Another possible reason Multires mod may not have been the ideal solutio, which may interest @Metin_Seven. 
I asked Jose about his Volume preserving smoothing:
‘Is there any way of implementing this as an alternative smoothing algorithm in the Sculpt Mode smooth brush?’
His answer:
'I’m not sure if it could work with new multires I will have to investigate. Plus on heavy sculpt meshes it may be slow since this is written in python… ’
So sculpt layers/corrective blend shapes?
It does seem like the logically better option but good luck convincing the people in charge… Everything must be ‘generic’ so you can spin a hammer to twist a wood screw instead of using a hammer as a hammer and a screwdriver as a screwdriver.
Yup. Take, for example, the auto-apply checkbox mockup image I posted a few posts ago. The only reason I made it was because realistically we’re going to be constrained by multi-res shoehorned as a modifier with dual use cases instead of getting a proper sculpt mode. A proper sculpt mode wouldn’t even need such a checkbox because it would be built in and separate from any modifier that was being used for other scenarios.
And if we ever get there, it could easily also act as a LOD system
Beyond a certain polygon count, but not very high.
I’ve submitted it as an issue, and Pablo has replied to it.
That’d be great, but Blender’s native Python module is indeed too slow for high-poly sculpting operations. There are a few faster Python modules that can be installed, such as NumPy and SciPy, but it will probably still be too slow.
The useful Mesh Fairing smoothing add-on makes use of the faster Python modules. You really notice the speed increase if you use the option to install the module(s).
I repeat that is not the case. You can put as many Deform modifiers as you want before.
And level 0 is editable in edit mode. It is not supposed to be high poly level.
It is supposed to be the low poly level.
So, it is not weird to not consider it as a sculpting level because sculpt tools are supposed to be useful for high poly mesh.
But I understand that may be frustrating if you create a mask with a gradient at an higher level to completely loose that gradient at level 0.
But if you created the mask at level 8, there are few chances that interpolated gradient at level 0 would be pertinent. And there are also few chances that working at level 1 instead would be a terrible experience, far away from what you would have done at level 0 or that Face Set from Masked would not be helpful.
So, to me, that does not seem to be a terrible issue.
On the other hand, changing multires into a basemesh property, that probably implies to make several adjustments not just to sculpt mode but also to edit mode & shape keys to ignore higher levels of multires.
IMO, that seems more reasonable to let devs evaluate that possibility during refactoring of modifiers stack into modifier nodes.
No. Each one has a certain degree of liberty.
Being responsible for an area of software means that you have to take decisions for that. But you can do that after reading users feedback and discussing with other devs.
And that is what they are doing when they are not tired or for whatever reason become mentally unstable. 
Grab is not generating geometry when dyntopo is used. And Elastic Deform does not, too.
So basically, Elastic Deform is already disabled like Grab in Dyntopo mode.
It looks like problem is more complicated than that.
Elastic Deform is not comparable to Grab because Grab is not analyzing the mesh farer than brush radius.
Elastic Deform is deforming the mesh beyond brush radius and tries to preserve details inside brush circle.
So, it has to use data of geometry the way it is stored in dyntopo mode. And the whole point of Pablo saying that dyntopo is slow and should be deprecated is about that way to store geometry different from standard mode or multires.
Yep. That’s why I hope a workaround is possible using a check to temporarily auto-disable Dyntopo as long as the Elastic Deform brush is active.
Yep. But despite using the old BMesh engine to divide / triangulate faces, Dyntopo works like a charm, even better than Sculptris Pro in ZBrush. If only the engine could be renewed, that’d be great. Otherwise, a whole new paradigm has to be introduced, like some sort of realtime voxelization, in the likes of 3D-Coat, but that feels like sculpting with mud, so it should have the feel of the current Dyntopo.
Overall I pretty much agree with @zeauro on what he said. Even if I think that some critics are justifiable when talking about how multires is embedded in the modifier stack.
The time spent on reworking the multires modeling / multires sculpting isn’t that long yet. Every solution starts with missing features and rough edges in development.
The fact that the basemesh isn’t seen as part of the multires levels has simple technical reasons I already wrote about. If that is covered ui-wise or adapted partially that won’t be seen any more and might even have benefits.
And what the performance and featureset for multires modeling and sculpting will be in the end should not be that much limited by the fact that it sits in the modifier stack.
Generally the quality of a multires modeling is rather determined by how effective the data will be modifiable based on the internal datastructures and brush algorithms and propagation algorithms. The rest is features features features. The stack doesn’t hinder it on that, and also allows to interact with other parts of blender or get a “deeper” more intuitive integration into the sculpting workspace
If anyone of you is hoping that everything nodes will be somehow usable in combination with multires modeling then be happy that it’s in the modifier stack right now, cause even if it is not exactly the same thing its more or not less a simple linearized version of a nodebased graphsystem and might help that it could be embedded there easier once.
Propagating the mask to the basemesh should easily be possible. But if it hasn’t been implemented yet, then it does mean it hasn’t to be seen as a bug.
But subdivision levels of multires would be a terrible Lod system. The levels resolution difference is almost always at 4:1. And the basemesh might be too finegrained but it would be the lowest level possible. There are much better meshrepresentations and algorithms out there, that have a much more fluent decimation method than subdivision surfaces and are fitting to a lod system much better.
I do agree that a modifier should be freely movable in the stack. Even if @Zeauro is right that it’s not totally fix but it still is limited and in principle it shouldn’t.
The first impression I got from his answer was more that the design was intentional and was fine the way it was. Sergey did however reply to the topic again, saying:
Face Sets were likely designed keeping propagation from level 0 all the way in mind. This is not how mask was initially implemented,
It is surely always possible to upgrade existing designs, but this happens outside of the bug tracker.
So it is up for consideration, but nothing seems to be decided yet. That to me is a bit more reassuring, but it is still important that they have a plan so that little kinks in the system like this problem get ironed out if Blender sculpting is going to feel polished and be of high quality. Long term the problems are going to come back and bite them if the devs ignore them for too long.
Just took a look at his post and yeah, pretty sure that will come.
Hey sculpting folks! I was working with photogramerty and I was wondering is there’s something like the spot healing brush from photoshop, I don’t have photoshp in english, but I’m referring to the one that is not automatic, where you can pick where to sample from. If not I think it could be a very nice feature for this field where we sometimes want a reconstruction, I want my photogrametry to look good looking not accurate.
Are you talking about sampling sculpture details like the Xtractor brushes in Zbrush? 
Or are you talking texture stuff only?
Ahh, yes, that’s exactly what I want, sculpting a few things in blender is easy but zbrush is alien for me. Maybe I can do something like this in substance painter with a LP object? It’s an ancient sculpture with some small grain in the surface and is an asignment for the uni so I don’t need it to be perfect.
Well, this kind of performance is absolutely awesome, do you think we can expect the same performance between different face masks instead of different objects?
Just imagine having a 100 million polygons model with sixty face masks and this kind of performance.
The only big handicap I can think of is painting between two different face masks.
I hope something like this is possible within the current code.
I don’t think, so.
Here, there is a multires modifier on each object. And basically, all objects that are not active one, are at lower preview level than Sculpt level of active one.
So, although scene is at 96 millions of vertices, viewport is probably displaying a lot less.
We could image to have an option to display Face Sets that are not under brush cursor at preview level. Contrary to radical current situation where we can hide all face sets except desired one, that would have advantage to maintain a silhouette of entire object while sculpting detail on isolated part.
But at some time, you will still need to create small face sets at high level from bigger ones, to sculpt without face sets restriction or simply to display object with all details visible.
So, I don’t think we should expect from Blender to be able to handle smoothly a 100 million polygons model in 2.9x.
You right, I totally missed that all the other iteration were displayed with a lower level of detail. anyway, trying to implement this in face sets is still super useful.
You see, I am not aiming for a perfect solution and I don’t think even if someone already working on this its gonna make it to 2.9 (actually with Pablo and Sergey/Brecht you can never know). but even with all the imperfection, as you mention, I still think its at least a nice feature.
Anyway, I just wanted to suggest this as a random idea that popped in my head 
, I didn’t really think too much about it.
Multires compatible Shape Keys with auto-apply base???
It gets crazy slow on lower levels and since it’s a decimate modifier, only certain topologies unsubdivide correctly.
Quick extract alphas would be awesome for Blender sculpting
