Jump to content

Uploading a linkset (Pt.2 - LOD's)


anselm Hexicola
 Share

You are about to reply to a thread that has been inactive for 3415 days.

Please take a moment to consider if this thread is worth bumping.

Recommended Posts

I have been trying to get a satisfactory result uploading a static mesh comprising 6 individual objects (drawn in Blender; total tri's = 1,300).

I have made 3 separate LOD versions (high-poly,low and very low) and a physics model.

...been trying out various permutations of the LOD's in the SL uploader - looking for an acceptable compromise between LI and visual detail with varying inworld camera zoom.

(In the end, high-poly for the High,Medium & Low fields and my very low-poly for the Lowest worked OK. My model is small, 1.2m x .5m x .5m, and LI came out at 6.)

SO .. 2 queries

    1. What I find frustrating is that not all members of the linkset change their LOD at the same time as the camera zooms in and out. Is there a way to prevent or even lessen this effect?

    2. I am a bit vague about this but I know that my own Viewer has a LOD setting of its own. Where can this be manually reset and what is a useful setting to represent that of most SL users?

 

... any advice much appreciated.

 

 

Link to comment
Share on other sites

hi :)

1: To get the LOD switching to occur at the same distance for each object in the linkset you would need to make the bounding box of each of those objects the same size.

 This could be done by adding geometry (a tri in diagonally apposite corners) and giving that extra geometry a separate material and then setting that material face to 100% transparent inworld.

 Or using one large triangle through your model to increase the size of the bounding box. Then selecting one of the vertices of that triangle and move it to the same position as one of the other vertices , but not merging it  so you still have 3 vertices but it looks like a single edge passing through your mesh object diagonally from one corner of the bounding box to another. This "triangle" does not need a separate material because it is so thin it will not be seen inworld.

Increasing the size of the bounding boxes will probably increase the LI .

You can see the Bounding box of an object in Blender by changing the Viewport shading type to Bounding box.

2: From the header at the top of your SL viewer, Advance menu > Debug settings > type in RendervolumeLODfactor   I think  the default is 1 or 1.25   What most people have it set to only LL will know. Mine is normally set to 1.25 some have it at 2.0 and others at 4.0

 

Link to comment
Share on other sites

Linked objects behave independently for LOD switching. As Aquila said, you can get close to coordinated switching by making all the objects have the same bounding box (technically more precisely, the same radius of the circumscribed sphere of the bounding box). However, if they are significantly different distances from the camera, they will still be a bit different. Also there are some "slop" and delay factors that might make small differences.

So the only perfect solution is to conbine your objects into a single mesh. Of course, you can only do this if you don't need them to be separate. The most common requirement for separate objects is that you need to move them separately, either for animation or for adjustments using inworld editing. Otherwise, you might need more than eight materials (unlikely for a small object). I'm sure there are other reasons too, but if you don't have to, I would recommend combining into a single mesh. Then the simultaneous LOD switch is guaranteed. For larger, spred-out onbjects, joining may cause increased LI, as a result of more distant LOD switch, but it sounds like that should not be the case here.

You are right that for an object this small, the LI is dominated by the lowest LOD geometry, rather than the intermediates LODs. If you want precise details of how this effect depends on the size of the object(s), you can find it here.

ETA: as for the "default" settings of RenderVolumeLODFactor, they are viewer dependent. For the LL viewer, you can find them in files called "low_graphics.xml", "mid_graphics.xml", "high_graphics.xml" and "ultra_graphics.xml". In Windows, these are in the "..\SecondLifeViewer\app_settings" folder  in the program files folder. The value for the viewer version I have currently are 1.125 for all except "ultra", where it is 2.0.

Link to comment
Share on other sites

Your post was very helpful. Here is how I got on :-

1. I liked the idea of the 2nd of your "added geometry" techniques, so that is the one I tried.

    As it happened, one of the cubes in my original physics model could provide a suitable trio of vertices to make that "almost-edge" triangle. Then I added it to some of the smaller objects that had been giving bother in my previous upload. Of course I had to this for 2 LOD's AND alter the physics model - so, a bit fiddly and time-consuming.

   Anyway, it all worked very well on upload to SL. The transitions between LOD's while zooming the camera were smoother and better. What's more ... the LI  remained at 6 because the extra geometry had only bumped it up by barely 0.2 - I was the right side of the rounding up/down watershed in this case.

2. After doing all that, I checked to see what my Viewer LOD setting was. It was 0.5 !!

     (Linden Lab obviously don't believe my machine is capable of much; I'll not disagree, but I moved the LOD up to 1.0 and THAT made a noticeable difference. Even my original upload behaved better with camera movements.)

... learned quite a lot - nice one, Aquila! 

Link to comment
Share on other sites

You are about to reply to a thread that has been inactive for 3415 days.

Please take a moment to consider if this thread is worth bumping.

Please sign in to comment

You will be able to leave a comment after signing in



Sign In Now
 Share

×
×
  • Create New...