It really feels like this is what he was talking about in his plan, particularly the normal map sculpting. It is really sad to think this could have been Blender, if Blender’s community wasn’t so idiotically poisonous, but at least I’m glad he got to implement his vision in his own project.
9 Likes
To be fair it wasn’t really blender community that was the nail in the coffin, Pablo had a vision for blender that conflicted with the direction the blender foundation wanted so he left to make his vision real, yes that could have been blender but it just was not meant to be, blender is not interested in being an asset pipeline first and foremost, blender is interested in being an app for creating art in and opening the software up to be more accessible, to be able to create what ever you want “Everything Nodes”.
6 Likes
Agreed.
I have a very strong feeling that they have already seen this recent feature and are already holding a meeting on implementing it.
Iirc, Zbrush implemented that cloth brush and some other brushes idea Pablo came up with when he was working on Blender some years back.
I swear I remember Modo having a sculpt to exr normal map texture file feature a lifetime ago.
You do not, though you aren’t supposed to clog that forum with bug reports or feature requests. That’s not what it’s for.
As a non-dev you can test out new alpha/beta features and give feedback on them. Usually bigger new features get their own thread there for this purpose.
1 Like
That is difficult, because there is a competition between issues.
Developers are really collecting feedback from a lot of places during long periods.
I am often impressed how their synthesis of the whole of complaints is coherent, when exposed as an article or design task.
Where I am disappointed, is when an aspect has been forgotten corrupting the whole or when their order of priority is not what I think should be.
Most of time, they are honest, able to defend their position and even able to recognize that they made a mistake or forgotten something.
But, sometimes, they don’t. They are humans.
They may lack time to discuss, support to achieve work before timeline, be tired and sincerely forgot things, lost track of long standing issues, have issues with real life.
When it is not the time to discuss about an issue ; they are blunt, saying that they have something else to work on, before.
But since they don’t want to be spammed all the time by thousands of users, about thing they will work on, later. That is their responsibility to open discussions on what they are working or will work on, in near future.
They really read a lot of complaints. They honestly try to make a synthesis. On that basis, they make proposals that may contained trade-offs between contradictory user requests.
When this hard work is done, for the part they brought answers to ; they are not inclined to change everything, just because somebody is vaguely complaining. They need solid elements to revise their judgement.
For the part, they did not succeed to find an answer, to make an arbitration, if the user community may debate endlessly on the subject or completely ignore it, they can make an hazardous choice and pick a bad or unsatisfying solution.
Accumulation of mistakes, hazard and long debates is a democratic process, taking a while to produce results.
Honestly, what they are working on, have been requested, years ago.
There are windows for users to influence development. The most efficient is when there is something concrete to test (patches, branches, betas) or when something has been officially written ( blog articles, devtalk threads, design tasks ).
During a period, they were sincerely picking some ideas from Right Click Select or Pablo Vasquez’s weekly livestream to integrate in the next release.
But the refactor started with 2.8 is monstrous and had to be spread on a decade.
We are on the middle of this decade. So, there are a lot of trains that have been launched (Vulkan, EEVEE-next, Grease Pencil 3.0, Simulation Nodes, Brush Assets, Character Animation) , without stops scheduled. And desynchronizing is hard to handle.
They have coordinators. But they don’t think like managers.
They try not to merge in main branch what is too unstable. They are hiding some stuff under experimental tag. But unless that, there are uncompleted, not polished features, that ends up in releases or last waiting for a second step during several releases.
There is still the idea that a release is more tested than betas, and that LTS will have QoL improvements and debugging, anyways.
So, they stick to their schedule and release current state of main branch.
That is not simple to find balance between deceiving users by postponing feature or deceiving them by landing a limited feature.
Because we still have half of a decade of big changes to happen, that is normal to have frustrated users until 2028.
Nobody is able to say what next release will look like. Because nobody knows how advanced will be next big thing.
And because nobody knows when something big will be there and, at what state, nobody can anticipate its integration in the pipeline, months ahead.
So, feedback from users, that are generalists, individuals will happen during beta.
But feedback from users, that are involved in big teams, will happen a semester or one year, after release.
So, developers may be praised for something at one moment, and roasted, years later for the same thing, by different people.
They are just humans. That is not possible to satisfy perfectly hundreds of other ones.
But users are facing a software that may be complicated for simple things, by persistence of obsolete parts of the design, and incompatibility of those with new stuff, that may lack polishing.
I was expecting a pause after 3.6. I would have preferred a bigger one, a 3.7 and a 3.8 to improve neglected texture workflow (painting, baking, UV mapping) to a postponed Blender 4.0 launching new trains.
I would not say that I have not been heard.
They launched Brush Assets and UV mapping work, too.
But I am not satisfied, either. I was expecting something with a bigger amplitude.
But I can not denied that Vulkan or Character Animation stuff are very long waited targets, too.
EEVEE-next, Grease Pencil 3 and Simulation Nodes have not miraculously emerged from developers mind. That are the third iteration of features, fruits of an important feedback of several years from users.
To live the transition without frustrations, amount of developers should have the double.
2 Likes
Zbrush started as a software to produce stills and collectibles, too.
Normal map is useless for 3D printing.
They simply reached the more universal target, handling millions of polys with ease.
That goal reached, they just had to convert to satisfy the others.
Because that ease is not possible in Blender, normal sculpting option was a debated thing.
But I am sure that I have seen lots of normal/vector painting plugins, addons for other DCCs, before.
A developer probably needs an experience with developing sculpting tools to do what Pablo did.
He will probably be imitated. But he is one step ahead.
I wish for him, that he will succeed to maintain a gap.
The following steps are obvious : supporting parallax mapping, converting the normal map in a vector displacement one or using AI to obtain real displacement.
2 Likes
Whatever happened to his retopo app. Cozy Blanket? Is Uniform going to have long-term updates and support or will it go to maintenance mode in a couple of years as he moves on to his next big idea?
Eventually, this will be a lot of code for a single developer, Pablo will eventually need a team if he makes enough sales. This appears to be a major missed opportunity for the BF it is turning out, but as a longtime user I am quite used to seeing them choosing to simply drop the ball instead of actually taking advantage of an opportunity to get ahead.
However, I am also used to the fact that it is not unusual for the feature to eventually find its way into Blender anyway, which is a good reason to not resign to hoping the project gets a hard fork so we can leave them behind.
Thanks @ zeauro. That was a really useful explanation. Should be pinned for those who wonder why BF doesn’t appear to listen.
He says CozyBlanket will receive updates. I think he doesn’t want to, and Uniform will have all the features CozyBlanket has, but he has responsibility to customers since it was commercial app.
And I think its more of a problem for Pablo than Blender, because as you say, what he’s doing is simply unmaintainable in long term. His best bet is to sell app to one of the big guys, like everybody else before him. Blender was his way to have some ownership on his creations long term. I don’t think Pablo and Blender’s story is fully over.
2 Likes
As per their Twitter account, I think there are at least two developers working on Uniform and Cozy Blanket.
—
In related news, a PC port will eventually come: https://x.com/sparseal/status/1732067369437057243?s=46&t=vkm0g7EyOP65qqUSYf9YLA
4 Likes
cgchannel says Sparseal is former Blender sculpting tools lead Pablo Dobarro and a Godot Engine developer Joan Fons.
4 Likes
Except we’ve been doing this for years already with Substance Painter.
This was one of the winners of the ‘Meet Mat 2’ Substance Painter competition from 4 or 5 years ago. This was 100% in Painter using painted disp/normal maps.
And this is blank Matt if you don’t know him. 
so yeah, this has been around for years with Sub Painter, and many game dev asset workflows moved away from sculpting the details in Zbrush. It all gets baked to a normal map anyway so it makes way more sense to create your normal details at the texturing phase, for many reasons.
7 Likes
And 3D Coat as well. 
2 Likes
I always forget the red-headed stepchild of 3D… 
I have tried this work flow in SP (not 3d coat) and honestly it never really feels like you are pseudo sculpting feels more like im trying to stamp detail, i could be doing the whole workflow incorrect but the control i got was nowhere near as good as just sculpting the details and baking down.
4 Likes
Well, it’s not “just” in Substance Painter from the looks of it. The original has smooth geometry, but the example given clearly has had the model modified with extra geometry. For example:
I’ve tried sculpting normal details in Substance before and it really doesn’t work that well out of the box. Yes, you can paint in details, but not to the point where it feels like I am actually sculpting the details. I still need to do a proper bake of stuff such as edge bevels and other details to achieve those same results. Whatever Uniform is doing makes the results look like actually sculpted in details without any additional complexities to set everything up, so it is definitely doing something different.
1 Like
Yes, it was never designed to be a true sculpting experience. It’s painting. I was just trying to point out that this isn’t some revolutionary idea, it’s been around for a few years now.
This is the whole point, you’re adding displacement details. There’s no extra geo added. What you’re looking at is displacement information. This competition was to be done completely in SP using the Mat mesh pictured above. I entered the competition myself. Same for this, not as complex, but painted displacement detail.
It’s very simple. You just add a layer and start painting on the normal or height channel. Height/normal info gets baked to a combined map. There’s no setup involved.
Yes, it does look pretty cool what it’s doing, as limited as the examples he shows are. It will be interesting to see how complex the details can go, but it looks like he’s aiming it at simple, low-poly stylised assets. Maybe that will change as the program evolves?
3 Likes
Painting on normal maps is terrible and not catching on for a reason. Its unnatural. You want to create fake geometry, you want to create facets, cavities, sharp edges. Doing that with painting tools is not good. You need to hack brushes and do multiple extra steps to make it look real
Simply using scrape brush and trimming edges on normal map is truly revolutionary. You dont need to think about anything or set something up. Youre simply sculpting.
2 Likes
Sculpting will always be more intuitive, but the result is exactly the same in the end, it all bakes down to a normal map. It doesn’t matter if you use an alpha in Zbrush or Painter, the result is exactly the same as it gets baked into a normal map. Plus you’re also limited in how many verts you can capture to your resolution. More than 4 million will be lost in a 2K map, for instance, whereas painting directly on the normal map is entirely UV based.
I’m not sure where you’re getting this information. The majority of the games industry moved on from full reliance on high/low bakes years ago. And many modern game workflows don’t do any normal map baking at all.
3 Likes