INTEL_lights_sunsky: a procedural sun-sky model as environmental light source

[johguenther]

Procedural sun-sky models are a wide-spread and flexible tools to provide natural illumination to scenes and objects or commercial products, complementing point lights (KHR_lights_punctual) and image-based environment maps (#1956).

[johguenther]

Any feedback/discussion/review?

[BruceCherniak]

There has been no feedback on this since Jun. I support this extension, and it is critical to some of our customers. Am I able to approve the "review" as a vendor extension without further discussion?

[lexaknyazev]

To move this further, the PR needs JSON schemas for the extension objects.

[donmccurdy]

• /cc Showing status of extension PRs #1970

I feel it would be useful to publish some visibility into what to expect when opening a PR for a {vendor, multi-vendor, or KHR} extension. There should be nothing blocking the authoring vendor from using their own vendor extension, with or without review. I assume the review process exists to ensure quality and consistency of documents published under the Khronos repository, as well as to offer technical feedback (if desired by the vendor).

[johguenther]

To move this further, the PR needs JSON schemas for the extension objects.

JSON schemas are now included.

[BruceCherniak]

To move this further, the PR needs JSON schemas for the extension objects.

JSON schemas are now included.

Intel is still interested in merging this vendor extension.

[javagl]

I cannot say anything from the perspective of a possible implementation. From quickly skimming over the text and the schema, it looks "clean", but that's not a thorough review.

A few high-level comments:

• The role of the node that the light is attached to might have to be clarified. Specifically: How does the node transform affect the light? I remember that this always raised questions elsewhere - for punctual lights and other things (even cameras).
• I wonder whether something should be said about combining multiple lights. Maybe that's obvious or a non-issue, as in "Just perform the computations - the specification is not responsible for defining what 'makes sense', but only to offer all the options"...
• The text mentions some horizonExtension parameter. I wonder whether this should be part of the schema...?