[Custom Build] Dyntopo improvement: TopoRake


(Jeacom) #1

I am developing a feature for blender hopefully developers make it official.

Revision can be found here: https://developer.blender.org/D4189

Many of us wanna a dyntopo that works with quads, unfortunally Its hard due to the way its build. But I could take advantage of the subdivide/collapse behavior and implement a feature alignment stroke setting, that align triangles to form “quads”.

Custom Build for Windows x64:
https://drive.google.com/open?id=19z9f6pHn0Q_vqLKlZfM2dxvvO7dd_k_c



Result after cleaning mesh by dissolving verts connected to 4 edges and merging triangles:
ezgif-1-f803ea8841ee

I even made a character using this feature,
character toporake.blend (4.9 MB)


Dyntopo tests
(Ace Dragon) #2

It may not be possible to create quad-based dynamic topology because of possible performance roadblocks. However, this toporake feature could be a good step in allowing the use of the Quad to Tris operator to be a one-click solution for somewhat decent quad-based topology.

I’m all for it, anything to make even the raw topology a bit more usable for other steps such as texturing, shading, and posing.


(Hadriscus) #3

Hey,

I like this ! Seems to give nice sharply defined ridges at low resolution. I’m all for it !

Hadrien


(Jeacom) #4

That was the Idea, of course its not like retopology but will save time and polygons to get the model shaped out, with more sharpy strokes.


(TheRedWaxPolice) #5

Useful… :+1:


(burnin) #6

:smiley: Good looking!
Wishing. Wanting. Waiting…


(0rAngE) #7

As someone who is into sculpting, and uses blender as the primary sculpting tool - wow, this is the most impressive thing I have seen recently!!!

You gotta talk BF to add it into the master!!! We The Dyntopo Crew ™ gotta have this.

I’m intrigued, from your testing, what is the performance hit with this algorythm compared to default blender dyntopo?

I feel using topoRake one could get a much finer sculpt under lower dyntopo resolution.
Looking forward to try this out.

Hoping this eventually gets in!


(Jeacom) #8

Its slightly slower since the algoithm runs a brute force search for the best edges to align but I dont feel many performance issues with my computer, AMD A4 6300 (dual core) 3GHz, its an average computer and doesnt have a potent graphics card also.

and…

Yup!, thats the idea!
Lighter meshes, finer details!


(Kologe) #9

Wow, this looks great and very useful indeed! I’d be all for having it in officially.
Btw., are there any builds floating around anywhere (not sure I’d find the time 2 compile myself)?

Anyway, I have my fingers crossed and wish you success in getting it reviewed.

Greetings, Kologe


(Jeacom) #10

I’ve made a build for windows x64, Hope you like

https://drive.google.com/open?id=19z9f6pHn0Q_vqLKlZfM2dxvvO7dd_k_c

example .blend
https://drive.google.com/open?id=1Q9Cp81PVqe7Ly1RbCHFYE99rZmV4Su4v


(0rAngE) #11

Thanks on sharing the build!

Tried it out. Was just messing around placing some strokes side by side with Toporake at 0 and 1.
I’ll have to try this out on an actual sculpt.

First impression is, that yes the topology with Toporake at 1 is a lot more favorable then what we get with regular Dyntopo.
Testing it on a very low poly subdivided cube, I did not see any considerable slowdown.

i7 2600K (3.8 GHz)
GTX 1060 6GB

I also like how you’ve made the setting local - per brush, rather than a global Dyntopo setting.

This looks pretty amazing already! Keep us posted with your progress on the feature and the review as well! It would be amazing to have this officially in the buildbots!

I’ll probably come back drop a word after I try it on an actual sculpt.
Thanks again, much appreciated! :beers:


(FreeMind) #12

Any reason not to keep this Fully on, all the time?..


(Jeacom) #13

That’s because some brushes might benefit from it but not every brush depending on what you are doing.
An example is the Smooth brush, it can use Toporake but can also be used to destroy the tesselation.
another example is the snake hook brush, if you are using it like a grab tool to move whole parts of your sculpt. you might not want it to destroy the flow that was aready defined.


(RaphaelBarros) #14

It’s been some time since I’ve been excited about a feature. This seems really useful for hardsurface sculpting as well! Thank you so much!


(Jeacom) #15

I’ve just playing for a while to test the feature, ended sculpting a character:
the hair was so much fun!




(Ace Dragon) #16

Brecht’s review.
https://developer.blender.org/D4189#94858

The good news is that he sees the functionality as useful and that there are no major issues with the code (with most of them rather minor like spelling errors and code style). I do hope this means the patch gets committed quickly, as dyntopo is currently an area that does not have any active development.


(Marc Clint Dion) #17

That’s a beautiful tool. Many years ago, a very talented sculptor named Doris was asking for a feature like this. She called it “Flow Edge Here” as a name for the concept she proposed. Nobody was interested in building it back then, now here it is built and looking amazing.

If Doris and Michalis are still around, I’m sure they will both help spread the word.


(Thinking Polygons) #18

@Jeacom This looks great.
You should meet up with Pablo Dobarro and together “fix” the sculpting system in blender once and for all hahaha. That would be nice. :smiley:

Good job. :+1:


(Michael Knubben) #19

This looks amazing!
Any chance you’d consider creating a tool that only redirects topology, but leaves the shape intact?


(Jeacom) #20

actually any brush can do this. just leave the strenght at zero and increase toporake.