That’s fantastic news! I remember being pretty bummed by how slow the boolean tools in Sculpt Mode were when they were first implemented, which made me not use them much. If the new boolean algorithm is 500 times faster than the current method, then I am very much looking forward to it! 
4 Likes
I am personally trying to figure out how it could even be possible to have a proven 500X speedup without any downgrade in quality, but I hope I am proven wrong on this.
Other sculpting programs have precise enough booleans that are several times faster than what we have in Blender, so it is about implementing the right type of algorithm that is efficient with high density models. The videos showing off the algorithm in Nomad look to me to be pretty darn solid in terms of performance. I doubt it will be as fast as in ZBrush, but it will be a substantial upgrade nonetheless.
4 Likes
There are downgrades. Most notably it doesn’t work with non-manifold meshes, so no planes or anything with holes. And there will be some other cases. But the thing is since it’s a choice, and will be alongside fast, exact (and in the future amber) user will have an option to choose what is best fit for every given context
7 Likes
Folks, anyone remind me if dyntopo not considering masked areas when subdividing or collapsing is a known limitation or something I just caught in the wild
I think it’s expected behavior, or at least that’s how I usually work when using dyntopo hehe; masking the parts I don’t want to mess up while adding more detail or simplifying other areas of the sculpt. It’s pretty handy IMO
2 Likes
Only subdividing and collapsing, where brush stroke effect is localized, is the principle of dyntopo.
According geometry, its remeshing can go further than brush radius to avoid ugly topology.
Mask attribute values lower than 1, are allowing subdividing and collapsing.
So, there can be geometry subdividing and collapsing at borders of a smoothed mask.
But that is making perfect sense ; that a mask at 1, supposed to annihilate brush effect totally, also avoids localized dyntopo remeshing.
That is the goal of Simplify Brush to subdivide and collapse without changing shape of mesh.
In fact, the limitation is rather that Face Sets automasking is not working, although Face Sets are preserved, now.
@julperado @zeauro thanks ! I completely messed up my question, I meant the opposite !
Look, this is a finely subdivided mesh that I partially masked, then did a big brush stroke with a much smaller resolution, and it definitely affects the masked area.
I think what’s happening is,… sculpt mode prevents the user from starting a stroke over a masked area, but if I brush over an unmasked area and the brush size is big enough that it overlaps the masked area next to it, then that masked area is still affected by subdivide/collapse. Mask values are definitely 1 here, I used a brush with constant falloff, but no matter what I do the outer border of the masked area is always affected.
4 Likes
Oh ok, yes that can be annoying, but I think it’s because the mask value isn’t actually fully 1 even if you have a constant falloff and no strenght pressure, I kinda got used to just press A and select Sharpen Mask and/or Increase Contrast a couple of times after I paint the mask, and then it works as expected
3 Likes
Thanks, it works ! the border is respected after using sharpen mask. 
Now, this means the constant falloff in brush settings isn’t actually constant. Or something else is at play. Anyway, this unlocked my progress. Thanks a lot !
5 Likes
MIght it be antialias at work?
Yes. In last dev meeting of Sculpt mode, they explained that they have to wrap up dyntopo refactor, focus on Brush code Refactor, fix multires bugs, think about redesigning multires for long term and work on texture painting project, too.
So, unless that is a project used to test brush code refactor to allow brush nodes, for later ; I doubt they will accept such proposal, now.
Anyways, that is too big to be ready to integrate, at the end of GSOC.
If it is accepted, that would not become a reality in master, before several years.
That would just end up as a fifth experimental branch about sculpt mode features.
At this point, a proposal about picking a feature present in previous experimental branch and making it rock solid would be more convenient.
These branches have a lot of cool stuff, absent of master.
- Brushes : Fairing brush, Scene Project brush, Sharpen Color brush, Symmetrize brush
- Lasso Project
- Auto Face Set creation
- Custom radius and projection for auto-smooth
- Custom Pressure Curves
- Roll mapping of brush texture
Because those features are bugged, not working or rashing experimental builds or nobody was able to split them and review them : all of that is stagnating in experimental branches and they are starting to be forgotten.
But if they were working and rock solid : they would be welcomed by users.
There are more than a dozen of potential ideas about sculpt mode, less ambitious than brush nodes : that could be the object of a GSOC.
2 Likes
It does seem precocious to me since sculpt mode is awaiting large architectural changes…
Sculpting something in the new Brush Assets Project branch. I forgot how to turn off the cursor following the mesh normals. Who can help me out? Please don’t tell me it isn’t possible. 
I remember that 2001’s projects for knife tool and curve edit were pretty successful with multiple actual merges. I would count shader node preview project too since it was actually merged even it’s only available as an experimental feature.
Also I think some projects (3D text usability improvements, many light sampling, and UV editor improvements, some VSE projects …) were later used as base or references to the actual commits by other developers. So some of them kind of worked as R&D projects.
UV vert slide project from last year failed, but devs implemented it from almost scratch for 4.2. So at least it gives devs incentive for finishing things.
And didnt Hans and Brecht started at GSoC? I think last 5 years skew the view of GSoC, but before that projects were often succesful. Its interesting tho, did Blender become awful at judging people and accepts wrong crowd, or are standards too high since 2.8?
2 Likes
I think? this is what you are looking for:
1 Like
Nope lol that’s the falloff of brush strength
2 Likes