Personally I would like to see this continue inside Blender:
I have not read anything here that convinced me it should go another way.
The argument not to work with Blender developers doesn’t seem sound either. That’s my opinion.
Rather I would like to see more money flow into Blender development for this. Not in a third party solution.
Again. My opinion.
I strongly disagree. I never said I think we should stop working on Blender’s sculpting mode; I said I think there should be an open-sourced alternative to ZBrush. Realistically, unless the devs overhaul most of Blender’s fundamental code and remake it to suit sculpting, it’s never going to be able to get close to what ZBrush is doing currently, and if the devs did that it would wreck the code for everything else.
As for why we should have this kind of program, I suggest you look up images of ZBrush sculpts and Blender sculpts. ZBrush is obviously far superior to Blender in sculpting, which makes sense because most of ZBrush’s code is based around sculpting. A third party solution is a good idea because we can’t decide Blender should primarily be a sculpting program and neglect the other stuff.
Last year I posted a wishlist for Blender’s Sculpt Mode at Right Click Select, and the vast majority of the wishes are ZBrush features. 
A number of these wishes have been realized in the Blender Sculpt Mode Features branch in the mean time, and those tools are now gradually being added to Blender’s 2.81 master builds, happy to see that.
One of the reasons zbrush is so much more popular than blender, is because of the tutorial videos. Blender is great program worth learning, but getting the learning information is hard. Zbrush makes it easy to learn zbrush, it’s part of their sales scheme.
If one wanted to make a program that is better than blender and could be as good as zbrush, it could be worth taking the blender source code and starting at that point. Pablo is making blender sculpting better. Someone might build ontop of Pablo after he is done.
Blockquote
…are talking about making a new data structure specific for Sculpting, so when you go to Sculpt mode your meshes get converted to something that are optimized for Sculpting than back to regular mesh when you exit Sculpt mode.
…
I hope in the future multi threading will be used. It’s sad to have 12 or 16 cores, and only one of them is used. The blender team said more than one causes too much overhead.
A bit of wishful thinking. That’s why i said it needs to be done outside of Blender community because those who already contrubting are working on blender, same as it happened with Pablo Dobarro he didn’t start developing new app instead he made small patches first and now he is hired full time, so anyone who is working with Blender wouldn’t be interested in starting a new project from scratch.
Actually I was not responding to you directly on all of the points I made which is why I left it open.
I don’t see the need for another half-baked sculpting application out there. And my opinion should not guide if this happens or not. It is just my opinion.
But it better kick some ass or I am not going to even think about switching over from Zbrush just because it is open source.
So if someone is going to do this, then please take the time to gather 5 years of resources and build it from the ground up with a small team. The question would be, on a practical level, where is this funding going to come from?
Because of the fact I can obtain one of the best perpetual licenses out there for an application - I have not paid for an upgrade of Zbrush in 6 years - and the fact there is already a number of Zbrush “wanna be’s” on the market, that don’t come close, I’d rather see Blender’s sculpting improved dramatically.
And also you are going to have better luck seeing money go into Blender than you are into a stand alone app. That is just my opinion. With a volunteer team it could take a long while to come up with anything even remotely useful. So this initiative would have to get money.
But the value of not wasting this energy and resources on a stand alone app - I mean you are asking it here on these forums after all - is that vast improvements going into Blender’s sculpting mode has a number of great payoffs over the short term than that energy paid (or even volunteered) into a stand alone app.
Zbrush has a number of limitations in transferring details and baking. And overall while it is the best sculpting app, (we won’t use anything else at my studio) you still have to do something with those meshes in another application to move down the pipeline. And this is a difficult bottle neck that takes some thinking and care.
At the very least, if Blender could easily handle similar numbers of polygons it would open the door to more baking options in Blender. And that with the retopo initiative would make a serious dent in the sculpting pipeline.
Another benefit is for those who can’t afford Zbrush, to bring better sculpting to Blender.
And I think the tools would go further quicker in Blender than a more long term 5-10 years to develop a competing stand alone app to Zbrush.
Of course it would be great to have an open-sourced zbrush, it’s just quite complicated to make it append.
Same statement goes to an open-source compositing application (like Natron), or something in the lines of Substance suite.
Zbrush have years of development time with really good engineers .You’ll have to find good engineers and pay them to work for at least a year before having something to show. And chance are that it won’t be useable for another year.
Finding users willing for a Zbrush alternative will be easy , finding enough funding and software engineers to work for years will be quite hard.
Most of the open sources softwares starts as a personal project lead by a developer, or is funded by a large company . Maybe you should start to find developpers and ask them how much time and money is needed to make a V1.0 , then study how much funding you’ll have to raise.
Also it will be easier to raise the funding if you show that you have a solid plan.
This is a misconception. Ultimately, all you need to do is handle input and draw the right values into the GBuffer, and you can do anything you want within Blender. You don’t need to use virtually any of Blender’s code.
Sculpt mode already is quite separated from Blender’s edit mesh mode, which is somewhat separated from the modifier stack.
So, you don’t need to write another application to have better sculpting. Not having to switch apps is valuable as well.
The real question is: Why would anybody do it? Why learn these really high-end software development skills and put all your energy into a product, only to give it away? Developing a sculpt app to show off your chops is one thing, developing and maintaining something to rival ZBrush on a FOSS budget is something else entirely.
That’s how sculpting works for many years already. Sculpt geometry is stored as a so called PBVH (painting bounding volume hierarchy), which is an optimized BVH structure for high poly geometry. There were also experiments with Voxel based geometry description though.
By the way, the conversion to regular mesh data may be quite expensive and cause annoying hiccups. That is also why auto-save is entirely disabled in sculpt mode, a super ugly issue users should be aware of.
which is not good enough to compete with zbrush
I thought bvhs were a sort of standard data structure that a lot of apps use. What type of data structure does zbrush use for its geometry?
While it’s certainly true that the tools and workflows in ZB are far more powerful, another factor is the quality of artists. ZB has a large pool of world renowned artists and veteran pros who have been working in the program for a long time. It also attracts(and was initially built with the world of trad sculpting as an inspiration) traditionally trained sculptors and VFX make-up artists(think Rick Baker)
Not to sound too offensive, but most Blender sculpts I’ve seen are not up to scratch because the artists themselves are amateurs or part-time hobbyists. I definitely have seen some really nice sculpts done in Blender, but that was entirely down to the skills/knowledge of the artists.
Insanely detailed bio-mech/hard surface and photoreal characters are a different story and this is where the power of ZB exponentially overtakes Blender.
Damn, I didn’t know about that. And yes, that’s an ugly issue, lol, even though I don’t use auto save in any app, ever.
ZB stuff is proprietary and patented as hell, they will never reveal it. 
I was thinking that this was just an assumption. If no one knows what type of data structure it uses, then there is no reason to think that its performance has anything to do with the data structures it uses. For all we know it could be some bvh type that is just like blender’s and the performance difference could be because of other optimizations.
That’s not an easy task, even companies with cash are struggling. Kanova I’m looking at you.
If it was patented we would know how it is done, but as far as I can tell, there are no patents.
At least Pixols are, so…
Sure we can’t tell exactly what kind of hack zbrush does to achieve that crazy performance, maybe it’s because it’s doesn’t use the GPU/opengl???
I don’t know man, maybe opengl is the bottleneck in blender? 
lol
because using the 4000 1ghz cores on your GPU instead of the 12 cores on your CPU is the problem, right, makes perfect sense.
and what exactly do you think is the difference between say, “metal” and OpenGL? are you going to magically not have shaders?
After writing metal myself, I can tell you, it’s like writing apple syntax OpenGL
Edit: Just to be clear, blender would lag at 100million vert whether you draw them to the screen or not. So, painting the viewport is not the slow part. I did investigate Zbrush, and the way they handle huge numbers of verts is through isolation. Similar to how in blender, you have a different viewport subd than render subd, in Zbrush you mask an area and in that small area it exposes them to allows you to sculpt a few million verts. It’s clever but not difficult to implement, we could likely have one single programmer implement it in blender.