Hi!

Maybe I'm going off-topic and I'm probably a little bit biased...

I think the main problem in the development process of some areas of Blender is feature incoherence. I feel like some times there are discussions, design proposals and commits implementing new advanced features on areas where the most basic functionality for an artist is still not right. The sculpt/paint module and the tool system are heavily affected by this.

This is a tricky problem to solve. We can't really expect artists to make useful bug reports about brushes or tools behaving incorrectly. For most of them, it would be nearly impossible to formulate the problem pointing at the exact technical detail for a developer to fix it. These kinds of issues are really obvious for an artist who has been working with these tools for years, but developers may think that the tool is working at it should. This communication problem often ends in misleading bug reports, proposals and discussions going nowhere while trying to fix the wrong thing.

The thing is: in some cases, those issues can be fixed by changing just a few lines, so we can expect artists with a coding background to fix them. But there are a lot of cases where a whole redesign of an area is required just to fix one those issues (that is essentially the problem of the sculpt branch). Some users are creating non-commercial addons to bypass these limitations by recreating entire systems through the Python API instead of contributing it to Blender, probably because they consider that a task for a core developer.

I think we should be extra careful about this when doing big refactors or creating new modules. In my opinion, the core team should focus more on designing the new systems in a way that makes the life of new contributors easier (who can probably take care of this) instead of making a final product with a planned set of features. We should put a little more effort to make sure the foundations are right from and artist perspective

Currently, some of those issues so bad that, in my opinion, they should not be considered as a feature request. They should probably be handled as a high priority bug, blocking a release. They make the experience of some areas much worse than Blender crashing every 10 minutes. This ends up hurting the opinion artists have about Blender and all the technical work that is behind it.

As an example, the Grease Pencil team spent a lot of time designing and tweaking the behavior of each brush, and that clearly shows. The result is so impressive that it almost feels like a pixel-based drawing software. An artist is not going to care about modifiers, effects or animation features if they feel that the brush engine is broken just by doing one brush stroke.

If we only focus on long term interesting technical projects and bug fixing, Blender will be able to paint PBR materials across multiple UDIMs, while using the current brush engine on a 2D view that does not rotate or flip. In my opinion, we should not touch a single line of the 3D texture projection code until the 2D view and the default brushes are absolutely perfect to a level they can be used for serious illustration work. Probably the sweet spot is in the middle of these two points of view.