LOD's switching sooner than expected.

[Swobbly Minogue]

Hi.

I am getting confused over LOD's and when they switch from a higher level to a lower level. I thought there were 3 factors that went into determining this.

1 - camera distance from the object.

2 - The size of the object

3 - RenderVolumeLODFactor setting in the viewer.

If this is true then the following problem that I am seeing does not make sense.

I have uploaded exactly the same mesh twice. The first one, "Mesh A" was uploaded at the weekend. It had custom built LOD's and physics shape. The second one "Mesh B" was uploaded today had auto generated LOD's and physics.

Other than the LOD's and the physics, these meshes are identical in shape, same size bounding box, same number of poly's and verts etc. The LI is obviously different as the LOD's and physics have not been optimized on Mesh B.

So when I rez them in world and place Mesh A and Mesh B next to each other, perfectly aligned and zoom my camera in to 1m they look identical. Then I pull my camera back to about 3m and Mesh B suddenly drops to the next LOD (medium). Mesh A stays the same, no LOD drop. I keep panning out and get to 10m and Mesh B drop LOD again (Low) and mesh A makes it's first LOD drop (medium)

10m is when I would expect an LOD drop on an item of this size (0.5m x 0.5m x 2m)

So why is Mesh B dropping LOD levels at 3 meters and Mesh A isn't?

Why aren't these 2 meshes switching LOD's at the same time. They are the same size, being viewed at the same distance with the same viewer?

What other factors drive the drop in the LOD?

Many thanks.

 

[Kwakkelde Kwak]

We've (we = several forum dwellers) seen this before on several objects. What we could make of it was very weird, something with materials. When there are 3 or 4 materials assigned, the z value of the bounding box is ignored. Did you by any chance change those? Or did you change the rotation of the original object so the axes aren't the same? If not, maybe you can narrow this bug down some more..

[Swobbly Minogue]

Yes, Mesh B, the one that dropped at 3m did indeed have 3 materials mapped on. Mesh A only had 1 material.

This would definitely explain it. I didn't even think that materials would be relevant which is why I did not mention them in my OP.

So until this bug is fixed we should only use 1,2.5,6,7 or 8 materials?

Thanks Kwak :smileyhappy:

[Kwakkelde Kwak]

Strangely enough...that's how it currently looks yes. You could also rotate the object before exporting so it's 2x0.5x0.5 or 0.5x2x0.5. There's a jira on it which I haven't looked at for a while, if I find it I'll post the link.

[Swobbly Minogue]

Great, thanks.

I just striped the materials down to 1 on the mesh and imported it again and it is now rendering as expected. What an annoying bug!

[Kwakkelde Kwak]

You could use it to your advantage, the LI bahaviour on these bugged items is very strange because one axis is ignored (or bugged at least). I don't think it's too wise to rely on bugs too often though...Definately annoying yes.

[Drongle McMahon]

"There's a jira on it..."

https://jira.secondlife.com/browse/SH-2837

 

[Bunnie Mills]

I'm having the same problem - or a similar one - with some mesh items i'm working at.

Adding a single triangle to the model results in the LOD steps happen in much shorter distances than expected. Like i have to move the camera within 0.1 meters from the prim center to make LOD3 appear (this is with LOD factor setting=1). The size of this model is about 0.4x0.5x8.0 and it is expected to display with LOD3 until quite far away.

After reading this thread i can relate the issue with the fact that these meshes behave ok until they use only two materials... and behave bad after adding even a single triangle that uses a new material.

I tried to browse for the JIRA referenced here but it seems deleted or locked (SH-2837).