Skip to content

Avatar Complexity

Ai Austin edited this page Aug 31, 2020 · 8 revisions

Information on avatar complexity and mesh rendering costs at:

Ruth2 v4 Draft 11 Avatar Complexity

Using Assembly Instructions as at 14-Aug-2020 and recommended LODs for Mesh uploader for each part.

2020-08-13-Ruth2-v4-Draft11-Avatar-Complexity

Experiments with main "Ruth2v4Draft9" (as at Draft11 10-Aug-2020)

  • All tests are with "Firestorm 6.4.5.60799 with OpenSimulator" mesh uploader using LOD: Default Default 100 100
  • Physics is either None selected or "Lowest".
  • NoP=No Root Prim; P=With 0.1m Cube Root Prim with 4x4white texture set 100% transparent
  • Alpha Mode is Blending or Alpha Mask with Cutoff 128

Avatar Complexity Experiment with Differences on Physics, Root Prim and Alpha Mode

No Texturing, No Root Prim, as Uploaded

Ruth2v4 (Body+Head+Hands) DefDef100100 Physics None Upload (as uploaded - no texturing, no root prim) 1,518

BoM Texture, Alpha Mask with Cutoff 128 Versions

Ruth2v4 (Body+Head+Hands) DefDef100100 PhysNone BoM Mask128 NoP 2,382

Ruth2v4 (Body+Head+Hands) DefDef100100 PhysNone BoM Mask128 NoP 2,719

Ruth2v4 (Body+Hands+Head) DefDef100100 PhysLowest BoM Mask128 P 2,725

BoM Texture, Alpha Blend Versions

Ruth2v4 (Body+Head+Hands) DefDef100100 PhysNone BoM Blend NoP 6,072

Ruth2v4 (Body+Head+Hands) DefDef100100 PhysNone BoM Blend P 6,409

Ruth2v4 (Body+Hands+Head) DefDef100100 PhysLowest BoM Blend P 6,433

Other Experiments

  • Use of a mesh cube (LOD: 12,12,12,12) versus a prim cube as root. Identical avatar complexity.
  • Use of a 0% root prim versus a 100% transparent root prim, lowers complexity a little (60 on Body+Hands+Head).
  • Uploading mesh with lower and lowest LO set to 0.. which actually gives a small number of triangles, does reduce the avatar complexity somewhat (1,962 versus 2,725 on Body+Hands+Head).

Conclusions:

  • Physics none or Physics lowest makes little difference.
  • Physics none, PhysicsCube PhysicsTriPlane earlier established as basically the same complexity.
  • Using a prim cube as root versus a simple mesh cube makes no difference.
  • Linking to a 0.1m 4x4white 100% transparent root cube makes little difference.
  • Alpha Mask Mode with Cutoff 128 adds about 1,000
  • Alpha Blend Mode roughly triples the overall complexity over Alpha Mask with Cutoff 128.