Voxel Remesher (only) sculpting vs. Multiresolution Modifier

[proc24]

Hey there,

Im almost finished with the lecture Retopology & The Multiresolution Modifier. Now in the previous lectures we blocked out the model, then sculpted the basic shape. In the Retopology & The Multiresolution Modifier lecture we then proceed to take our basic shape model with a lot of face count (about 800.000 for me), used the quad remesher to get the face count down (to about 12.000 for me). Afterwards we add a multiresolution modifier to get back details so we can sculpt the mid level and finer details Im assuming (coming up in the next lectures).

Now i have done multiple other blender courses on the GDTV site and when I compare this process to the Blender Character Creator v2.0 for Video Games Design Course I was wondering why we did not just sculpt the mid level and fine level details directly on our initial sculpt with a lower voxel size, then use the quad remesher later to get our low poly dragon and use the high poly dragon to bake our textures.

Now this course of course is working on blender 2.8/2.9 I think, while the other courses are newer of course, so this process might be due to an older blender version, but I was just curious as to why Grant did it this way in this course. Is there any benefit to either approach? Or should you not skip the quad remesher and multiresolution step we did in this lecture at all?

I would kind of argue from my knowledge sculpting the mid level and fine level details on the basic shape mesh would be fine, we just need to lower the voxel size then, no?

Thanks in advance!

[NP5]

There are many ways in Blender.

Some suit people better than others. Some may suit situations better then others.

MultiRes had had some fixes or improvements if I recall correctly shortly before the Dragon course.

It is probably the best route. (Personal preference obviously heavy here!)

Basically you are starting with good topology, or can/should be. Multires then makes higher density of good topology.
Retaining the lower levels, which can be baked to from the highest version.

So you make it once, the other way is make a messy topology high poly and then have to make a copy properly by hand, or hope to get away with cop out one-click remeshers product.

[proc24]

Honestly, this feels like a conversation we could spend hours debating on with a beer in a pub somewhere, or maybe its just me

What wondered me is with the multiresolution approach we have these steps

• Basic Shape Sculpting Starting from Block (high face count, 800.000 for ex.)
• Quad Remesh or Manual Remesh (low face count, 13.000 for ex.)
• Multiresolution Modifier (low face count 13.000 and high face count 800.000 or so depending on how many times you subdivide in your multiresolution modifier)
• Next step would be Mid level and fine level sculpting and adapting the multiresolution modifier as you need

Whereas when we skip it as we did in the Blender Character Creator v2.0 for Video Games Design course our process looks like this:

• Basic Shape Sculpting Starting from Block (high face count, 800.000 for ex.)
• Mid Level and Fine Level Details Sculpting, work with remesh as you need detail
• Quad Remesh or Manual Remesh (low face count, 13.000 for example)

Then after both the processes we would be off to texture painting and or baking I assume.

I guess its just a minor detail because if you remove the multiresolution modifier step both ways are the same with the sole exception that in the multiresolution approach we first do the retopology then do the fine sculpting and in the latter approach we first do the fine sculpting and then do the retopology. So both the order of the steps and the extra multiresolution step got me confused for a moment there.

I think my reason why I think I would opt for the 2nd approach is that I still can adapt the shape more or less freely while the multiresolution modifier allows for minor adaptions of the base shape, but not greater ones if Im not mistaken, but then again I havent done much sculpting outside of the GDTV scope.

I guess the main take away is that both ways are valid and we can choose the process we prefer, thanks for clarifying that @NP5

[FedPete]

With multires, you can efficiently modify the sculpt on many levels of details (LOD), which is important for game design. For a building in the distance you need probably a cube (8 vertices), or simply a bitmap of a building (4 vertices).
Closer by more details on the mesh are needed (increase of vertices, more CPU). This is called Level Of Detail.

But if the model is ready, then yes your solution of re-mesh will work. But remember most challenges are just about creating something. No need for smart way of working. But if you say you need a good model for animation (as animation, or in game), then it will be important to know how you use the model. It usages, and how to maintain the model even when it seems to be finished.

Game dev courses don’t discuss these project management skills in much detail.

[proc24]

Hey there,

I read a bit of documentation and watched some videos about the multiresolution modifier and I thought I would share my “findings”. Now its not exactly something new. As @NP5 correctly pointed out both ways using a voxel remesher approach and a multiresolution modifier approach is valid and each has their pros and cons. @FedPete also correctly pointed out that using a multiresolution modifier you can efficiently modify the sculpt on many levels. If you havent used the multiresolution modifier enough like me though, chances are you are not really understanding what @FedPete meant by efficiently sculpting on many levels of detail.

So i went ahead and watched this youtube video “Multires modifier with Sculpting tools is awesome!” in which the creator explains in a bit more detail what sculpting on many different levels of detail means.

Now putting all the knowledge from this thread, the blender docs and the YT video linked above together, here is what I understood: If you are sculpting on the multiresolution modifier say on sculpt level 2 (so you subdivided 2 times) your multiresolution modifier will look like this:

You then make some changes for your current level of detail, rougher changes as we did throughout the course. Then the time comes where you need more topology to work on finer details, for example hands, eyes, feet, maybe scars and so on. So you subdivide 3 more times, making your multiresolution modifier look like this:

Now what I thought this meant is simply doing a subdivision 3 more times and thats it. What it actually means is that you have sculpted your rough details on sculpt level 2 and you are now sculpting your finer details on sculpt level 5. More specifically, the sculpt details which you have sculpted on sculpt level 2 are stored on sculpt level 2 and the details you are sculpting now on sculpt level 5 will now be stored on sculpt level 5. This also means, if you want to make changes on which you would prefer working with a lower level of detail again, you can actually go back to sculpt level 2 by setting the sculpt level back to sculpt level 2. Blender will still keep your fine details which you have sculpted on sculpt level 5! This comes in really handy for example using the smooth brush in which we know the more topology the less of an effect the smooth brush has.

The voxel remesher on the other hand does not have that capability. If you remesh to say 100.000 faces and make your rough shape and then remesh to 1.000.000 faces to make some finer details, you cannot remesh to 100.000 faces again to work with a lower level of detail again that easily, because you will actually lose the finer detailed sculpt information.

Now this doesnt automaticly mean the voxel remesher is bad and we should always use the multiresolution modifier instead. The multiresolution modifier is not able to adjust the level of detail as fine as the voxel remesher is. Say for example you are using a multiresolution modifier to sculpt and you are working on sculpt level 5 with 1.000.000 faces. Now if you need just slightly more topology, maybe 1.500.000 faces, you cannot do that with the multiresolution modifier, because if you subdivide again you will be at 4.000.000 faces if my math skills havent given up on me.
The voxel remesher on the other hand can do that. Need just a little bit of more detail? No worries, just slightly decrease the voxel size, remesh and off you go!

This means you have more fine grained control of the level of detail you want to work with using the voxel remesher, which is maybe really usefull if you need just a bit more detail, but your computer will burn up if you add a lot of detail using a subdivision step. There are many more use cases, but I think you get the idea.

So in conclusion:
voxel remesher → more fine grained control over level of detail you are working with
multiresolution modifier → ability to “go back in time” and sculpt on lower level of detail without loosing sculpting information on your higher level of detail

I really recommend watching these two youtube videos if you want to know more about the multiresolution modifier, they greatly helped me:

• Multires modifier with Sculpting tools is awesome!
• ONE BY ONE - Everything about the Multiresolution Modifier in Blender.

Now given that this is really new knowledge I encourage others to correct me If I wrote something that is wrong in any way, would really appreciate it

Thanks again for @NP5 and @FedPete for both of your previous answers on which I was able to expand with this one here!

[FedPete]

Well done explanation!