SL Mesh Update Week 35

The Mesh User Group meeting is moving back to ADITI. The meeting has been in AGNI in the Mesh Sandboxes. Those will disappear tonight. So, a new meeting place is being selected  in ADITI.

Mesh Meeting

The majority of meeting participants feel ADITI is a better place for the discussion. I think the big selling point is the free mesh upload available in ADITI. Whatever the reason, Charlar decided to go with the majority and have the meeting in ADITI.

I have mixed feeling about the location. It is easier for new people to get to the meeting if it is held in AGNI. But, then we have to explain so much to new people and rehash things for them. Having an agenda really helps in that respect.

What the SL User Groups page for the new location.

Mesh Import Fail

Drongle McMahon found a way to crash the mesh upload. So, the Lindens have a new hole to patch up.

Drongle has been digging through various PE (Prim Equivalence) Cost scenarios. A mesh will change its physics shape to a convex hull when it is scaled down enough. The limit is unclear, somewhere around 0.5m.  If one links a prim and a mesh the prim will not change to a convex hull as it is scaled down. The problems come in when that causes explosive PE growth. Drongle was able to make a link set with more than 15,000 PE by linking a simple mesh cube to three 4-turn, hollow, profile cut tori, which then begins to create problems with it being returned or failing to rez with no warnings.

Mesh Materials

There is a forum thread where people are listing the material enhancements they would like the Lab to add. See: Materials Project Wish List.

Features in Process

There is a mesh phase two in progress. Everyone wants to know what’s in it. Charlar says it is small tweaks and fixes that they couldn’t get in before release. Vir Linden was a bit more specific listing; adding UI to expose additional accounting info, cleaning up the uploader dialog, and bug fixes.

Pivot Point control is something several people want. Pivot points are not going to be in the next release of fixes. There are changes that must be made on the server side before they can be implemented.

More information shows in the Uploaded and Build Editor if you turn on showAdvancedBuilderOptions. The setting is in Debug Settings. This will let you see more of the PE factors.

Tall Mesh Objects

Someone was having trouble uploading a mesh tree 240m tall… O.O

The error messages given by the uploader make no sense. But, I thought it would be intuitive that anything larger than 64m on any axis would be a problem. Whatever, the error message problems will eventually get fixed so they make sense. If you run into an error that makes no sense, file a JIRA.

I am confused by the size limit on mesh. One can make meshes larger than 64m and upload them. I’ll have to play with that one.

Attached Mesh Selection

Trying to select mesh attached to an avatar can be a problem. There is a JIRA about that somewhere. It will eventually get fixed.

Inconsistent Mesh PE on Upload

Several people have noticed that the PE of an initial upload is higher than the second upload. The Lindens are looking at the example given in the meeting. But, if this is true it is a really good reason to upload first on ADITI where it is free then upload in AGNI.

Small Objects with High PE

Some people are noticing that small version of a mesh are more costly than larger version of the object. This is an aspect of the Physics Weight cost. The weight goes up as a physics model is down scaled.

Second Life

Mesh Purse - Notice Lots of Shape and Detail - Blog - Store

Think of the physics engine trying to figure out which of a bunch of triangle an avatar is colliding with. With a large physics model the triangles are large and the engine needs to consider only a few close to the avatar. As the physics model gets smaller more triangles are in the space near the avatar. Think of a castle with physics model that allows you to walk through it. Normally the tower four stories up would not be in the collision calc. A nearby door, wall, and floor would be. Shrink the castle down to a meter tall and all the castle would need to be checked, floors, towers, all the walls, etc.

So, a complex physics shape can drive up PE very quickly as it scales down.

There is a technical article on mesh’s physics cost in the SL Wiki.


We are still learning about mesh and how different things affect PE. The Sojourner tutorials show how nice things can be build at low PE cost compared to prims. We are also finding things that cause the PE to grow explosively, like hooking twisted tori to a mesh.

I suspect the next few week will be a bit anti-climatic as mesh is polished.

The exciting developments for the next 3 or 4 weeks are going to have to do with Direct Delivery. Those Lindens don’t speak much with residents. So, that development is old Linden style; build, release, find out it is unusable for purpose, get yelled at by customers, then patch to some state of mediocre usability.

I like the newer Linden style where users can point out problems early on. Then we get things much more usable. It never comes out the way everyone will like, but it is much nicer. Mesh seems to have worked pretty well and in spite of what many say, is proving to be prim-cost effective.

Leave a Reply

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