The Deformer project remains pretty much on hold. The JIRA thread STORM-1716 has turned in to a massive discussion where the uninformed and technically illiterate rant about what should be done. It is now a useless JIRA report too long to read with good technical information buried in thousands of words of useless rhetoric. The purpose that report best serves is to demonstrate why the JIRA was closed.
So… until Oz Linden can convince Linden management to assign people to the Deformer/Avatar project I see it remaining on hold and the JIRA item having devolved into a waste of time.
With In-Worldz deciding to adopt the Deformer, we are likely to see it added to viewers that support non-Second Life grids. But, for those viewers to be used in SL, they have to be classed as experimental viewers. That likely means little use by residents and lagging support. But, it might increase the Lab’s priority for the Deformer… might…
But, there is a need now and an annoying problem with Mesh Clothes.
State of Mesh
The result of the Deformer/Avatar project delay is we are using Mesh Clothes made for the standard shapes. But, even those do not fit well. Have you noticed how the butts and breasts of many Standard Size shapes seem bigger than they should be? I don’t mean the exaggerated shapes that are popular with many. What I mean is what I was talking about in: Second Life’s Annoyances and shown in the image above. Do, you think all the designers really are into those big butts?
The Standard Sizes are the work of SL residents Minnu Palen & Thora Charron (of LeLutka), Siddean Munro (of SLink), Elie & Anouk Spot (of Mon Tissu) and Jaden Celoe & Shay Sibrian (of Celoe). They did a great job and their efforts helped the community. But…
NOTE: You can get the shapes for free from the Market Place: Second Life Standard Sizes. These are the shapes you need to be, to wear clothes made for the Standard Sizes. Supposedly most avatar shapes fit into these basic size groups. Also, supposedly designers know how to use the 3D models of these shapes to make clothes. I strongly suspect many do not.
As I show in the image, clothes made for the standard shapes tend to exaggerate our curves. When I look at the 3D Standard Size models I don’t see those exaggerations.
I hate they way most Standard Size clothes shape the butt and breasts. I think something like 90% of of the mesh clothes made now have this problem. In the months since mesh clothes came out, I have found one pair of pants that I think don’t exaggerate my curves. Their medium size is close to my shape without the big butt.
There is a reason rigged mesh clothes don’t even come close to fitting right. I suspect it is because most clothes makers do not understand the SL processes that affects rigged mesh in SL. It is sort of a problem of SL being too helpful. The result is some huge percentage of mesh clothes are only supposedly made to fit the Standard Size. The makers are trying. But, what they don’t know is distorting the end result.
First one must understand the problem. Next one must find a method to compensate for the problem. Understanding is not that hard. Finding the compensation method is a bit of a problem, or has been for me.
To understand the problem check out Gaia Clary’s explanation. It is the clearest explanation I know about. See: Avastar vs. Workbench Skeleton.
The quick explanation is: No matter what shape you use to model your clothes, Second Life applies some of your shape settings to the resulting rigged mesh. So, if you use a custom or standard shape and take it into Blender or 3DS to model your clothes, the result will essentially be distorted once the model is taken back into SL. Thus the too big or too small butts and breasts.
So, if you increase butt size 10% in SL’s appearance, export the XML shape to Blender/3DS, model a skirt or pants, then export and import to SL… SL will apply a 10% appearance adjustment on top of the 10% you added in the modeling program, resulting in the butt being too big by 10%. So, you must adjust the shape in the opposite direction for certain settings by twice the appearance adjustment… confusing?… to make it come out correct after SL adds the adjustment back in… yes, it is confusing and a pain.
How many SL clothing designers know to do that? From my personal experience, not many.
You can get the Blender file with the Second Life 5 Standard Size models here. They are the models one needs for making standard size clothes in Blender, 3DS, or whatever modeling program. But, one still must compensate for the adjustment the SL system is going to apply to your rigged mesh.
I’ll admit this adjusting compensation was getting way too confusing to deal with. So, I decided to use Avastar and avoid having to deal with the problem. Gaia Clary and her friends and associates worked out the solution to the problem. They modified the Collada exporter to apply the compensation as part of Blender’s SL export process.
I like Avastar and I have used both the original pre-Avastar workbench files made by Gaia and company and the avatar files available from Linden Lab and others in the past. Unfortunately, the avatar situation is complicated as you have just read and those did not work as well as I had hoped. Gaia’s tutorial is the only tutorial explaining what is going on.
The tutorial explains the problem of using Blender alone to make clothes for avatars. With custom shapes, shapes exported from SL, we need to calculate the anti-custom-shape… which is no easy task. But, failing to do that means any close fitting clothes are NOT going to fit well when taken into SL. It is pretty much a trial and error process until you get it figured out.
There is also the problem of getting the correct shape into Blender or another program as a model for making clothes. Gaia does not mention the problems of moving SL shapes to Blender or other modeling programs. I covered that in: Second Life Shape Export. There is a work-around for getting a model correctly sized so it matches what you have in SL. Avastar also handles this problem nicely saving you a bunch of work.
Back to the Deformer
I like tight skirts and tops. So, I think I’ve tried about every work-around I can think of to get around the XML import to Blender problem and SL’s double adjustment issue. With the Mesh Deformer (STORM-1716) one could pretty much eliminate the problem.
Karl’s code compensates for the ‘double’ adjustment of the imported mesh. It essentially calculates the anti-default or anti-custom shape, whichever you select, so the result is what you intended. So, if you can get the precise custom shape into Blender, model to that shape, then the clothes will fit right once imported to SL, provided you use the Mesh Deformer viewer once in SL.
The problem I ran into was the failure of SL export and various Blender import programs to correctly size the avatar shape brought into Blender. That meant my work when taken into SL didn’t fit as well as I wanted, even using the Mesh Deformer. The Deformer helped a lot, but still did not give me what I wanted. I was going nuts trying to figure it out. Many think the problem is the Deformer not working well, but it does work well. People just do not understand the problem and thus do it wrong.
Now that I know there is a sizing problem going into Blender and that SL re-adjusts rigged mesh imported to SL, I can make tight fitting clothes that work with or without the Deformer. I am working on filing a JIRA (BUG-3499) feature request to get an option to turn on or off an ‘anti-default-shape’ calculation in the viewer’s importer as an interim fix. Putting that in a form the Lindens will understand and see the necessity of is a challenge.
While Gaia and associates have been working out solutions and building them into a Blender add-on, I’ve spent those months trying to find a work-around. One can do it. But…. dang it is painful. One has to learn a bunch and then do it manually without mistake for each rigged item made for SL. And if you need to add an animation… zOMG! …triple the complexity and problems.
Not all the problems are due to SL’s implementations. Blender has its own problems. But, the Lab could make things easier. I hope they adopt my feature request. I won’t be holding my breath.