Does blender 2.8 will have smart material like 3d coat and substance painter ?


(Richard Culver) #41

I think there is a disconnect here between vision and reality. There is no point in following a vision that does not lead to a reality.

Additionally all visions need support to become a reality. Gaining support and effectively bringing about a vision comes from first a realistic assessment of current reality which includes interests but also realistic audit if resources.

Everything else is just pointless talk.

What might have been relevant to Blender 10 years ago as a realistic vision is not relevant today.

A very cursory look around the 3D world that Blender lives in will paint a clear picture.

Further investigation would enligbten why.

But the bottom line is to get development money it takes support.

That requires understanding the current needs of most users. That has changed drastically in the last 10 years.

And yes. This does matter. Blender does not live in a financial bubble.


(ambi) #42

This is very true. But I also think implementing near real-time texture map nodes like Substance Designer would make a lot of these possible. Which wouldn’t be a (really) huge amount of hard work.


(Richard Culver) #43

From my experience these systems are very complex. And I am just taking a wild - but educated - guess that they are far more complex and expensive to make than people in this thread have hinted at.

With Blender’s current funding set up, I think it would make more sense as a 3P addon.

I could be wrong. :slight_smile:


(Grimm) #45

I would say you are spot on with your experience on just how complex these systems can be. Just looking at programming languages, they all only have 20 to 40 core commands. Which are very general in what they do so they can be combined into much more complex systems and have a known effect. It doesn’t take many more additional commands to make the system too complex to maintain and troubleshoot.


#46

Really if we just get better edge detection and maybe an asset library we’d already be 90% of the way there.


(Romanji) #47

Nope i don’t think so, and i think that has been discussed in this thread already. Your definition of what is appropriate to call “there” is fundamentally different than what other people have in mind.

A proper edge detection, curvature/cavity/position and thickness node would be a nice start, …

But i think the whole painting system needs an overhaul, and instead of going the same direction as Substance and 3DCoat i would rather prefer when Blender also takes some notes from Mari instead.


(Richard Culver) #48

Precisely. However., the biggest and probably most important note to take is performance with large assets and multiple UDIMs.

I would say full support for UDIM and a performance boost would be the starting point. And that would require a massive refactoring of Blender.

Think Blender 3.0 which if you are doing the math should arrive around 2026.

And I wonder if some of the people responding to this thread are grasping that it is Substance Painter. The operative word being substance.

And when you grasp what that is, a product of Substance Designer, it is another level to the complexity.


(Unreal3DFX) #49

For start 2.8 should have just easy texture or better say, material painting feature from external images(Base color, Metallic, Roughness, Normal, etc…) per material. For example you generate textures from Substance player or download from freepbr, cc0textures and create material then paint to model and export painted material for model you painted.

I have faith in new development fund that they think something eventually and upgrade current painting system.


#50

What can mari do that substance can’t?(actual question)


(Ace Dragon) #51

Color-based AO/curvature shading in Cycles can now be done with a brand new node found in the buildbot builds (both master and 2.8).

Eevee however does not yet have it.


#52

This sounds great!

What’s it called?


(0rAngE) #53

Could not disagree more with some of the initial posters here.

I’m fully on the Wazou ™ train!!!

Blender has the BEST sculpting tools of all full 3D Packages (so not counting ZBrush or Mudbox as they are standalone niche).
I’ve been employed for the past 10+ years working in studios as a 3D artist. Last 4-5 years I started integrating blender into my toolset, as much as it can fit within the studio pipeline I’m in.
I sculpt professionally in blender, even though the studio (I’m currently at) has a ZBrush (and Mudbox) license. Why? Because I genuenly prefer sculpting in blender than those 2! And my sculpting is not suffering for that, I can keep up with the guys that use ZBrush - easy. There is also a lot of benefits of me having an environment that is a full 3D package with modifiers, insanly good modeling, retopo, cycles (now eevee), etc etc. I use all of that in my workflow!

Why give up on such a gem?!
And eevee coupled with the node editor is 80% substance/3d coat. Why not push it further.

If you guys suck at sculpting and modeling, and love rigging and animation - cool. But don’t think we’re all like that!
There is a large and vibrant blender sculpting community!

Just check the latest from Jama Jurabaev to see weather it’s worth investing into sculpting. He’s an amazing ZBrush artist, yet he still resorts to sculpting in blender if it’s something he feels he doesn’t need to go to Z for. That says a lot.
And then our own BA Dyntopo Crew ™, I could name 10 pretty amazing sculptors, many professionals, that use blender for sculpting. And all of these guys have tried/used ZBrush at some point.

SCULPTING FTW LOL
…and texturing :slight_smile:


(Romanji) #54

Mari can paint on insanely high texture resolutions (8k, 16k, 32k) , over several UDIM’s (10…100)
It is more like a 3D Photoshop, with tons of tools available (reads Photoshop brushes), but it also has a fully procedural workflow and a Nuke like Node-editor.
Its incredible robust when dealing with such extreme amounts of data (if you have the hardware).

The main technical difference is that you don’t paint directly on the model, you always paint on a so called paint buffer which is basically the screen and then project from paint buffer onto the model (like a camera projection). This makes it possible to paint/project on countless UDIM’s.
It is this workflow which i think could be very helpful in Blender when dealing with UDIM’s.

Substance Painter has a different workflow. Substance safes all your paintbrushes in memory, so that you can change the texture resolution and the program basically redraws all your paintbrushes in a higher or lower resolution. This makes it quasi independent from resolution and kind of non-destructive.
But the downsize is, that it fills the memory and it is difficult to adapt this system so that it can deal with huge amounts of textures and texture resolution. That’s why Substance can only do 8K and a couple of UDIM’s so far. You also can’t draw over multiple UDIM’s simultaneously.

Substance still has the edge over Mari when it comes to create stuff fast and automatically using smart materials and smart masks, but ultimately Mari gives more granular control over anything, even if that means more manual work.

So Substance is ideal for props and hard surface stuff and Mari for Hero assets and characters.
You can use both for both, but at a certain point when dealing with large Data-sets, Substance becomes too limiting and too slow while Mari just keeps on going.


#55

just do add to Romanjis excellent explanations, Mari can also import Photoshop Layers from a psd.