Mesh Deformer 0.2 Update

The Mesh Deformer is still alpha software. However, it is now in its second iteration. I think it works better. Well… some of it does and some of it doesn’t. Until more viewers have Deformer 0.2 code I can’t tell if it is the viewer or the Deformer that hiccups.

Another gotcha at the max breast size setting

I downloaded Nirans Viewer because Niran let me know the new Deformer code was included in version 1.25, which is a Large Address Aware (LAA) viewer. I’m on a 32-bit system so I’m not sure how that may or may not affect things. Whatever, the viewer runs and the Deformer deforms.

Enable Deformer

In Nirans Viewer 1.25 the control for the Deformer is in Preferences->General->UI Settings. Look in the upper right of the panel for Enable Qarl Fizz’s Mesh Deformer.

Niran has reorganized the Preferences panels. Check boxes are used. I suppose that is better for those with some visual challenge. I get grumpy when things change because I have to learn numerous user interfaces. It takes some time for a UI to grow on me.

The Deformer

Look in the top menu in The Stuff under Mesh Related. You’ll find articles explaining how the Deformer 0.1 worked. Basically 0.1 deformed mesh by figuring out where the closest vertex in the avatar is in relation to a vertex in the mesh item. It then adjusted (deformed) the mesh item’s vertex based on the single vertex and the appearance settings. That created some odd deformations. I used the image below of bare feet to show what was happening in an exaggerated way.

SL Mesh Feet - Deformer 0.1

A quick trial using the same mesh feet with Deformer 0.2 produces a better result. However, it was much more awkward getting the deformer going. I put the feet on to check them. THEN I tried to turn on the Deformer. That did not work. My feet did not deform. I had to take the mesh off, have the deformer enabled, and then put the feet on. I don’t remember having to do that before. However, Qarl has added a ‘Deform or Not’ switch in the release. So, that may be causing the behavior.

Mesh Feet - Deformer 0.2

The Deformer 0.2 image shows that the previous deformation that made feet so bizarre is much better. But, there are oddities with Deformer 0.2.

When you put on a mesh item, the viewer freezes for a few seconds as the Deformer analyzes the mesh item then you will see the mesh item rez then re-rez into the deformed shape. The new 0.2 is doing more math.

This brings up a problem. When mesh was in beta several people were hoping to be able to animate mesh by swapping mesh the way we swap sculpty maps to animate sculpties. The Lindens would not allow it because of the load it throws on the servers. There are some other issues too but my memory is fuzzy one those. So, does this re-rez throw load on the servers? I don’t know. Someday I’ll experiment with it on OSGrid and may be find out. But, if it does this could be a problem for acceptance by the Lab.

In the 0.2 version the middle frame shows what happened when I adjusted my foot from size 0 to 100. The toenails changed size, the foot didn’t. I took the foot off. Left the size at 100 and put it back on. When I did the food rezied and deformed to what you see in the left frame.

In the 0.1 image the left frame is the same foot at size 100. The comparison between the 0.1 and 0.2 images shows the improvement of the deform of the foot at size 100.

The right frame of the images shows the foot at size zero. There is improvement there too.

Mesh Top with Deformer 0.2

I tried the mesh top again. In the previous tests as I adjusted size to the maximums the mesh got bumpy or crumpled. There are still problems with oddities as size is increased and there is still crumpling at the minimums. I doubt this is solely the fault of the Deformer. It could be in large part how the mesh is made.

The mesh does seem to deform more nicely at moderate appearance settings. Values around the norm seem to work best. In the image above I have take my breast size to zero. It is a little hard to tell what is happening. The right side (image left-right) shows a yellow line on the crumple. The left breast has the same crumple but the agle and the texture effectively hide it.

The left breast crumples too.

The two images above use the same breast size settings. Both breasts are crumpled the same. The texture and angles hide some of the distortion in photos. The crumpling is more obvious in-world where things are moving.

The feet show a good improvement. The breasts still have odd problems as the maximum and minimum shape settings are reached. But, in general the deformation around the breasts is smoother.

I doubt the Deformer will get much better in regard to the extreme cases. There will have to be some compromise between what we do and what we expect the SL system to handle. We may still see some need for clothes sizes.  But, creative types may find a good way around that problem.

OK, these are way too big... and things are going wrong.

I doubt that we will have to make our shape conform to some standard. I think this improved Deformer version shows that this process works. Obviously there are ways to make deformers that work because they are in various games. But, we may have a challenge with the SL Avatar. The mesh layout of the SL Avatar is not ideal for what we want now. When the Lab may consider changing the avatar is up in the air. It is a controversial issue because changing the avatar is likely to break a large amount of SL content.

Summary

At least we have a pretty good deformer in the works. Over the next few days we’ll see some of the other third party viewers up date to the 0.2 code. Then we’ll be able to tell what is implementation and whats inherent in Deformer.

Thanks to Niran for quickly getting a viewer out so we can see what the new Deformer code does.

8 thoughts on “Mesh Deformer 0.2 Update

  1. Pingback: Parametric deformer: Qarl updates (2) | Living in the Modem World

  2. Pingback: #SL Mesh Deformer Project Update Wk16

  3. Pingback: Second Life Mesh Clothing Tutorial

  4. Pingback: The Great #SL Deformer Debate

Leave a Reply to Henri Beauchamp Cancel reply

Your email address will not be published. Required fields are marked *