I stumbled on this thread today: https://forum.unity.com/threads/scr...p-should-be-much-easier-to-work-with.1131433/ And I think points being made there are still valid to this day In the initial proposal on SRP dated like ~2017 or earlier, you guys did already had in mind idea that render pipeline doesn't even need to know about cameras. But then something seem to have gone wrong. Currently in SRP we either do things Unity way or completely custom rendering way, it seems there's no middle ground. There are still APIs like ComputeDirectionalShadowMatricies which are used in HDRP/URP and comments near them say that "we should move this logic to C#". To summarize, I'd really like some changes to be made in order to make SRP less of a black box with less coupled APIs. First of all, maybe do something about CullingResults struct I wish SRP followed principles stated by Casey Muratori as far as in 2004 in his Reusable Components talk.