#SL Mesh Deformer Discussion

Additional Base Shapes

When we were testing Deformer version 0.3, or say the just previous one, people ran into the problem of poor deformations as the shape sliders reached the minimum and maximum setting values. While we were looking at removing the need for clothes to be made in multiple meshes of different sizes, it became apparent that we would need various base shapes to start from to get good deformations for all sizes of people.

So rather than making multiple mesh copies in different sizes as we see now (think standard sizing), something different is going to happen. The current mesh standard of sizing clothing is somewhat inflexible. We can’t change the clothing PROPORTIONS as we change out shape proportions, which is the whole reason for the Mesh Deformer. We need a mesh item for each shape. The idea for the Deformer was to have one item and size that fits all.

But, with a single base shape we can only stretch the clothes so far before they start to look odd. One size is only going to fit most, not all sizes of avatar. Clothes deformed on big size avatars and small size avatars start to look bad. The deformation gets ugly and textures compress or stretch. But, if we use a larger base shape we do not have to stretch so far for large avatar shapes and things look better. The deformation is better and the textures are stretched less. The same is true for smaller shapes.

So, Karl, going beyond the scope of work, added in the ability to use multiple base shapes. Currently only the female and male base shapes are included. Karl has left it to others to extend this work and add more base shapes. Some Third Party Developers are already looking at doing that.

Hold Up

In the JIRA some think the holdup is not having the User Interface (UI) for the multiple base shapes. Multiple sources are pointing out that is not a holdup.  The UI is laid out and has two shapes, female and male. Extending it is relatively simple. If that is all that is available when it is release time, the Lab will likely release the feature.

However, testing the Deformer is a holdup. Oz says the people in the Lab that will be testing it are busy with other things just now. But, they will get to it. Remember. The Linden and user time scales are different. What we think is a long time is often Linden ASAP.

The real question is will there be enough clothes available for testing when the Lab’s staff becomes available? Only clothes uploaded with the new Deformer project viewer can be used for testing. Oz says only a couple of items have been provided. With a shortage of testing clothes, the Lab staff tends to put off testing. It makes sense to wait until there is a good variety of items before starting testing. So, the lack of test items is contributing to a psychological slow down…

Project Complete

For all practical purposes Karl is complete with his work and then some. If a bug is found, that will get handled and Karl will be available to help. The bug we know of now has a fix. Not getting the fix out is not holding anything up. Deformer testing just has to be done with the deferred render turned off… or is on? Whatever. Testing can proceed once we have things to test. So… say something to whoever you know that is making mesh clothes.

2 thoughts on “#SL Mesh Deformer Discussion

  1. Yeah, I been saying for almost a year now that creators have to tweak the weights. I know that it can be a huge pain in the butt, but it must be done to get good results. I’ve not made much clothing, but I did make my own mesh avatar. Basically, I must’ve redid the weights on my mesh avatar about 20 or more times. Once you have a good set of weights, you can drop those weights onto many of your clothing items with minimal reweighting. You can’t just use the default weights and think that is going to work on everything, especially uniquely stretching outfits.

    It’s also probably a good idea to add some verts in areas that will stretch the most. This will also help with the textures stretching too much in those areas. If I were LL, I would give a suggested total vert counts for clothing, as so many people have no idea what is reasonable.

    Although I was totally against Standard Sizing, I do like that Karl added in the ability to use a custom shape. I said from the beginning of the Standard Sizing fight, that the real solution was adding custom shapes. This could make it possible for most clothing to fit most sizes, even custom mesh avatars, as long as the custom shape is not imbedded in the mesh clothing.

    What would be super cool to see happen, would be for Blender to automatically reshape mesh clothing to whatever shape avatar you are using. I could very much see people asking for custom fit clothing, but I don’t see this as a viable option unless it is super easy for a creator to do quickly.

  2. Adding my agreement to the necessity to learn weight painting and make manual tweaks, or weight completely from scratch. Not only does copying the weights from the default avatar not usually give the most satisfactory results, but the Blender files most of us have aren’t actually 100% accurate when it comes to the weights on the avatar in world. (I use the fine ones provided by Gaia Clary on the Machinimatrix site.) These inaccuracies only get exaggerated further doing a bone weight copy.

    This is actually not too difficult to see if one imports the vanilla avatar from the one of these files in world in certain poses. For example, a person might put the avatar into a simple sitting pose, with the knees raised and level to the hips, and take a look at the buttocks. You can see that one of the buttocks is strangely flattened, while the other stays rounded. It’s probably only a single vertex that’s not weighted quite right, but it makes a noticeable difference.

    So further refinements to the files we use as a base might also help reduce cases of user error, although that’s not a substitute for doing the hard work yourself.

Leave a Reply

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