View Previously Used LOD Triangle Settings for Uploaded Meshes
tracked
B
Brigantia Mole
Feature Request: View Previously Used LOD Triangle Settings for Uploaded Meshes
As a mesh creator, I frequently adjust the triangle limits/settings for each LOD level during upload in order to achieve the best balance between visual quality and reasonable land impact.
The challenge comes when a mesh later needs to be corrected or updated and reuploaded. Currently, there does not appear to be any way to review the LOD triangle settings that were successfully used in the original upload. Unless creators manually document every upload externally, the original working values are effectively lost.
This results in unnecessary trial-and-error reuploads, inconsistent optimization, and difficulty maintaining the same visual/LI balance across updated versions of a product.
Suggested enhancement:
Provide a way to inspect the previously used LOD settings for an uploaded mesh object.
One possible implementation:
* Right-click mesh object
* Advanced / Developer submenu
* “View Uploaded LOD Settings” or similar
The displayed information could include:
* High LOD triangle limit
* Medium LOD triangle limit
* Low LOD triangle limit
* Lowest LOD triangle limit
* Physics settings (if available)
* Upload optimization method used
This would greatly improve creator workflow efficiency, consistency, and asset maintenance, especially for creators managing large product catalogs.
Log In
animats Resident
Unfortunately, that info is not retained after a mesh upload. Here's the internal mesh format: https://wiki.secondlife.com/wiki/Mesh/Mesh_Asset_Format Interestingly, the creator's UUID is in the mesh asset, and, potentially, creators could be allowed to securely re-download their old mesh objects into glTF form. That would be useful for creators who want to clean up their old meshes. Only the creator/uploader would be allowed to do this, of course. (Risk: makes copybot-type code more widely available.)
Nobody is really happy with how mesh LODs work. The built-in LOD generator isn't great, the LOD accounting system encourages creators to ask it for much lower triangle counts than it can deliver with anything approaching acceptable quality, and there are no limits to prevent creators from asking it for more mesh reduction than it can provide.
There are better LOD mesh reduction algorithms. Simplygon, Unreal Engine 5, and Houdini all have good ones. If you have to ask how much they cost, you can't afford them. Available open source code is either too naive (generates bad LODs) or too brittle (crashes on imperfect geometry).
In addition, SL suffers from a tradition that textures and meshes are independent. A good LOD generator should be generating textures which express small details with textures and normals as it downsizes the mesh. Put in a house, and at the lower LODs, window and door indentations are just normals, not triangles. Houdini does that. SL could do this now that glTF upload is a thing, and meshes and textures can be uploaded together. But it doesn't.
It's worth revisiting this in the glTF era.
I'm interested in this because of my interest in larger draw distances. I'd like to be able to draw a region 250m away at Lowest LOD and have it look decent.
Andrey Kleshchev
animats Resident
> SL could do this now that glTF upload is a thing, and meshes and textures can be uploaded together.
That isn't accurate. The only thing gltf upload does is converts gltf model into internal format, one optimized for much faster transfer and selective lod downloads, same as with collada. All the logic is the same as it was before gltf.
Ikaros Alpha
animats Resident > A good LOD generator should be generating textures which express small details with textures and normals as it downsizes the mesh.
A designer showed me he always does that in Blender himself. "Bake higher to lower" (textures) and add normals for low rendering cost of details.
SL Feedback
marked this post as
tracked
SL Feedback
Issue tracked. We have no estimate when it may be implemented. Please see future updates here.
Wicked Nightfall
We need the moles to rise up and demand change from the Lindens!!
Bambi Bubbles
Example from Firestorm
Perhaps we could have a universal "advanced" icon that could be used in many places in the UI.. that was like, click this for advanced features. that could show this additional 'advanced mode' info.
Photo Viewer
View photos in a modal
Jelly Doll
The uploader is great .. for one singular use case.
High poly quad based models (and now knowing that, you too can cheese insane content in for very low Li)
Handy that's exactly the type of model we want in a real time environment.
So rather than adding ways to fudge the problem, shouldn't time and resources go to solving this properly?
Candide LeMay
May I suggest you lead the way and create custom LODs for your meshes as it's supposed to be done, instead of relying on the crutch that is the uploader's LOD generator?
S
Star Mole
Yes please! Will save a lot off resources in uploads.
Jelly Doll
Star Mole Presumably moles don't pay for uploads?