UI change for the Armature Modifier #93374

New Issue

Maxime Casas · 2021-11-25T14:10:43+01:00

Maxime Casas commented

2021-11-25 14:10:43 +01:00

This task proposes a small update of the Armature modifier's UI.

The issue the UI currently suffers from is that it is unclear that checking "Preserve Volume" actually changes the deformation method, instead of just enabling an additional option in the modifier.
This is addressed by replacing the checkbox with a dropdown, containing the two available deformation methods: "Linear Blending" and "Dual Quaternion".

On the contrary, a similar method cannot be used for the "binding" checkboxes, since those two options are apparently not mutually exclusive.

Mockup:

Tooltips:

Linear Blending: Deforms the mesh without preserving volume. Prone to candy-wrapping artifacts, but does not suffer from bulging issues.
Dual Quaternions: Attempts to preserve the volume when deforming the mesh. Generates bulging issues around joints, but does not have any candiwrapping issues.

This task proposes a small update of the Armature modifier's UI. The issue the UI currently suffers from is that it is unclear that checking "Preserve Volume" actually changes the deformation method, instead of just enabling an additional option in the modifier. This is addressed by replacing the checkbox with a dropdown, containing the two available deformation methods: "Linear Blending" and "Dual Quaternion". On the contrary, a similar method cannot be used for the "binding" checkboxes, since those two options are apparently not mutually exclusive. Mockup: ![mockup.png](https://archive.blender.org/developer/F12138465/mockup.png) Tooltips: - Linear Blending: Deforms the mesh without preserving volume. Prone to candy-wrapping artifacts, but does not suffer from bulging issues. - Dual Quaternions: Attempts to preserve the volume when deforming the mesh. Generates bulging issues around joints, but does not have any candiwrapping issues.

Maxime Casas self-assigned this 2021-11-25 14:10:43 +01:00

Maxime Casas commented

2021-11-25 14:10:43 +01:00

Added subscriber: @troopy28

Juan Pablo Bouza commented

2021-11-25 14:16:47 +01:00

Added subscriber: @jpbouza-4

Juan Pablo Bouza commented

2021-11-25 14:16:47 +01:00

Well, as long as you make explicit what each method does in the tooltip, it's ok. I think it's important for users to be able to know what dual quaternion does, so just write something about preserving volume there.

Maxime Casas commented

2021-11-25 16:47:37 +01:00

In #93374#1259936, @jpbouza-4 wrote:
Well, as long as you make explicit what each method does in the tooltip, it's ok. I think it's important for users to be able to know what dual quaternion does, so just write something about preserving volume there.

Oh indeed, I didn't include the tooltips in the task. I've put them in the Python mockup but forgot to put them here. Updating the task now.

> In #93374#1259936, @jpbouza-4 wrote: > Well, as long as you make explicit what each method does in the tooltip, it's ok. I think it's important for users to be able to know what dual quaternion does, so just write something about preserving volume there. Oh indeed, I didn't include the tooltips in the task. I've put them in the Python mockup but forgot to put them here. Updating the task now.

Sybren A. Stüvel commented

2021-11-26 14:56:57 +01:00

Added subscriber: @dr.sybren

Sybren A. Stüvel commented

2021-11-26 14:56:57 +01:00

This was discussed in yesterday's Animation & Rigging module meeting.
Although this is not described in this task's description, it was assumed that the future anti-bulging feature could become a 3rd option in the drop-down. That would make the drop-down worth the extra effort (in terms of number of clicks required to make a choice) it introduces.

The proposal:

Rename “Dual Quaternion” back to “Preserve Volume”
Keep the drop-down, to be able to add “Anti Bulging” option later.
The fact that both Preseve Volume and Anti Bulging use Dual Quaterions should be mentioned in the tooltip, but not as the main choice for artists too make. It would be too unfamiliar for many.
The tooltips should explicitly mention “Dual Quaternions” and not just “Quaternions” (as they do currently), so that it’s clearer which exact algorithm is used.

This was discussed in [yesterday's Animation & Rigging module meeting](https://devtalk.blender.org/t/2021-11-25-animation-rigging-module-meeting/21536). Although this is not described in this task's description, it was assumed that the [future anti-bulging feature](https://devtalk.blender.org/t/2021-11-11-animation-rigging-module-meeting/21157) could become a 3rd option in the drop-down. That would make the drop-down worth the extra effort (in terms of number of clicks required to make a choice) it introduces. The proposal: - Rename “Dual Quaternion” back to “Preserve Volume” - Keep the drop-down, to be able to add “Anti Bulging” option later. - The fact that both Preseve Volume and Anti Bulging use Dual Quaterions should be mentioned in the tooltip, but not as the main choice for artists too make. It would be too unfamiliar for many. - The tooltips should explicitly mention “Dual Quaternions” and not just “Quaternions” (as they do currently), so that it’s clearer which exact algorithm is used.

Maxime Casas commented

2022-01-18 11:23:20 +01:00

I agree with everything proposed, except for having the "Anti Bulging" option as an item of the dropdown (I would still keep the dropdown, even for two items only, to make it clear that two different skinning methods are available). Instead, I think the Anti Bulging should be a separate checkbox.
I think this is a better idea for three reasons:

The Anti Bulging option is not a skinning method, it is only a post-processing step that can happen after any skinning method. For now it is made available in Dual Quaternion (exposed as "Preserve Volume") only because the Linear Blending method does not introduce bulging. This also means that if/when blender will have some newer skinning algorithms (potentially introducing some degree of bulging too) the option would still be available for them.
The idea behind having the dropdown for the skinning methods is to show different possibilities for the same step performed by the armature modifier: the skinning itself. In contrast, having a checkbox for the Anti Bulging option makes it clear that it only enables/disables an additional step, performed separately from skinning.
The Anti Bulging option will have a vertex group option. When fed an existing empty group, it will have zero effect on the final deformation. This would mean that two options of the dropdown could lead to the same result, despite being displayed as separate, "different" methods. I think this would make things confusing.

I agree with everything proposed, except for having the "Anti Bulging" option as an item of the dropdown (I would still keep the dropdown, even for two items only, to make it clear that two different skinning methods are available). Instead, I think the Anti Bulging should be a separate checkbox. I think this is a better idea for three reasons: * The Anti Bulging option is *not* a skinning method, it is only a post-processing step that can happen after any skinning method. For now it is made available in Dual Quaternion (exposed as "Preserve Volume") only because the Linear Blending method does not introduce bulging. This also means that if/when blender will have some newer skinning algorithms (potentially introducing some degree of bulging too) the option would still be available for them. * The idea behind having the dropdown for the skinning methods is to show different *possibilities* for the same step performed by the armature modifier: the skinning itself. In contrast, having a checkbox for the Anti Bulging option makes it clear that it only enables/disables an additional step, performed separately from skinning. * The Anti Bulging option will have a vertex group option. When fed an existing empty group, it will have zero effect on the final deformation. This would mean that two options of the dropdown could lead to the same result, despite being displayed as separate, "different" methods. I think this would make things confusing.

Sybren A. Stüvel commented

2022-02-07 11:25:58 +01:00

In #93374#1289895, @troopy28 wrote:

The Anti Bulging option is not a skinning method, it is only a post-processing step that can happen after any skinning method.

Ah, that I didn't realise. It then makes sense to have it as a separate checkbox, indeed. And all the other reasons you mention, I also agree with :)

> In #93374#1289895, @troopy28 wrote: > * The Anti Bulging option is *not* a skinning method, it is only a post-processing step that can happen after any skinning method. Ah, that I didn't realise. It then makes sense to have it as a separate checkbox, indeed. And all the other reasons you mention, I also agree with :)

Brad Clark commented

2022-02-07 17:08:20 +01:00

Added subscriber: @BClark

Brad Clark commented

2022-02-07 17:08:20 +01:00

"The Anti Bulging option is not a skinning method, it is only a post-processing step that can happen after any skinning method. For now it is made available in Dual Quaternion (exposed as "Preserve Volume") only because the Linear Blending method does not introduce bulging. This also means that if/when blender will have some newer skinning algorithms (potentially introducing some degree of bulging too) the option would still be available for them."

In other systems it is a different interpolation than linear skinning, so do you mean by saying post-process it isn't how the skinning/vertex groups get assigned, like heat map etc.. or are you saying that it is always linear skinning and then the Dual-quat= preserve volume checkbox does another layer of processing on top of the linear skinning?

"The Anti Bulging option is not a skinning method, it is only a post-processing step that can happen after any skinning method. For now it is made available in Dual Quaternion (exposed as "Preserve Volume") only because the Linear Blending method does not introduce bulging. This also means that if/when blender will have some newer skinning algorithms (potentially introducing some degree of bulging too) the option would still be available for them." In other systems it is a different interpolation than linear skinning, so do you mean by saying post-process it isn't how the skinning/vertex groups get assigned, like heat map etc.. or are you saying that it is always linear skinning and then the Dual-quat= preserve volume checkbox does another layer of processing on top of the linear skinning?

Maxime Casas commented

2022-02-07 19:51:51 +01:00

In #93374#1303270, @BClark wrote:
"The Anti Bulging option is not a skinning method, it is only a post-processing step that can happen after any skinning method. For now it is made available in Dual Quaternion (exposed as "Preserve Volume") only because the Linear Blending method does not introduce bulging. This also means that if/when blender will have some newer skinning algorithms (potentially introducing some degree of bulging too) the option would still be available for them."

In other systems it is a different interpolation than linear skinning, so do you mean by saying post-process it isn't how the skinning/vertex groups get assigned, like heat map etc.. or are you saying that it is always linear skinning and then the Dual-quat= preserve volume checkbox does another layer of processing on top of the linear skinning?

The armature modifier is not responsible for providing the weight groups. It works with the groups the vertices already have. How those groups are computed is completely separate from this.

Blender currently has two ways to do the deformations: linear blending, and an option called "preserve volume", which is the dual quaternions method. These two methods are not changed by what is being proposed here. What is changed regarding them is that the choice between the two methods will now be made using a dropdown, to clearly tell the user that using Linear Blending or Preserve Volume is doing a different operation. Currently it is a checkbox, which does not make this clear.

The "anti bulging" option is something that will be performed by the armature modifier immediately after having applied any of the two deformation methods. This option simply checks, for each vertex, if its distance to its "main bone" (in weights) has grown between the current pose and the rest pose. If it has grown, then the vertex gets reprojected on the bone in order to make it keep the same distance, thus removing the bulging artifacts.
This option only needs, for each vertex, to know the original distance (at rest pose) and the new one (current pose). It can therefore be applied after any deformation method, be it linear blending or dual quaternion. The thing is the linear blending method does not produce bulging at all, so it is useless to enable it when using it. For this reason, the option will not be available (grayed out) when the user is using linear blending.

To sum up how the modifier will work:

Step 1: Deform using the bones - either with linear blending or "preserve volume" (dual quaternions)
Step 2: If bulging reduction is enabled, compare if a vertex has "bulged" compared to its position in the rest pose. If yes, reproject it on the bone to reduce bulging.

> In #93374#1303270, @BClark wrote: > "The Anti Bulging option is not a skinning method, it is only a post-processing step that can happen after any skinning method. For now it is made available in Dual Quaternion (exposed as "Preserve Volume") only because the Linear Blending method does not introduce bulging. This also means that if/when blender will have some newer skinning algorithms (potentially introducing some degree of bulging too) the option would still be available for them." > > In other systems it is a different interpolation than linear skinning, so do you mean by saying post-process it isn't how the skinning/vertex groups get assigned, like heat map etc.. or are you saying that it is always linear skinning and then the Dual-quat= preserve volume checkbox does another layer of processing on top of the linear skinning? The armature modifier is not responsible for providing the weight groups. It works with the groups the vertices already have. How those groups are computed is completely separate from this. Blender currently has two ways to do the deformations: linear blending, and an option called "preserve volume", which is the dual quaternions method. These two methods are not changed by what is being proposed here. What is changed regarding them is that the choice between the two methods will now be made using a dropdown, to clearly tell the user that using Linear Blending or Preserve Volume is doing a different operation. Currently it is a checkbox, which does not make this clear. The "anti bulging" option is something that will be performed by the armature modifier immediately after having applied any of the two deformation methods. This option simply checks, for each vertex, if its distance to its "main bone" (in weights) has grown between the current pose and the rest pose. If it has grown, then the vertex gets reprojected on the bone in order to make it keep the same distance, thus removing the bulging artifacts. This option only needs, for each vertex, to know the original distance (at rest pose) and the new one (current pose). It can therefore be applied after any deformation method, be it linear blending or dual quaternion. The thing is the linear blending method does not produce bulging at all, so it is useless to enable it when using it. For this reason, the option will not be available (grayed out) when the user is using linear blending. To sum up how the modifier will work: * Step 1: Deform using the bones - either with linear blending or "preserve volume" (dual quaternions) * Step 2: If bulging reduction is enabled, compare if a vertex has "bulged" compared to its position in the rest pose. If yes, reproject it on the bone to reduce bulging.

Brad Clark commented

2022-02-07 20:10:27 +01:00

Ah okay, thank you. and yes I agree that having the preserve volume as a dropdown instead of a check box makes it clear.

The ant-buldge seems like it should be a modifier like delta-mush for the armature so it can be more controlled ? in other software the bulge issue is a problem but users are allowed to mix or paint/blend the linear with the dual quat to fix this vs. having it try to automate it?

Either way, much more clear, thank you.

Ah okay, thank you. and yes I agree that having the preserve volume as a dropdown instead of a check box makes it clear. The ant-buldge seems like it should be a modifier like delta-mush for the armature so it can be more controlled ? in other software the bulge issue is a problem but users are allowed to mix or paint/blend the linear with the dual quat to fix this vs. having it try to automate it? Either way, much more clear, thank you.

Daniel Salazar commented

2022-02-07 20:11:45 +01:00

Added subscriber: @zanqdo

Daniel Salazar commented

2022-02-07 20:11:45 +01:00

Could this be a separate modifier/node instead?

Maxime Casas commented

2022-02-07 20:35:52 +01:00

Currently it is also possible to mix dual quaternion and linear blending, but it requires two armature modifiers. This requires going over all the vertices twice, which is slower. This is the main reason why I also decided to include this in the armature modifier directly instead of in another modifier. Currently it is done in a way that reduces as much as possible the amount of computations required to do the bulging reduction (vertices are iterated over only once, among other things).
The other issue with making it a separate modifier is that it would require a binding step (like for corrective smooth when using the option based on binding).

Finally, a separate bulging reduction modifier could only work if an armature modifier already exists in the stack. This is because it needs to know the armature, and it would not be a good workflow to be able to pick a different armature using a field (users would always pick their armature twice, it wouldn't be optimal). This system needs the bones used for the armature deformations to work, so it would always be the same armature, and if the user forgets to change the armature in one modifier and not the other, the results would be very strange.

I think simply adding a checkbox (and a vertex group) to manage this new option directly in the armature modifier is the best solution, at least for now, and if a clear use case for a separate modifier is found, then perhaps making it a separate node/modifier would be the best idea.

Currently it is also possible to mix dual quaternion and linear blending, but it requires two armature modifiers. This requires going over all the vertices twice, which is slower. This is the main reason why I also decided to include this in the armature modifier directly instead of in another modifier. Currently it is done in a way that reduces as much as possible the amount of computations required to do the bulging reduction (vertices are iterated over only once, among other things). The other issue with making it a separate modifier is that it would require a binding step (like for corrective smooth when using the option based on binding). Finally, a separate bulging reduction modifier could only work if an armature modifier already exists in the stack. This is because it needs to know the armature, and it would not be a good workflow to be able to pick a different armature using a field (users would always pick their armature twice, it wouldn't be optimal). This system needs the bones used for the armature deformations to work, so it would always be the same armature, and if the user forgets to change the armature in one modifier and not the other, the results would be very strange. I think simply adding a checkbox (and a vertex group) to manage this new option directly in the armature modifier is the best solution, at least for now, and if a clear use case for a separate modifier is found, then perhaps making it a separate node/modifier would be the best idea.

Brad Clark commented

2022-02-07 20:45:33 +01:00

Thank you.

Juan Pablo Bouza commented

2022-02-08 03:39:23 +01:00

Will this auto-bugling feature also solve the killer bulging problems that occur when you scale a rig that deforms with Dual Quaternion?

There was a discussion about this in the blender chat and someone showed that Maya has a Scale option for Dual Quaternion, similar to what we have now in the Corrective Smooth modifier.

Anyway, if anti-bulging also addressed this, it would be awesome.

Will this auto-bugling feature also solve the killer bulging problems that occur when you scale a rig that deforms with Dual Quaternion? There was a discussion about this in the blender chat and someone showed that Maya has a Scale option for Dual Quaternion, similar to what we have now in the Corrective Smooth modifier. Anyway, if anti-bulging also addressed this, it would be awesome.

Maxime Casas commented

2022-02-08 12:51:12 +01:00

I am not exactly sure about what you mean for the "killer issue". Could you please show some images, so that I investigate whether I can add it to the antibulging (unless it is already there) ?

The current implementation of antibulging allows to go from this:

To this:

By the way, note that even mixing linear blending and dual quaternion would not give similar results, since the second image does not suffer from candywrapping artifact we would get using linear blending. The following picture is using linear blending:

But I am afraid that when bones are scaled, the usual artifact that DQ creates cannot be easily fixed using the method being used here. It goes from:

to:

So in both cases it is incorrect. But this artifact isn't really "bulging", so I am not sure it is the issue you were thinking about when you said "killer bulging problems". Note that the images here show are rather extreme case, where the bone is rotated and then rolled on top of it (and also scaled for the last 2 images).

I am not exactly sure about what you mean for the "killer issue". Could you please show some images, so that I investigate whether I can add it to the antibulging (unless it is already there) ? The current implementation of antibulging allows to go from this: ![image.png](https://archive.blender.org/developer/F12855102/image.png) To this: ![image.png](https://archive.blender.org/developer/F12855105/image.png) By the way, note that even mixing linear blending and dual quaternion would not give similar results, since the second image does not suffer from candywrapping artifact we would get using linear blending. The following picture is using linear blending: ![image.png](https://archive.blender.org/developer/F12855107/image.png) But I am afraid that when bones are scaled, the usual artifact that DQ creates cannot be easily fixed using the method being used here. It goes from: ![image.png](https://archive.blender.org/developer/F12855114/image.png) to: ![image.png](https://archive.blender.org/developer/F12855116/image.png) So in both cases it is incorrect. But this artifact isn't really "bulging", so I am not sure it is the issue you were thinking about when you said "killer bulging problems". Note that the images here show are rather extreme case, where the bone is rotated and then rolled on top of it (and also scaled for the last 2 images).

Juan Pablo Bouza commented

2022-02-08 19:32:07 +01:00

Hi Maxime!

ok, so here's DQ with the rig in the normal scale:

And here it is with the root bone scaled to 0.05:

As you can see, when you scale the rig, DQ tends to keep joint deformation in its original size or something.... So if the rig scales down, the joints deform with huge bulging. The opposite happens if you scale the rig up.

You can only see the effect of this if you rotate the joints, if the rig is in it's rest pose, nothing bad happens.

So this is why Maya have the Scale option for DQ. Needless to say is that there are workaround for this... For example if you parent the whole rig and meshes to an empty and you use the empty to scale the character, then you bypass this issue. So It seems that this only happens within pose space calculations.

Hi Maxime! ok, so here's DQ with the rig in the normal scale: ![image.png](https://archive.blender.org/developer/F12855697/image.png) And here it is with the root bone scaled to 0.05: ![image.png](https://archive.blender.org/developer/F12855699/image.png) As you can see, when you scale the rig, DQ tends to keep joint deformation in its original size or something.... So if the rig scales down, the joints deform with huge bulging. The opposite happens if you scale the rig up. You can only see the effect of this if you rotate the joints, if the rig is in it's rest pose, nothing bad happens. So this is why Maya have the Scale option for DQ. Needless to say is that there are workaround for this... For example if you parent the whole rig and meshes to an empty and you use the empty to scale the character, then you bypass this issue. So It seems that this only happens within pose space calculations. ```

Maxime Casas commented

2022-02-08 22:24:07 +01:00

When I test with my own models (root scaled at 0.3), I get much uglier artifacts, like this:

And this is not simply bulging, the entire knee goes out of its expected position. The best this method can do for such artifacts is reducing the extent of the bulging, but the side of the back of the knee that went in the front will remain incorrect:

That said, when the artifact is only bulging (like near the upper thigh / posterior), then antibulging does a decent job at fixing the issue. DQ has inherent issues with scaled bones, and this is not what antibulging is trying to fix, but it looks like it can help there. It goes from:

To:

The left spike is something I'm working on getting rid of (without having to use an influence mask), but otherwise it would be easily solvable thanks to the vertex group available to control the influence of antibulging on the mesh.

When I test with my own models (root scaled at 0.3), I get much uglier artifacts, like this: ![image.png](https://archive.blender.org/developer/F12856121/image.png) And this is not simply bulging, the entire knee goes out of its expected position. The best this method can do for such artifacts is reducing the extent of the bulging, but the side of the back of the knee that went in the front will remain incorrect: ![image.png](https://archive.blender.org/developer/F12856134/image.png) That said, when the artifact is only bulging (like near the upper thigh / posterior), then antibulging does a decent job at fixing the issue. DQ has inherent issues with scaled bones, and this is not what antibulging is trying to fix, but it looks like it can help there. It goes from: ![image.png](https://archive.blender.org/developer/F12856147/image.png) To: ![image.png](https://archive.blender.org/developer/F12856149/image.png) The left spike is something I'm working on getting rid of (without having to use an influence mask), but otherwise it would be easily solvable thanks to the vertex group available to control the influence of antibulging on the mesh.

Juan Pablo Bouza commented

2022-02-09 01:22:51 +01:00

Will there be an "amount" option for anti-bulging? In case you less or more of the effect?

Brad Clark commented

2022-02-09 02:58:28 +01:00

https://developer.blender.org/T43188 related to the dual quat. scale problem and mentioned in the to do workbook
The Dual Quaternions algorithm behind the "Preserve Volume" function (Armature Modifier) needs an update (#43188) - [http://rodolphe-vaillant.fr/?e=59 Paper]

https://developer.blender.org/T43188 related to the dual quat. scale problem and mentioned in the to do workbook The Dual Quaternions algorithm behind the "Preserve Volume" function (Armature Modifier) needs an update (#43188) - [http://rodolphe-vaillant.fr/?e=59 Paper]

Brad Clark commented

2022-02-09 02:59:59 +01:00

In #93374#1304310, @jpbouza-4 wrote:
Will there be an "amount" option for anti-bulging? In case you less or more of the effect?

If we can use a vertex group to blend it yes.

Will it have that option?

> In #93374#1304310, @jpbouza-4 wrote: > Will there be an "amount" option for anti-bulging? In case you less or more of the effect? If we can use a vertex group to blend it yes. Will it have that option?

Maxime Casas commented

2022-02-09 11:03:06 +01:00

Hi!

In #93374#1304310, @jpbouza-4 wrote:
Will there be an "amount" option for anti-bulging? In case you less or more of the effect?

The current implementation indeed has a vertex group option. Do you think having a separate "amount" parameter on top of it would be useful, similarly to the MeshCache modifier?

In #93374#1304341, @BClark wrote:
https://developer.blender.org/T43188 related to the dual quat. scale problem and mentioned in the to do workbook
The Dual Quaternions algorithm behind the "Preserve Volume" function (Armature Modifier) needs an update (#43188) - [http://rodolphe-vaillant.fr/?e=59 Paper]

There is definitely some work to do in the implementation of DQ itself, but the paper you linked goes beyond "simple skinning" like is currently done in blender and actually performs simulations on top of it, making the results nondeterministic (you are not guaranteed to have exactly the same result every time you look at a given frame). The results look really nice, but I am a bit confused about how to read their performance table, because it does not seem to be real time (650ms per frame with a hand mesh of 37k vertices). Reworking blender's DQ implementation is something that will need to be discussed in depth, since there is a lot of super interesting research going on. There will be a lot of techniques to pick from :)

Meanwhile, antibulging provides a way to mitigate the current issues while having virtually no impact on playback speed (in my measurements). It also adds the possibility to remove bulging from any skinning method blender could have in the future and that might introduce some degree of bulging.

Hi! > In #93374#1304310, @jpbouza-4 wrote: > Will there be an "amount" option for anti-bulging? In case you less or more of the effect? The current implementation indeed has a vertex group option. Do you think having a separate "amount" parameter on top of it would be useful, similarly to the MeshCache modifier? > In #93374#1304341, @BClark wrote: > https://developer.blender.org/T43188 related to the dual quat. scale problem and mentioned in the to do workbook > The Dual Quaternions algorithm behind the "Preserve Volume" function (Armature Modifier) needs an update (#43188) - [http://rodolphe-vaillant.fr/?e=59 Paper] There is definitely some work to do in the implementation of DQ itself, but the paper you linked goes beyond "simple skinning" like is currently done in blender and actually performs simulations on top of it, making the results nondeterministic (you are not guaranteed to have exactly the same result every time you look at a given frame). The results look really nice, but I am a bit confused about how to read their performance table, because it does not seem to be real time (650ms per frame with a hand mesh of 37k vertices). Reworking blender's DQ implementation is something that will need to be discussed in depth, since there is a lot of super interesting research going on. There will be a lot of techniques to pick from :) Meanwhile, antibulging provides a way to mitigate the current issues while having virtually no impact on playback speed (in my measurements). It also adds the possibility to remove bulging from any skinning method blender could have in the future and that might introduce some degree of bulging.

Sybren A. Stüvel commented

2022-02-14 12:23:04 +01:00

Changed status from 'Needs Triage' to: 'Confirmed'

RedMser commented

2022-12-06 20:04:33 +01:00

Added subscriber: @RedMser

Philipp Oeser removed the

Download

What's New

Blender Studio

Manual

Developers Blog

Documentation

Benchmark

Blender Conference

Development Fund

One-time Donations

UI change for the Armature Modifier #93374