• add docstrings

I tried to debug your code, but it seems that materialize is causing the crash even though the parameters seem correct from testing.

• rebase (fixing the previous merge)

I do experience the same issue on

• rebase (attribute refactor)

@Germano Cavalcante (mano-wii) Good move. Explaining code is sometimes difficult.

Probably a duplicate of T85752. Also don't confuse "Exclude from View Layer", "Hide". The Eye Icon is called "Hide". What you meant to write in your report was "Exclude from View Layer".

Thanks for the feedback, I have some questions though:

• requested changes

• cache clipping planes in custom

Not sure about the names of the functions and structs.

It does crash on 2.93.0 Beta, branch: master, commit date: 2021-04-21 20:15, hash: rB0f66dbea904f on Linux.

Can confirm.

Cannot reproduce on Linux. Neither with 2.92.0 nor with current 3.0.0. Tested restarting blender, loading the file and selecting, 6 times. Also tested the "Select Linked" command with no crash. Maybe it's Mac specific, someone needs to test this on Windows.

Cannot reproduce, neither in 2.92.0 nor in current master 3.0.0. Tested on Linux.
Is it consisstent? If it is consistent, somebody with windows should test, you can also try to make a video clearly showing it.

This patch is still up for review

• requested style changes
• align layout (more consistent with modifiers)
• fix memory leak on apply

Tested and confirmed with

I have done some ablation tests and reduced the file down to two booleans. You can see the result switch not only when reevaluating the modifier, but also when moving some of the vertices of the tube in that file.

I don't know if this was discussed anywhere else already (didn't find anything), but If you have all the outputs at once (I guess for performance if multiple are needed like mean and standard deviation) then it would be good to keep it at a minimum. Meaning, you should not have variance if there already is standard deviation, a simple square will give you the variance (no big deal for the user). Also meaning there should not be Range (which, for some users, could also imply a two value object like an interval) if there is Min and Max (again trivial to do as a user). For sum and median I think keeping both would be better, even though it is again trivial to do as a user.

@Hans Goudey (HooglyBoogly) can you give me a hint where to look for the code that can transfer all attributes at points from input to output? I had a hard time figuring this out.

So I just downloaded the build from https://blender.community/c/graphicall/djbbbc/ (Thanks NiCapp) to see what exactly is missing from this patch.

• requested changes

@Jacob Z (TooMuchFun) You are describing something completely different. Please open a new bug report with all the required information (you need to specify a working version for these kind of reports).

Oh sorry maybe I was a bit unclear, let me post some more images.
My plan was to make a convex hull around any position data passed to the node.

The problem is, that unsubdivide is currently executed multiple times one after the other, as many times as you specified in "iterations". So after one unsubdivision you get this criss-cross pattern, where there is really no way to not mess up your UV's and the next iteration has to work with that. So a "fix" here would look more like a full redesign of unsubdivision.

The log looks like its a gpu hang with the intel drivers much like in T80458. I think the issue was raised around the time of 2.82 or 2.83 release, so I think it roughly matches the experience there about 2.82 being more stable. I also remember using 2.82 on my laptop completely stable. I can't say though if it was the driver that was available then, or the blender version, which made it stable. Also, there are definitly multiple bugs at play here! The random crashes look the same as the crash that people are getting with that file in T80458, but it might be something different, as there is no obvious high gpu load on those random crashes, while the gpu hang with the file from T80458 is likely caused by a too big render call.

please check if this is a duplicate of T80458.

• add option to dna defaults

Thanks for the patch! This is exactly what I wanted to see in the weld modifier.

• change name to ui_scale

• rebase
• add error check for negative ranges on log scale sliders

• fixed wrong indices for more than one segment

@Campbell Barton (campbellbarton) I would like to get a review on this patch now.
I used find and replace commands for most of the names cleanup, so there should not be any errors.
I used copy and paste to get the code into the functions and again replace to change it to using the new context struct.
I also tested it on my (animated) test file with all previously known difficult cases and it didn't show any changes nor did it crash.

@Gaia Clary (gaiaclary) I just now noticed, this feature was removed.
I was actually using it in my projects (with my own hacky importer...).
Any chance on getting shininess and specular export (or some equivalent) back anytime soon?
Currently I don't know how to workaround this other than reverting this commit manually on a local build.

In T85752#1115246, @Philipp Oeser (lichtwerk) wrote:

you can also enable it again here! Very unexpected!

Why is this so unexpected?

In D4960#245812, @Alberto Velázquez (dcvertice) wrote:

Voxel remesher needs a way to block the slider from very low vixel size respect the object. Calculate the bounding box and that the slider won't change below some value like 1/500 of the size without input the text directly. Some for default value, that need to be a round number using the size of the object (1cm, 10cm, 1m, 10m, etc)

I did some more testing. Some of the errors I was seeing were due to the naive rebase, so thanks for the update.
I can see that your code for the edges is working good in a lot of cases.

I know this is WIP, but I wanted to mention, that this solution has some bigger flaws to tackle.
I tested it with bevel weight for example and it gives completely wrong results.
As this patch currently only aims at 2D extruded bevels (AFAICT), intersection fills are not working. But also in the 2D case it seems like sharp angles dont work?

can confirm with the reported file with:

I like the splitup of the functions.
I looked through the code and didn't catch any functional changes, so I am accepting this patch.
(I also compiled and tested it)

• add comment for axis and flip
• use branched logic with count_bits_i for readability

@Germano Cavalcante (mano-wii) I looked at the file for roll handling, but it seems to be all over the place. It's is used from the field val but sometimes saved in td->extra. I don't understand it.

Ok I though a bit more about the scewd spacemtx. For the general mirror operation I always (if possible) want to mirror the visual geometry in global space as that is the most intuitive and useful. Other types of mirrors can still be achieved by scaling.