What is the most bat-guano-insane thing in Blender that you can't believe they haven't fixed yet?

[zeauro]

It is simply not finished.
UDIM work has been stopped by work on new painting brush abilities and project of new paint mode.
Its developer worked on V2 of principled shader and bugfixes of Cycles.

There is no support of animated image sequences as UDIM tiles, yet.

@Romanji

Why do I have to mess with creating UDIM tiles in the UV editor?

It depends on inter-exchange format.
You have to recreate tile for obj, fbx. You don’t have to, for Alembic or USD.
(but in that case, you have to recreate material)

@Romanji

Why does it take 14 seconds on my 8Core CPU to create one, even if I don’t fill it with pixels - what kind of bullshit is this?

No idea. You probably should search to see if there is an existing bugreport.

@Romanji

why does Blender refuse to load certain UDIMs

Since 3.1, it supports virtual filenames. But only 2 substitution tokens are supported.

UDIM: Support virtual filenames · 180b66ae8a

[Romanji]

Thanks for the information, I got it fixed after messing around with it, but I am not even sure how…all in all - 45+ minutes wasted for nothing.
I guess exporting the asset to USD and importing it again will prevent this shit from happening again.

[Harley]

@Harley

On the Windows platform the placement and sizing of multiple (non-temporary) windows among multiple monitors should be working well. So let me know if they aren’t.

As in you should be able to have any arrangement of monitors (horizontal, vertical, a block, etc) and put multiple Blender windows (almost) anywhere you wish and they should be saved and restored correctly. “Save Startup File” and Blender will start that way, with multiple windows all over the place. It should even work correctly when the monitors differ in scale and resolution.

I say “almost” anywhere in that it won’t (currently) allow a window to be restored in a position that straddles multiple monitors at once. I should look at that one day. I also said “non-temporary” in that the positions and sizes of temporary windows (Preferences, File Browser, Render) aren’t saved and restored the same way.

Just quoting myself above for context. For 3.6, on the Windows platform, we’ll be able to work with windows that straddle multiple monitors. https://developer.blender.org/D17048

[joseph (J)]

Ooh… this is extremely exciting to hear, I’ve been wanting a good way to span a node graph across two monitors for a long time

[thinsoldier]

I’m red green colorblind. Don’t know which diagram is the one that won’t work.

[joseph (J)]

The two on the left will work, the one on the right won’t

[Harley]

It will allow a window to be created at any size and position as long as each of its four corners are within the working area of any active monitor.

Obviously even today you can stretch a blender window over multiple monitors in any way you wish. But if you saved the layout with the blend, or saved as default with “save startup file”, when those windows are loaded they are currently resized and moved to fit within the single nearest monitor. With the patch it will allow window creation that spans monitors.

I have updated the image and the text on that patch to be more colorblind-friendly. Sorry about that.

[AlphaChannel (Dr Salty Lout)]

@thinsoldier

I’m red green colorblind.

Instant eyedropper could help you?

a really handy little app @joseph recomended.

[Panther_Modern]

How Blender treats Vertex Groups as Float values, but the node editor tells you it expects Boolean, and then forces you to use a Float value node to try and find the correct value to address your Vertex Group.

Absolutely insane.

If you have a named Vertex Group, you should be able to address it by name, and not have to scrub through a range of values to try and find your vertices!

Completely mental.

[Okidoki]

With Geometry Nodes… ? …but…

…you can… ?!

[Panther_Modern]

See this thread I posted a while back for an example of the behavior:

Geometry Nodes - Extrude Mesh Problem

Hello all, Thanks in advance for any insight you might be able to provide here. I’m trying to use Geometry Nodes to procedurally build parts of a spaceship model, and I’m running into a problem where a Vertex Group selection does not seem to be honored. First, the working part: I have an 8-sided cylinder which I am using Extrude Mesh + a Mirror modifier, and Geometry Proximity to dynamically build this structure: [image] The extrusion is using a Vertex Group which is then extruded on the …

[Okidoki]

Ah yes now i understand the first part of this a bit better… well… as far as i understand this:
Every vertex is weighted individually in/for every vertex group… but the usual (boolean) interpretation is: if the value is zero then it does not “belong to” or “is in” the group…
(…it’s just one possibility to implement this…)

But in the other thread i didn’t understand why a greater than is need… in my example (now) in 3.3.2 a threshold of:

• t=0: selects everything
• 0 < t <1: selects the vertices in the group (after i understood ) according to the weight
• t=1: selects nothing

So the selection is made on a boolean basis but the values itself represent the weight…

Maybe (if you want) further discussion about this should be done in the original or a new thread ?

The named attributes are even in the 3.0 docu… https://docs.blender.org/manual/en/3.0/modeling/geometry_nodes/attributes_reference.html
but i don’t remember/found yet where the toggle i showed in the screenshot is mentioned or if i just discovered it while looking and was mouse pointer moving in the GUI…

[AlphaChannel (Dr Salty Lout)]

@Panther_Modern

How Blender treats Vertex Groups as Float values, but the node editor tells you it expects Boolean, and then forces you to use a Float value node to try and find the correct value to address your Vertex Group.

Absolutely insane.

I disagree, its great, means you can do things with weights

it’s more the boolean selection will be triggered by ANY value.

[Panther_Modern]

…This is… an insane and unintuitive way to handle it vs how Maya handles it. Ugh. I love Blender, but, honestly, I hate Blender sometimes because of how these things are handled. Stuff like this is one of the reasons Blender isn’t having any meaningful marketshare of the Hollywood production space. -_-

[Okidoki]

Blender is not aiming to/on the market… and everybody is free to make a fork… see for example BforArtists (or something like that… yeah… thsi one)…

Home - Bforartists

they use different icons workflows etc…

[Panther_Modern]

This list of differences is the exact type of thing I’m talking about.

The Differences to Blender - Bforartists

That Blender has not attempted to make itself more “industry standard” is one of the big reasons it has such a small percentage of adoption and marketshare vs something like Maya:

All major 3D packages provide better tooling than Blender does.

This post isn’t me hating on Blender - like I said before, I love Blender: for a piece of open source software, it’s amazing - but let’s be realistic: Blender’s shortcomings have a lot to do with the fact that most of the people working on Blender are programmers and not artists.

The vertex grouping / vertex weighting thing is a prime example of this:

You should be able to address a vertex group as a static group - AND - paint those vertices with weights - and optionally address them that way - but treating membership as a weight and then interpolating it as shown in the thread I linked is just…crazy counter-intuitive to the way other packages like Maya handle this. It’s just absolute madness - and it’s not my imagination either: I sent that thread to a friend of mine who is the lead technical artist at a major studio, using Maya and Houdini, and he was flabbergasted that Blender handled it that way.

[thinsoldier]

@Panther_Modern

Blender’s shortcomings have a lot to do with the fact that most of the people working on Blender are programmers and not artists.

Most of the time when artists need programmers, they hire them. I’d expect 99% of the programmers freely contributing to Blender to be programmers before artists. What would solve this problem is more money to hire more paid sheep dogs to shepherd a bigger herd of contributors, both artists and UI designers and programmers. But these things take time. I would prefer a hard stop on all new features until several hundred outstanding bugs and performance issues are fixed 100%. But when people come along with undeniably cool stuff and are willing to work for free until it’s usable (but years away from production ready) you kind of have to fly the plane while installing the windows.

[Laserschwert]

Just noticed that the flipped camera bug on FBX imports is STILL not fixed. Is it 10 years already?

[nickberckley (Nika Kutsniashvili)]

@Panther_Modern

Blender’s shortcomings have a lot to do with the fact that most of the people working on Blender are programmers and not artists.

I’ve thought about this too. The amount of attention Geometry Nodes are getting from developers is amazing considering just how bad painting is in Blender. Texture painting is at this point unusable simply. You can’t paint and thats the end. I know they’re working on adapting painting to sculpt mode code, but the pace is very slow.

Get even 1 artist in BF decision-making position and they’ll say painting should be the priority, every other module could use help and improvement, but painting just needs to exist its so behind, but because its just programmers they choose to pay attention to whatever is cool and interesting to them. Unsupervised nerds I call them, lovingly

Edit: I’ve also noticed on developer forum the how different Blender Studio people view development from developers. Studio people mostly suggest improvements and features they know from their experience to be needed for artists, but getting developers attention is very hard with those issues.

[zeauro]

That are different tasks.
Painting workflow needs improvements of UI, mastering of 3D View, mastering of images, being familiar with assets and libraries, being familiar with tools.
The target is to recreate the whole painting mode.
That is requiring experimented developers.

Geometry Nodes have been, rethought several times, before their introduction in 2.92.
They have been refactored in 3.0 series. But basically, when design is ready, the work consists in converting code already existing in modeling modes into nodes.
That is a massive task but more accessible to new devs, less experimented, and volunteers.
On the other hand, that is an ubiquitous domain and experts of a module can contribute to a node relative to their expertise.

That gives the impression that there is a different treatment. When, in fact, that are two tasks at different states of their development.

In 2.8 initial design proposal, vertex paint should have been improved, there was a gsoc about it, same year of 2.8 codequest. But quickly after, 2.80 release, Pablo Dobarro came in with an ambitious design proposal about painting in sculpt mode. The proposal was great but incomplete on many points that were very theoretical. Pablo did in sculpt mode the proof of concept. And other devs finalized what became Sculpt Color Attributes painting, released in 3.2.

3.2 — blender.org

What is currently in master, as experimental feature, is a canvas popover allowing to paint images, in Sculpt mode. That is there since months. We could think that development of painting is frozen.
But, in Blender 3.4, Sculpt mode dev worked on new auto-masking and face sets. Those tools are as useful for Paint brush than for Sculpt brushes.
Workbench engine progressed, recently.
https://developer.blender.org/rBba982119cdcacfa8e603457face1d8ebc763ef75

In Texture mode, textured materials also use the material properties. (Previously, only non textured materials would)

I don’t know what this sentence means. I perceived no functional change, for the moment.
But what is important to understand is that painting of textures is related to progress of 3D View engine displaying them. So, other devs, than Sculpt mode dev, are working on it.
There are 5 branches of development about rewriting Worbench or texture painting, that had last commit done, during last week.
It was clearly announced during a dev meeting, that EEVEE next, Workbench Next, Vulkan, Viewport Compositor was stuff that will be land after 3.5.
There is also a branch that was started this month, about asset shelf, that should become new way to access assets in 3D View. Brush assets should use it.
As a consequence, paint modes should be redesigned after those changes, not before.

So, waiting for other modules to progress is a completely different situation, than progressing by assimilating what exists in other modules.
When developers will work on new UI of new paint mode, with a testable build, that will make noise. But that will be the end of a long road of things, not testable by users, during a long period.