I wonder if someone used to work at Pixologic making nice features that never got released and now they’re sending feature suggestions directly to Pablo as a way to mess with their former employer.
2 Likes
Better yet, they should make quality code pull requests and Pixologic would never find out.
Blockquote
- Joe Eagar finished his work in the dyntopo next step is clean up
- The patch still needs to be split for review
- There will likely still need to be adjustments
- Pablo will get an update on that
- Pablo will write a document with limitations of multi-res
- A discussions can be scheduled afterwards
- EEVEE sculpting for 3.0
- Dalai suggested Pablo to create a new branch for sculpting to handle UV and tangent data
- Then the draw manager team can work on the API to populate only the required data into the batches.
9 Likes
I also hope that the vertex paint or texture paint improvements are on the plan as well. Polishing the sculpting workflow is great but after that texturing is the next step. But EEVEE sculpting is pretty nice.
6 Likes
That is some deep conspiracy right there.
1 Like
I’m confused. Pablo already said that Multires had some issues. He even suggested to rewrite it from scratch as a possibility. Sergey insisted in fixing it. But… now multires kind of works but looks like it still has its limitations and Pablo is asking again for a proper solution.
Am I wrong?
Am I missing something?
What is love? Baby don’t hurt me, don’t hurt me, no more?
4 Likes
" Tangent space sculpting layers: This is a feature that even specialized software doesn’t support and which is really important for retouching VFX animation. By having multiple CD_MDISP datalayer, it should be possible to tweak the influence of each one of them on the final displacement. As these layers are in tangent space, this supports any transformation on the base mesh without applying the layers displacement in random directions."
This would be really cool to have. Not sure what shotsculpting has to do with multires or how it would possibly work with mesh sequence caches.
I have to agree with Brecht here - “I don’t think there is much point changing Multires to be aimed at animation and rendering.”
Multires aside, Tangent space sculpting layers would be huge, but so would be anything in that department as shotsculpting inside Blender is kind of impossible at the moment.
5 Likes
For Multires, I imagine the workflow to be like:
- The user creates a subdivision based sculpt with a dedicated subdivision based sculpting system (not Multires).
- When done, the user adds a Multires modifier to the high poly version and uses rebuild subdivisions. This will convert the sculpt to a base mesh + displacement. It is the same concept as baking a displacement texture from the high poly mesh into level 0, but slightly more convenient as it does not require UVs and geometry will look exactly the same as the sculpt.
- After having the new low res based mesh, the user can rig and animate the mesh with displacement.
This sounds like a pretty good workflow to me, make the subdiv sculpting system part of the mesh data as in Zbrush or Mudbox, and separate it from Multires, leaving the modifier to be used only with animation in mind.
I really hope they go this route 
9 Likes
Is there any way to approach this? I mean, sculpting already is on a much better stage than two years ago. Maybe one year dedicated to painting wouldn’t hurt and it should improve a lot the curren workflow.
1 Like
I would sum it up a little bit, differently.
Pablo said that Multires was too much broken to have a sculpting workflow making sense.
It was urgent to fix it.
They did a first brainstorming.
Result was : " we will quickly fix what we have and retake the discussion about a better solution, later."
Now, it looks like they are agreeing ; that current multires system can be mainly maintained as is and optimized to handle animation and rendering, and a new additional one with other benefits and limitations could be created to handle sculpting practice.
In other words, it seems to be impossible to have one multres system handling all desired goals. Let’s have two.
7 Likes
If anyone knows both Mixer&Painter then with everyone different opinion it is hard to tell but seems that Mixer is easy to iterate material mixing and Painter is very good to get material act right and smart materials can be controlled and created very well…with hundreds of other features too long to describe. But as far i have looked then Mixer+Substance Painter UI would be best mix in upcoming Material Paint. Blender bonus is that it could be little bit a Substance Designer too thanks to nodes.
I do not talk about sculpt itself which is now very good already even for beginner like myself but i personally focus on hard surface stuff(with realtime render rules) and sometimes sculpting is used for normal generation (currently hipoly and lowpoly with Substance Painter as Blender baker is pain)
1 Like
Completely agree. That’s what I was thinking when they started tackling it : why not make a separate “multires object”, since it has such specific needs ? data structures, layers, tools… it being a modifier on regular geometry feels really tacked-on.
1 Like
I still listen to that song every once in a while.
4 Likes
Yes, it feels too weird and limited for a pure multilevel sculpting system, but I can see its use for animation and mixing with other modifiers. I don’t think the Blender devs planned to have a sculpting system rivaling the big boys at the beginning, so the design must have been thought out to be useful in a “generalist” workflow, not like a separate, independent thing.
The multires modifier as it is now has proven very useful for me to quickly add details to models without having to deal with UV’s and displacement textures, and making changes or variations on the same model is also pretty fast. Clients love it when you can deliver changes quickly 
Hopefully, we’ll end up with both systems, that would be an amazing boost to Blender.
I don’t get it. What’s the difference between a sculpting layer system and the displacement modifier? Code should be already there… 
Why keep MR if there is already another SubD sculpting in place? I do not see the point of adding extra overhead to get some functionality in place, given MR is still prone to major pipeline issues, unless they will code it from scratch. Plus everyone agrees that MR code is hard to work with.
The displacement modifier needs texture inputs to control the displacement, while the Multires modifier, as we all know, allows to sculpt directly on the mesh. That’s extremely helpful for adding extra detail to animated models for example.
Well I don’t know much about the state of the code, but I do know that when the multires is used only as a way to add detail to animated/deformed meshes in combination with other modifiers there’s no issues or bugs.
The issues we all have been complaining about Multires for years come from using the modifier thinking of it as Blender’s attempt at mimicking Zbrush, which is not, and I don’t think it was designed to be something like that in the first place. Multires works just fine as long as you don’t change the basemesh, well minor changes shouldn’t be a problem, but in general that’s how it works.
Problem is, many Blender users including myself always tried to push Multires way beyond its capabilities.
So now what Pablo proposes is to keep multires for the purpose it was designed for originally and design a proper subdiv sculpting system, that works directly on the mesh (like Zbrush). Both would have its place and should work together pretty good.
Having the ability to transfer all the details and sculpting data from a highres model (new subdiv sculpt system) to a rigged lowres model with the modifier (multires) means you won’t have to bake displacement maps, given you’re working on a Blender based pipeline of course. Also, from a rendering point of view, rendering a mesh with multires uses a bit less memory than a displacement map, so that’s also a plus.
7 Likes
I agree. I purchased Retopoflow and used it for a project a while back. While the tools/visualization were nice, the hotkeys/environment were different and kind of felt like I was using another software. I’d much prefer an integrated set of tools that are perfectly native to Blender and work in harmony with the existing modeling tools.
5 Likes
Well and for me who likes auto retopo, i would like to have a better solution than the quad remesher (which is not that good).
4 Likes
I assume you mean Blender’s Quad(riflow) remesher is not that good? To my experience the Quad Remesher add-on ( = ZRemesher ) is pretty satisfactory.
2 Likes