Sean Kim Sean-Kim
  • Joined on 2023-12-14
Sean Kim approved blender/blender#123447 2024-07-01 20:44:37 +02:00
Sculpt: Initial data oriented refactor for grab brush

Re-tested. No crash and behaving as expected

Sean Kim commented on issue blender/blender#124002 2024-07-01 20:40:07 +02:00
Sculpt: Undo issues when using multires

Bug doesn't show up in 4.2

Sean Kim commented on issue blender/blender#124002 2024-07-01 20:04:12 +02:00
Sculpt: Undo issues when using multires

Can confirm; actually unable to undo the mask at all in current main

Sean Kim commented on issue blender/blender#124002 2024-07-01 19:58:58 +02:00
Sculpt: Undo issues when using multires

Currently in a debug build for current main, this fails on the following assert after attempting to undo the mask.

`BLI_assert failed: source/blender/editors/sculpt_paint/sculpt_undo.cc:920,…

Sean Kim commented on pull request blender/blender#123447 2024-07-01 19:47:50 +02:00
Sculpt: Initial data oriented refactor for grab brush

Oh, my mistake, I completely looked over the fact this wasn't inside the factor loop

Sean Kim commented on pull request blender/blender#123447 2024-07-01 19:45:11 +02:00
Sculpt: Initial data oriented refactor for grab brush

I'm fine with that, but I think we should also do a cleanup commit of the other usages of it in the recent sculpt files for consistency's sake and so we don't rehash this conversation in the future

Sean Kim suggested changes for blender/blender#123447 2024-07-01 19:31:38 +02:00
Sculpt: Initial data oriented refactor for grab brush

Testing results:

Sean Kim commented on pull request blender/blender#123447 2024-07-01 19:12:05 +02:00
Sculpt: Initial data oriented refactor for grab brush

@mod_moder Is the reaction because of the mix of MutableSpan + const? I agree it reads weird, but for better or worse it's fairly common in the newly refactored code

Sean Kim commented on pull request blender/blender#123447 2024-07-01 19:00:20 +02:00
Sculpt: Initial data oriented refactor for grab brush

Haven't done validation of the brushes just yet - do you have any performance changes measured for this brush?

Sean Kim commented on pull request blender/blender#123447 2024-07-01 19:00:19 +02:00
Sculpt: Initial data oriented refactor for grab brush

Why do we pass Object in here if it's unused? Do you plan on making use of this in a future refactor?

Sean Kim commented on pull request blender/blender#123447 2024-07-01 19:00:18 +02:00
Sculpt: Initial data oriented refactor for grab brush

nit: Move the function docstring to the header

Sean Kim commented on pull request blender/blender#123447 2024-07-01 19:00:17 +02:00
Sculpt: Initial data oriented refactor for grab brush

BLI_assert(normals.size() == factors.size())

Sean Kim commented on pull request blender/blender#123447 2024-07-01 19:00:16 +02:00
Sculpt: Initial data oriented refactor for grab brush

Should this and orig_normals be part of TLS or otherwise initialized outside of the parallel loop?

Sean Kim commented on pull request blender/blender#123447 2024-07-01 19:00:15 +02:00
Sculpt: Initial data oriented refactor for grab brush

Do we support const here?

Sean Kim commented on pull request blender/blender#123447 2024-07-01 19:00:14 +02:00
Sculpt: Initial data oriented refactor for grab brush

We normally do BLI_NOINLINE before static void right?

Sean Kim commented on pull request blender/blender#123447 2024-07-01 19:00:13 +02:00
Sculpt: Initial data oriented refactor for grab brush

const

Sean Kim commented on pull request blender/blender#123447 2024-07-01 19:00:12 +02:00
Sculpt: Initial data oriented refactor for grab brush

Is it worth passing in a normalized version to this function from somewhere to avoid normalizing on every vert? It seems like a bit of a waste to explicitly make it non-normalized on line 228 only to revert it here.

Sean Kim closed pull request blender/blender#123518 2024-07-01 16:58:00 +02:00
Sculpt: Fix clay brush sample factor application
Sean Kim commented on pull request blender/blender#123518 2024-07-01 16:57:57 +02:00
Sculpt: Fix clay brush sample factor application

Thanks @DanielBystedt for the detailed feedback! I'll close this PR then since this new behavior isn't desirable.

Sean Kim pushed to main at Sean-Kim/.profile 2024-06-30 00:05:40 +02:00
3a9ae194ba Update reports/2024.md