Gaia Clary

There is a huge misunderstanding about how fitted Mesh is supposed to work. And i believe that the name Fitted Mesh for the used method is one part of the misery. Actually i think that it should be named Approximated Mesh or so. The basic problem is that the Avatar Shape system in SL is based on a mixture between Skeletal Animation and Blend shapes (Morph Shapes). While the Mesh shape system in SL is only based on Skeletal Animation. To make this more precise: There are about 130 Appearance Sliders in the Avatar Appearance. About 30 of these sliders affect Mesh characters while the other 100 Appearance sliders only affect the SL Avatar. As a consequence any attempt to match a fitted Mesh to all Slider settings of the SL Avatar will somewhat deviate from the corresponding SL Avatar shape. Now there is another bad misconception, that is: The Weight maps for the default SL Avatar do not work nicely for fitted mesh creations. However Avastar does currently only provide the default weight maps (exactly as they are defined in the SL Viewer). There is no such definition for the fitted mesh weight maps, hence we can not provide any official data here! As a consequence creators need to weight map their own fitted meshes by hand. Of course you can use the default weight maps as starting point. But you must modify them to get decent fittings. So to make your pants work you need to tweek the weigfht maps. I found a couple of methods how you can actually do that without breaking the overall weighting of your character and how it animates, but i can tell you its not a cakewalk :matte-motes-impatient: Here is Lesson 1 of my SL Skeleton course (it includes a video) where i introduce the basics of SL character animation. There i try to explain how blend shapes and skeletal animation work in principle: http://blog.machinimatrix.org/lesson/sls-1/ Furthermore i am working on an alternative set of Weightmaps which might be a better starting point for fitted Mesh creations. Here is an image of where i am at the moment. This is a silhouette made for the most extreme shape that is: All slider values set to 100, except the knee angle slider (which i kept at 50): Fitted Avastar Shape: This is how a mesh clone of the SL Avatar will be shaped when it is rigged as fitted Mesh using my (not yet available) alternative set of weight maps Avastar Shape: This is how the Avastar character gets shaped in Blender. SL Shape: This is how the Avatar is shaped in SL (this is actually derived from a photo that i made inworld) What you can learn from this image: Avastar's Slider System acurately recreates the SL Appearance editor. So what you see in Blender/Avastar will be exactly what you get later in SL (provided you use the same shape in Blender and in SL) The Fitted Mesh shape deviates from the Avatar shape even with my alternative Weight Map set. Note: The Fitted Avastar shape gets totally broken when i try to use the default SL Weight maps! I have not included a picture of that here. I am not yet done with my alternative weight set. I know that i can be better at the arms and the lower legs. The hips are the most troublesome area. There i suspect i won't be able to create something that behaves similar to the default SL Avatar. For all other regions i think i am OK so far (but again: This is not a precise solution!) Finally: Here is a proof of concept video that uses the new Weight map set together with a fitted Mesh Avastar and a fitted mesh pants: BTW When you make tights then you always have trouble with poke through. There is no other general solution for that other than using slpha masking!

the first image is most probably due to a bug in the SL Importer. I can see these holes as well, but after i imported to SL all is well again. This issue happens for me only with fitted mesh. The second issue might be due to the viewer that you used for importing. If you used the fitted mesh character from the wiki and you used the SL Importer from the default viewer, then this issue should not happen. At least not when you also have used blender to create the collada file and if you have used the Exporter presets for Second Life Rigged Mesh...

Jake Koronikov wrote: Nobody took the challenge? :matte-motes-silly: nah... i am just a bit slow But i am working on it right now as well and i will need another couple of days for my answer on this :matte-motes-evil:

Jake Koronikov wrote: As a creator-point-of-view (and eventually customer-point-of-view), the fitted mesh concept just does not work. That is the only correct answer. No matter what kind of tools there are, the idea does not work. I do not mean to critisize Avastar or anyone making those tools, you have done a great work. I do not feel personaly criticised All of us hate fitted mesh in one or the other way. Anyways. Can you give me an obvious example (blend file) where rigging to the Colliiosn Volumes does not work ? I would like to look at that in detail. Maybe there is something we can do at the end ... thanks

So, after reading a bit (thanks to Bobbie Faulds's insights in this thread), chatting a bit, and experimenting a bit ... :matte-motes-grin: The new Mesh Avatars are not a replacement of the default Avatar, but they are just mesh characters like all other mesh characters made by other creators. Avastar is healthy and up to date as far as i can tell. I am not aware of any "new way to create mesh clothes". Please correct me if i missed any other new development. It would be nice if we had some mesh mannequins for making alternative clothes for the Linden Mesh Avatars. But on the other hand those mesh characters are not the coolest ones on the planet. So it may be an alternative to look at the character market and make mesh clothes for whichever brand seems convenient to you. As a sidenote: Since Avastar Release (1.3) we have a Conversion function which allows to convert a foreign SL compatible Rig to the Avastar Rig. So i think that we are now prepared to work with any Mesh creator's development kit as long as it is rigged to an SL compatible Skeleton (non human meshes should also work). I tested yesterday with Belleeza. Next to test is the Mesh Project. And if there are other creators who want their development meshes be tested with Avastar, or want to provide Template Blend files, just contact me. An IM or notecard in SL would be sufficient. cheers, Gaia

1.) Open the Modifier stack of your mesh, search the Armature modifier and enable Displaye Modifier in Edit mode Adjust Edit Cage to Modifier Result 2.) Rig and weight the pieces separately. When all animation (weighting) is working as you want it to be, then join the parts into one Object.

hi. Can you explain in more detail what the new mesh bodies are and where we can get their definition files ? Thanks

Sassy Romano wrote: In Blender, I also believe that you're going into the texture window and just doing "Export UV map" (or whatever the menu option is). You mean the option Export UV Layout in the UV Image Editor. The UV Layout is a human readable visualization of the UV Map using a 2 dimensional wireframe image. The UV Layout can be used as a guide for colorising your image textures within a 2D image editor but has no other relevance (as far as i know). Sorry if i sound a little bit picky here, but keeping the terms right is important to avoid confusion :matte-motes-wink:

Hi, Sanjaria; This sounds to me like you use the term UV Map for something else. You neither can save a UV Map separately from its mesh (with Blender). Nor can you upload UV Maps separately from meshes to SL. Can it be that you actually mean UV Texture, which is an image? However i am not sure what you did when you "got the UV by just selecting my mesh and then saved it." What exactly did you save here? And how did you do that? Some pictures of what you call UV Map and how this is misplaced in SL might be helpful.

I think this video matches nicely to your explanation: http://blog.machinimatrix.org/coffee-cup-tutorials/uv-mapping/

Here is a brief documentation for how to export collada files from Blender: http://www.blender.org/manual/data_system/files/import/collada.html Note the operator presets for Second Life!

Cathy Foil wrote: Awesome first video Gaia! I am really looking forward to the rest of the series. Hi, Cathy; Thank you for your answer and insight into your own travel into unknown spaces :matte-motes-smile: Regarding the series i have plans as follows: I have finished the 7 out of 8 lessons which cover what the course video introduces in much greater detail. This part is finished and ready available. It even contains a few references to other videos from my youtube channel which just fitted well to the lesson. But to be honest all these videos are public available from my channel. So you can find them there (in the Playlist) I will add more (practical) information to the lessons over time and depending on the feedback that i get from users i will improve the lessons if necessary. I also plan to cut the lessons into smaller pieces because some have grown too big already (this tends to happen often as soon as you dig a bit deeper into the details :matte-motes-wink: I originally did not plan to add a special lesson about fitted mesh to the course because i have already explained all basic information about how the skeleton realy functions in the existing lessons (lesson 6 and 7 if i recall correctly). But meanwhile i have put it on my todo list because fitted mesh actually has a lot of practical pitfalls as you mention as well. So i will add a practical lesson to the course some time next year. I am not sure if i can add more videos. The one that i have published in the course took me almost 4 months to make it (as i can not work full time on them). Maybe i will add some with less gimicks... When you change your knee angle instead of just letting them be children to the "m" or Normal Bones LL decided they should rotate so the axis of the leg Collision Bones stay basically straight up and down. Yes, the collision volumes have very odd scaling. However, the main issue is that the SL Avatar has been bound to the rig in a certain bind position. This is the main reason for the difference between the free Avatar used in Blender (namely in avatar.blend and the Avatar workbench ) and what you see in SL (see in the course video). Our main issue with Blender is that it only supports binding to the Rest Position which is a realy annoying limitation. Honestly LL should not have used the original Collision Bones at all. A good number of them are at strange rotations and or scales not to mention some are offset a good distance from their "m" or Normal Bone counterpart which makes skinning harder since the initial binding process weights those vertices closest to the bone the most. The collision Volume Bones have never been meant to be used for weighting. The reason why their Joints are at offsets from their corresponding parent bones is due to how they actually control the primitive collision Volumes which in turn are used for calculating when the avatar bounces into other objects/Avatars (the octahedrons which you can see in the SL Viewer when you enable "Display Collision Volumes" in the Debug Menu) BTW the collision Volume Bones and all other bones of the SL Skeleton can be used for weighting because the skeletal animation technique allows a complete transformation matrix for each Joint (so that it can rotate, translate AND Scale) Luckily LL did not think about mesh 12 years ago. And that someone could want to add weights for arbitrary Bones of the Avatar. so they never added constraints to suppress such modifications. So it is NOT a bug that all of this is working. All of the standard "m" bones or what I call Normal Bones have zero rotations and scales all set to 1 in the standard T-Pose. This is pretty much an industry standard but some of original Collision Bones and some of the new ones like I said rotations and different scales. This just adds to the complexity of rigging Fitted Mesh unnecessary. I do not know about industry standards. All i know is there is a difference between Rest Pose and Bind Pose. But i agree it would have been great if the SL Avatar meshes and the Collision Volumes had been bound to the Skeleton in Rest Pose (with all joints set to unity) IMHO this would have avoided a LOT of issues. In my opinion LL, if LL were going to use just the bone scaling method, it would have been better to have created a new category of bones called Fitted Mesh Bones and not use any of the original Collision Bones. Another option would have been to put a bit of guesswork into how we could reuse the Mesh Morphing for custom meshes. That(!) would have solved all problems imho. Look at how nicely you can create attachments for the DAZ figures for example... To be perfectly honest my opinion is that the Mesh Deformer would have been a much better overall system than Fitted Mesh. Way less work for content creators and it did respond to the morphs and physics since it deformered worn mesh by by using the avatars body to change the shape. Even the facial and hand morphs worked. Imagine the custom avatars that could have been made using it. I am not completely sure if the Mesh Deformer is the solution for all issues. But i have seen that it could solve a few of the major ones (like facial animation for example). I have no idea why exactly the Deformer was dropped and Collision Volume weighting was made official. maybe it was because the SL Avatar was prepared for the latter anyways... Remind that we had collision Volume weighting (aka Liquid Mesh) for quiet a long time before LL decided to rename it to Fitted Mesh and add 5 extra bones to support that better.

Hi all; I have been working on a course about the SL Skeleton and how to create Mesh attachments. Here is Lesson 1 (of 8): http://blog.machinimatrix.org/lesson/sls-1/ I have tried to make the course as independent from any 3D tools as possible. However i added some practical tips for Blender users. The Lesson 1 also contains the course video. The course is an ongoing work and i plan to add more information over time depending on the feedback that i get :matte-motes-sunglasses-1: merry christmas

ChinRey wrote: Flat (or sharp) normals doesn't necessarily result in higher download weight than smooth ones. In many cases smooth normals gives considerably higher weight than flat ones. Can you give an example for this ? All i learned about normals indicates that flat shaded faces need 3 times more coordinates (one verterx normal for each vertex of the flat face) which would reslt in a raise of LI.

an increase of the download costs by a factor of 3 points to flat shading for all faces. Sorry i do not know how it is called in Maya. But the point is: When every face has its vertex normals set parallel to its face normal, then the number of "vertices" as counted by the SL Importer tripples -> this could be the reason for the increase of LI... But i am just guessing :matte-motes-sunglasses-3: There may be other reasons for this...

Amoralie Triellis wrote: Which avatar base should I import into blender to mesh clothing off of? I would recommend to take the Avatar from our Workbench. This character has following benefits: It is prepared for export to Second Life (it has all necessary bones) It is based on the Viewer's Avatar definition files. The newest version of the workbench also contains meshes which are prepared for being useable with Fitted Mesh. There is also a set of work files available from Linden Lab. You might prefer to use these as they are "official". More info is on the official Second Life Fitted Mesh page. I am curious about 'standard sizes' and if this is something I have to consider, and if avastar / workbench offer support for this. You can organise your projects easily by using multiple armatures besides each other, or rig meshes for different sizes to the same armature. This can all be done with Blender. Neither Avastar nor the Workbench offer any direct support for this. Does rigged mesh make sizing obsolete? Fitted Mesh allows users(customers) to modify the shape of mesh attachments to some extend. So in the best of all cases you might be able to get away with just one mesh atachment that "fits all purposes". However i believe that in most cases you still get better results when you create 2 or more variations of your mesh depending on its complexity. Somehow fitted mesh has made the situation better but not good (in my opinion). However no general rule seems to apply here.

ahimsa Balut wrote: ...And there are no other tutorials that teach how to do things, i am still learning. why do users tend to begin with the most complex projects ?

Maybe you misunderstood my post then. I meant that you should begin by making the current pose (the one in your picture) the rest pose and then go ahead. in the last steps you revert the pose back to T-Pose, apply the pose to your mesh, and finally you rig your mesh to the SL standard T-Pose. Does that make more sense now ?

Here are a few chat groups (mostly for blender) from where you could get more opinions and some insight into the tool(s): "Blender Mesh development" (1000 Members) "Blender " (1350 members) "Mesh Dev" (1500 members) "blender users" (1900 members) "Mesh Volunteers" (2600 members) "blender avastar" (4200 members) I apologize but i do not have many chat groups for other tools in my list. Maybe other creators can add suggestions for chat groups here as well ?

You could rig this poncho to the current Pose, then bend the rigged avatar into T-Pose and finally bind it again to T-Pose. That should work for normal stand poses at least. You will need to take special care for when the arms move sidewards. But it could work as well to some extend at least. Here is the workflow in a nutshell: The current pose (original mesh pose) is made the armature’s default pose. The mesh is parented to the new default pose The mesh gets weighted until it fits the needs The skeleton is bended into the SL T-Pose (and the mesh gets bent into T-Pose) The Armature Modifier gets applied (and mesh unbound from the armature) The armature’s default pose is set back to T-Pose The mesh is parented to T-Pose. When now the armature is posed into the original mesh pose then the mesh looks exactly like it looked when it was not yet parented.

Hi; This is an announcement for creators who need better support for importing rigs from other tools into Blender. Reasons for doing this are: Create assets for other creator's CharactersUse Blender's Animation system to create SL animationsI guess a few more reasons ...If you don't see a reason for an improved Blender importer, then the following is not for you. I have been asked to look into the Blender Importer to find possible improvements. And we ended up adding some import options to Blender which might be very helpful for some of you. I can imagine that especially character creators who provide Development rigs for attachment makers may be interested in this Blender improvement. The new options will officially become available with Blender 2.73, but you can get them already in the nightly blender builds starting today (2-dec-2014). New Collada importer options: Leaf bones are now better placed. I believe that the leaf bone orientation is now preserved as given in the Collada fileBut i have also added a "Fix leaf bone orientation" option, which can be helpful when the exporting program is a bit sloppy with its Joint definitions.And finally i added a bone chain finder which allows to find "most reasonable chains of bones" which will be automatically connected.What we have for 2.73 is documented in the Release_Notes/2.73/Import-ExportI would appreciate if creators would contact us (me) for feedback. cheers,Gaia

If you found a collada file that breaks blender's collada importer, then please report a bug to the blender bug tracker at: https://developer.blender.org/ Please add the .dae file that causes Blender to break. Your issue will certainly be inspected. If possible the problem will be solved.

Hi. Can you explain what you mean by "Apply physics to the mesh when uploading" ? Is this something you do in the SL Importer or in Blender before you export ? Also do you rig to fitted mesh bones (collision Volumes) or only to the classic mBones ?

valerie Inshan wrote: The LOD is more for sculpties than mesh. I always thought it is one of the major advantages of mesh that we can optimize the LOD, while for Sculpties the LOD's are predefined.

Is there a particular reason for why you want to bring them in to SL as separate Objects? Well, i could imagine it makes sense when you need one object as demo... But otherwise what benefit would you get ?