, visionOS, tvOS & watchOS	macOS	Linux	Windows
✅ 🧪 style	✅ 🛠 ios	✅ 🛠 mac	✅ 🛠 wpe	✅ 🛠 win
✅ 🧪 bindings	✅ 🛠 ios-sim	✅ 🛠 mac-AS-debug	✅ 🧪 wpe-wk2	⏳ 🧪 win-tests
✅ 🧪 webkitperl	✅ 🧪 ios-wk2	✅ 🧪 api-mac	✅ 🧪 api-wpe
	✅ 🧪 ios-wk2-wpt	✅ 🧪 mac-wk1	✅ 🛠 wpe-cairo
	✅ 🧪 api-ios	✅ 🧪 mac-wk2	✅ 🛠 gtk
	✅ 🛠 vision	✅ 🧪 mac-AS-debug-wk2	✅ 🧪 gtk-wk2
	✅ 🛠 vision-sim	✅ 🧪 mac-wk2-stress	✅ 🧪 api-gtk
	✅ 🧪 vision-wk2	✅ 🧪 mac-intel-wk2	❌ 🛠 playstation
	✅ 🛠 tv	✅ 🛠 mac-safer-cpp
	✅ 🛠 tv-sim
	✅ 🛠 watch
	✅ 🛠 watch-sim

weinig · 2025-06-27T18:37:52Z

Can you expand on how a color space affects tone mapping? Commonly, it is only the headroom that has affect as all changing from extended SRGB to extended P3 does is change the coordinate space.

shallawa · 2025-06-27T22:08:50Z

Can you expand on how a color space affects tone mapping? Commonly, it is only the headroom that has affect as all changing from extended SRGB to extended P3 does is change the coordinate space.

I really do not know the difference between the tone-mapping to SRGB versus the tone-mapping to P3 on iOS. This was recommended internally by macOS system frameworks teams. I think iOS might have tone-mapping calculations tuned for P3 better than it for SRGB.

smfr · 2025-06-27T22:50:20Z

Source/WebCore/platform/graphics/ContentsFormat.cpp

@@ -45,7 +42,11 @@ std::optional<DestinationColorSpace> contentsFormatExtendedColorSpace(ContentsFo
 #endif
 #if ENABLE(PIXEL_FORMAT_RGBA16F) && ENABLE(DESTINATION_COLOR_SPACE_EXTENDED_REC_2020)


If you stop using Rec2020 below it's' weird to still have this #if ...DESTINATION_COLOR_SPACE_EXTENDED_REC_2020

webkit-early-warning-system · 2025-06-27T23:10:04Z

EWS run on previous version of this PR (hash 08ec995)

Misc	iOS, visionOS, tvOS & watchOS	macOS	Linux	Windows
✅ 🧪 style	✅ 🛠 ios	✅ 🛠 mac	✅ 🛠 wpe	⏳ 🛠 win
✅ 🧪 bindings	✅ 🛠 ios-sim	✅ 🛠 mac-AS-debug	✅ 🧪 wpe-wk2	⏳ 🧪 win-tests
✅ 🧪 webkitperl	✅ 🧪 ios-wk2	✅ 🧪 api-mac	✅ 🧪 api-wpe
	✅ 🧪 ios-wk2-wpt	✅ 🧪 mac-wk1	✅ 🛠 wpe-cairo
	~~🧪 api-ios~~	✅ 🧪 mac-wk2	✅ 🛠 gtk
	✅ 🛠 vision	✅ 🧪 mac-AS-debug-wk2	✅ 🧪 gtk-wk2
	✅ 🛠 vision-sim	✅ 🧪 mac-wk2-stress	✅ 🧪 api-gtk
	✅ 🧪 vision-wk2	✅ 🧪 mac-intel-wk2	✅ 🛠 playstation
	✅ 🛠 tv	✅ 🛠 mac-safer-cpp
	✅ 🛠 tv-sim
	✅ 🛠 watch
	✅ 🛠 watch-sim

weinig · 2025-06-27T23:25:13Z

Given this can give different results, this feels like something we should be really clear on. I think before changing this kind of thing, getting more details on exactly what is different, and making sure those differences are tested, makes sense.

weinig · 2025-06-27T23:26:50Z

Also, if macOS and iOS are giving different results, that seems like a bad thing, no?

weinig · 2025-06-27T23:28:29Z

ExtendedDisplayP3 produces better tone-mapping to SDR on iOS. On iOS we have to
explicitly choose the backing store colorSpace.

The PR also changes things to use ExtendedSRGB on all other platforms, instead of ExtendedRec2020, so the commit message should state that.

webkit-early-warning-system · 2025-06-28T00:25:25Z

EWS run on current version of this PR (hash 2364e58)

Misc	iOS, visionOS, tvOS & watchOS	macOS	Linux	Windows
✅ 🧪 style	✅ 🛠 ios	✅ 🛠 mac	✅ 🛠 wpe	✅ 🛠 win
✅ 🧪 bindings	✅ 🛠 ios-sim	✅ 🛠 mac-AS-debug	✅ 🧪 wpe-wk2	⏳ 🧪 win-tests
✅ 🧪 webkitperl	✅ 🧪 ios-wk2	✅ 🧪 api-mac	✅ 🧪 api-wpe
	✅ 🧪 ios-wk2-wpt	✅ 🧪 mac-wk1	✅ 🛠 wpe-cairo
	✅ 🧪 api-ios	✅ 🧪 mac-wk2	✅ 🛠 gtk
	✅ 🛠 vision	✅ 🧪 mac-AS-debug-wk2	✅ 🧪 gtk-wk2
	✅ 🛠 vision-sim	✅ 🧪 mac-wk2-stress	✅ 🧪 api-gtk
✅ 🛠 🧪 merge	✅ 🧪 vision-wk2	✅ 🧪 mac-intel-wk2	✅ 🛠 playstation
	✅ 🛠 tv	✅ 🛠 mac-safer-cpp
	✅ 🛠 tv-sim
	✅ 🛠 watch
	✅ 🛠 watch-sim

shallawa · 2025-06-28T00:31:05Z

ExtendedDisplayP3 produces better tone-mapping to SDR on iOS. On iOS we have to
explicitly choose the backing store colorSpace.

The PR also changes things to use ExtendedSRGB on all other platforms, instead of ExtendedRec2020, so the commit message should state that.

I changed the PR to use ExtendedDisplayP3 for all non compositing scenarios on macOS and for everything on iOS. I think ExtendedDisplayP3 should be fine for all drawing except the compositing drawing on macOS.

shallawa · 2025-06-30T16:36:08Z

Also, if macOS and iOS are giving different results, that seems like a bad thing, no?

On macOS, we rely on one of two things to get the colorSpace of the backing stores:

NSScreen by calling [NSScreen mainScreen].colorSpace
NSView by calling [[m_view window] colorSpace]

NSScreen and NSView are available only on macOS. So on iOS we have to use a predefined colorSpace. This has been the behavior for many years.

I am not aware of a more robust way to calculate the colorSpace on iOS. Please let me know if there is a better way to do that.

shallawa · 2025-06-30T16:54:18Z

Given this can give different results, this feels like something we should be really clear on. I think before changing this kind of thing, getting more details on exactly what is different, and making sure those differences are tested, makes sense.

There is no real technical reason for choosing ExtendedRec2020 for HDR on iOS. I just tried it when I added the support for HDR and things seem to work. I used by eyes to compare the display on macOS and on iOS and I did not notice any significant difference.

The recommendation was to use an extended color for HDR backing store so I chose ExtendedRec2020. I even used it for both macOS and iOS at the beginning. Then I changed it on macOS to use the DisplayColorSpace. Then I changed again it to the CGColorSpaceCreateExtended of the DisplayColorSpace.

Recently we received reports about differences in the display between macOS and iOS when displaying gain-map images and when tone-mapping HDR images to SDR images.

mwyrzykowski · 2025-06-30T18:06:35Z

Can you expand on how a color space affects tone mapping? Commonly, it is only the headroom that has affect as all changing from extended SRGB to extended P3 does is change the coordinate space.

I really do not know the difference between the tone-mapping to SRGB versus the tone-mapping to P3 on iOS. This was recommended internally by macOS system frameworks teams. I think iOS might have tone-mapping calculations tuned for P3 better than it for SRGB.

System tonemapping is not applied for non-extended color spaces.

https://bugs.webkit.org/show_bug.cgi?id=295118 rdar://153919243 Reviewed by Mike Wyrzykowski and Simon Fraser. For compositing HDR drawing on macOS, the DisplayColorSpace should be used to avoid extra colorSpace conversion. For compositing HDR drawing on iOS, using ExtendedDisplayP3 ensures any blending operations happen in the same gamma space as the compositor. For non compositing HDR drawing, ExtendedDisplayP3 will be used for everything else on macOS and iOS. * Source/WebCore/platform/graphics/ContentsFormat.cpp: (WebCore::contentsFormatExtendedColorSpace): Canonical link: https://commits.webkit.org/296814@main

webkit-commit-queue · 2025-06-30T18:28:35Z

Committed 296814@main (c52a198): https://commits.webkit.org/296814@main

Reviewed commits have been landed. Closing PR #47313 and removing active labels.

shallawa self-assigned this Jun 27, 2025

shallawa added the Images For bugs in image handling. label Jun 27, 2025

shallawa force-pushed the eng/HDR-Use-ExtendedDisplayP3-as-the-HDR-backing-store-colorSpace-on-iOS branch from 340e1ff to 0125679 Compare June 27, 2025 18:36

webkit-ews-buildbot added the merging-blocked Applied to prevent a change from being merged label Jun 27, 2025

shallawa requested review from weinig, smfr and heycam June 27, 2025 22:08

smfr reviewed Jun 27, 2025

View reviewed changes

shallawa removed the merging-blocked Applied to prevent a change from being merged label Jun 27, 2025

shallawa force-pushed the eng/HDR-Use-ExtendedDisplayP3-as-the-HDR-backing-store-colorSpace-on-iOS branch from 0125679 to 08ec995 Compare June 27, 2025 23:09

shallawa force-pushed the eng/HDR-Use-ExtendedDisplayP3-as-the-HDR-backing-store-colorSpace-on-iOS branch from 08ec995 to 2364e58 Compare June 28, 2025 00:25

smfr approved these changes Jun 30, 2025

View reviewed changes

mwyrzykowski approved these changes Jun 30, 2025

View reviewed changes

shallawa added the merge-queue Applied to send a pull request to merge-queue label Jun 30, 2025

webkit-commit-queue force-pushed the eng/HDR-Use-ExtendedDisplayP3-as-the-HDR-backing-store-colorSpace-on-iOS branch from 2364e58 to 5fe83f4 Compare June 30, 2025 18:26

webkit-commit-queue force-pushed the eng/HDR-Use-ExtendedDisplayP3-as-the-HDR-backing-store-colorSpace-on-iOS branch from 5fe83f4 to c52a198 Compare June 30, 2025 18:28

webkit-commit-queue merged commit c52a198 into WebKit:main Jun 30, 2025

webkit-commit-queue removed the merge-queue Applied to send a pull request to merge-queue label Jun 30, 2025

		@@ -45,7 +42,11 @@ std::optional<DestinationColorSpace> contentsFormatExtendedColorSpace(ContentsFo
		#endif
		#if ENABLE(PIXEL_FORMAT_RGBA16F) && ENABLE(DESTINATION_COLOR_SPACE_EXTENDED_REC_2020)

[HDR] Use ExtendedDisplayP3 as the HDR backing store colorSpace on iOS #47313

[HDR] Use ExtendedDisplayP3 as the HDR backing store colorSpace on iOS #47313

Conversation

shallawa commented Jun 27, 2025 • edited by webkit-early-warning-system Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

c52a198

Uh oh!

webkit-early-warning-system commented Jun 27, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

webkit-early-warning-system commented Jun 27, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

weinig commented Jun 27, 2025

Uh oh!

shallawa commented Jun 27, 2025

Uh oh!

smfr Jun 27, 2025

Choose a reason for hiding this comment

Uh oh!

shallawa Jun 27, 2025

Choose a reason for hiding this comment

Uh oh!

webkit-early-warning-system commented Jun 27, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

weinig commented Jun 27, 2025

Uh oh!

weinig commented Jun 27, 2025

Uh oh!

weinig commented Jun 27, 2025

Uh oh!

webkit-early-warning-system commented Jun 28, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

shallawa commented Jun 28, 2025

Uh oh!

shallawa commented Jun 30, 2025

Uh oh!

shallawa commented Jun 30, 2025

Uh oh!

mwyrzykowski commented Jun 30, 2025

Uh oh!

webkit-commit-queue commented Jun 30, 2025

Uh oh!

Uh oh!

shallawa commented Jun 27, 2025 •

edited by webkit-early-warning-system

Loading

`c52a198`

webkit-early-warning-system commented Jun 27, 2025 •

edited

Loading

webkit-early-warning-system commented Jun 27, 2025 •

edited

Loading

webkit-early-warning-system commented Jun 27, 2025 •

edited

Loading

webkit-early-warning-system commented Jun 28, 2025 •

edited

Loading