Blender already has the base for material creation form scratch.
But many advanced nodes allowing lot more advanced creation are missing, like spawner and randomizer, or shapes modifications with gradients and many more.
Not saying Blender should be as good as Substance Designer, but will it get more advanced shader nodes for procedural textures creation ?
I think they are currently working on the procedural stuff - I’m sure I’ve read something about it somewhere. But I did see some pretty bonkers stuff being made with nodes in Blender last year on Twitter. #Nodevember I believe was the tag - if you have Twitter, take a look.
It seems an awful lot is possible with what we have, it’s just having the knowledge to do it.
Like this one
It’s gets too complicated with limited nodes available, it needs more basic features and some advanced nodes.
More nodes will be really welcome (spawners, masks, edges garident dither, edge offsets, and more) for advanced textures and make it as easy and fast similar to Substance Designer.
They start using Substance Designer for their rocks making, because it’s high quality results and it’s unimited variations once you made one texture.
An overwhelming amount of math is necessary for a lot of those. In some cases the artist knew that math in other cases they purchased a node system that gave them a lot of utility nodes that removed a lot of the math overhead.
Most usefull nodes i used are not math and not available in Blender nodes.
Many mix modes, many clamp, spawners, geomtry shapes, advanced gradients, edges dither, height map nodes, bump mixing nodes, many more different noise and patterns generators; and lot more nodes that are work on texture and not math advanced.
I seen some cool Blender materials, but it’s too much Blender nodes needed because lot of nodes are missing, so it’s lot more work and work around compared to Substance Designer, sand many things are not possible in Blender shader editor nodes.
In Substance Designer you can make great materials without going advanced complex nodes.
Just using noiss generators, mix and masks nodes, random spawners and color ramps.
While Blender only have very few of those nodes and some does not exist in Blender.
While it could improve a lot if there was people working on it.
Absolutely agree with what you say. With some enhancements in the procedural texturing system Blender could come very near to the abilities SD has.
This is not a texture issue, it is a shading issue but let mention it too: A fully fledged edge detection node for having the opportunity to produce wear and dirt effects in objects! A so evident need and still it has not been dealt with (as far as it seems there is not even a discussion about it).
Today’s bevel node does some work but it is very limited. It generates its effect undiscriminately on all edges of an object. It is not offering the option to choose between concave and convex edges, it does not have an edge degree preference and an edge sharpness (or softness) choosing feature.
As said before, Blender has all the base structure for including such tools in its texturing and shading side. All the needed thing is taking some steps forward and… we will have something very near to a SD and SP in it!
For Cycles at least, it used to be that the devs. couldn’t just add nodes willy-nilly because the CUDA and OpenCL compilers could not handle large kernels (which impacted development even though you can also render with the CPU).
Nowadays, the CUDA compilers at least are far better, so the result over the last few years was a lot of new texture and effect nodes. Right now, the biggest issue is resources, as not all of them are around at all times and Brecht can often be busy with other tasks (if not on break). Then there’s the other very important task of optimization (like OIDN and adaptive sampling), your fancy new nodes are useless if the rendering can’t be done in a decent timeframe.
You can just as well say that it’s useless to render fast if you can’t render what you want.
Both these statements are silly.
You need both speed and features, so the resources for both ought to be balanced.
And they mostly are.
I agree that Blender needs procedural texture nodes, but it’s a false premise that this will happen with Cycles nodes. Render time is not the right time to evaluate your procedural brick texture or whatever a million times only to get the same result over and over and over. That would be inefficient and stupid. It’s more likely procedural texturing will happen by either repurposing the Compositor, or designing a new system from scratch.
While I think bringing this capability to Blender is good however most artists honestly lack the understanding - skill to do this well.
To me it seems this is such a specialized thing that you need to do it a lot to keep fluid at it.
This is like procedural wood. In theory easy but in reality requires a lot to know to quickly jump into it.
Maybe a Substance designer bridge will be sufficient simply keeping compatible with that system which is now established in the industry.
But dont get me wrong I would welcome any addition to the nodes to have more play in Blender.
It would not be possible though to have a bridge that would work with anything other than Cycles (because of the GPL). Thanks to that any use would be very limited, as it would not be able to actually interface with Blender’s editors directly.
However, a bridge to a FOSS alternative like Material Maker could work (though how complex it would be depends on whether or not the node setup gets baked to an image first).