Texturing, Baking, Brush System improvements

I am very aware of the huge benefits a node system has over layers. But it’s just not practical for layering images the way you’d do it within most 2D Applications.
But I also agree with you my proposed solution of just using the material editor for texture painting is not a great idea. Mabye adding a new datatype of a layered image, that utilizes the material/compositing nodes backend for blending and stuff might be a better solution.

Also It’d really be beneficial to discuss this somehow in person.

(And trust me I’m also not happy with my own color pallete system atm and my posts are really not meant to be about providing perfect solutions but about discussing possibilities to make blender more usable for the needs of 2D-artists, that often deal with color painting…)


Regarding the nodes vs layers, honestly I think we need to simply look at what has made both designer and painter a success. Designer uses nodes to design substances, aka smart materials, masks…ect Painter, takes those smart materials designed with nodes, and lets you color in or paint on the assets with layers.

Layers are the way to go when actually painting and texturing an asset. Like in Modo, layers can also exist in the same way an object tree or outliner exist, if this makes it easier to implement. 3D Coat and Substance Painter use the Photoshop approach, which would probably require more work but also becomes instantly familiar and thus more functional to the end user.

Nodes -> designing the things you paint or apply to a mesh, but the layers are where those things exist in order to be easily managed.

ArmoryPaint for Blender probably has the right generalized idea in this regard, taking the best of both and combining them (which a lot of substance users wanted originally anyhow), node & paint. The workflow could probably be a lot better, and deeper but the combination seems to work well.

1 Like

Blender have powerful baking opportunities, imho. Possibilities to bake absolutely any type of texture map, also pack them into rgb channels. Big possibilities to fix texture map in texture piant mode. But with a high grid density and a lot of objects, baking becomes very long. Before baking, the blender has a delay and this is a big problem. In blender internal baking very fast, use all cpu cores and don’t have long delay, like on cycles. But internal dont have cage. If internal baking can have cage it will be big improvement.

1 Like

“lacking interface between texture painting and node editor (therefore lacking support for filters or procedural masks)” + layers system

yeah, thats all what we need

I did not want to discourage you to make a layer system. I was just warning you that connection between a Layering system and material node editor probably can not be absolute.
It does not mean that partial/relative one can not exist.

Currently, Slots panel have 2 modes :

  • Material Mode, listing all images used by current material
  • Single Image Mode, to allow to paint an image that is not used by material but that is used to influence effect a modifier or to tweak particles.

We can easily to replace actual material mode by several material modes.
An absolute material mode working as current one that could list all images in material.
And a layered material mode that would only support nodetrees made of a principled shader + mixRGB nodes, with an admitted known limitation of not being able to handle all nodetrees.

So workflow would be : when artist starts to paint to prototype the asset, he would roughly use this layering system.
When he wants to become more technical, to insert procedural effects, to decouple things into different shader nodes, to emphasize some things with math/curve nodes : he stops to paint and modifies the nodes.
And then, to finally adjusts the image textures, he jump back to absolute material mode.

Your last post reminds me that there was also an old idea of supporting painting on a layered image format like .ora. All textures of material could be layers stored into one .ora file.

Guys just check Multitile udim addon by Antonio Mendosa

It is what we need with additional features and nicely planed so there is no need to have node editor opened. Everything should work under the hood.

1 Like

Have had a little dig/whinge about how normal map baking in 2.80 is generally taking me almost double the time to get the same result I could get in much less time using 2.79.

Thinking of just sticking to 2.79 until something is down about all the additional steps needed to bake textures like Normal maps using Cycles in 2.80


I have never said a lack of features in baking. It’s about a lack of utility in baking.

So yeah you can pretty much bake anything, but it takes just way to long to do so. You can’t bake multiple textures at once, it is a big challenge to replicate Substance Painters name matching to prevent undesired projections on objects close to each other, you can’t bake multiple objects into the same image texture and the process of adding a new image Texture, selecting the correct resolution, inserting it into the correct material and having it active at all times during baking is just tideous.

And this is time you can’t afford to spend in any kind of professional production unfortunately.

Okay that kind of looks like a step in the right direction and fits one of my ideas nicely. But I still don’t like the idea of this being an addOn. We want to move people from substance painter to blender. And I don’t see this happening with some “weird” addOn

I’ll try to gatherr some thoughts, ideas and solutions soon and list pointed out downsides of those solutions. (just give me a bit of time, I’m currently writing a thesis. ^^;)

1 Like

This is an odd hangup to have considering they add good addons to the Blender distribution all the time.

I don’t personally think addons are a bad thing. In fact I love, that there is a blender addon for pretty much every problem you might encounter during using this software.

But I don’t think addons are that great for “advertising” a software as a valid alternative :c

At least that’s my impression based on opinions, I’ve heared from non-blender using artists when it comes to using blender: If texture painting is supposed to be a strong feature of blender it has to be properly done in “vanlla blender”.
Because as it is now, everyone thinks you don’t want to use blender for texure painting and I don’t see this opinion change with the help of an addon unfortunately.

1 Like

The biggest issue with Blender texturing is missing UDIM support. If you are coming from another application as a texture artist, you will feel like you just downgraded to 1999 all over again due to this particular matter.

Other missing texturing features can be compensated this way or that way, I can even paint my textures in MS Paint, but lack of UDIM cant be compensated easily, especially if you are contracting for a studio that expects UDIM textures from you.

I just did a project where I had to use many 8k textures in Blender because of lack of UDIM forced me to go that way, and it is not fun production wise. And single 8K (per layer) texture was not even enough resolution wise but we had to deal with it.


There needs to be a way to paint to individual RGBA channels. Using RGBA as mask channels is the norm for doing 3d game modeling development. This is a major omission in blender. You should be able to select/lock channels instead of having to create 3 or 4 separate textures and merge them externally with Photoshop.


There is a way using nodes, where you create three textures and use an rgb curves node to disable the other channels. A layer manager as discussed here (and similar to BATS or Bpainter) could easily do that behind the scenes, without any work in the Blender core.

It’s important to note that nothing else works the way you describe. Substance Painter just combines different textures into the channels during the export process

1 Like

Yeah … I think the main concerns regarding this topic can be solved by imporiving the UI. In fact for most of my own complaints I know a bodgy and (and sometimes a bit time consiuming) solution, you can pull of yourself with minor scripts or clicking.
A big plus of blender has always been the big abmount of control given to users. (I’ve even porposed a bodged UDIM support in a private message)

Big changes to the core will likely only be nessecary, when it comes to baking, brush system and stabilizers.

1 Like

I would welcome the use of cloning tools for texturing. Many times, I just want to clone a texture to cover a seam. Pick a ‘layer’ and clone pixels from another nearby part of the mesh, or have a scaleable overlay of an image that can be pasted in place. These are the elements that I do ‘elsewhere’ along with what has been said on texturing in general.

You can already do that.
Blender has a clone brush and there is the Stencil mapping mode for texture brush.
There are tiling options (in Image Editor, in paint mode, when repeat image option is ON) that have to be restored from 2.79b. They will probably come back for 3D View, too like they are present for sculpting.

I am confounded by my ignorance of this. I had thought that I had Googled the topic, but apparently my inquiry didn’t bring any answers. I have now watched several videos, thanks to you. Back to work.

  • Maybe seamless noise generator? (regular, musgrave, etc).
    Actually you can do it in a torus geometry (https://www.youtube.com/watch?v=CJUzDoK_PrM but is time consuming and not so accurate.

  • An option to bake multiple maps at once, ie.: diffuse, normal, height, ao.