After last Friday meet up and in the impromptu SVVR-meeting yesterday, I noticed that I have alot of issues getting Playa to render with upto 10-20 avatars at the same time, when wearing the Vive. The scene now loaded pretty nicely, but with alot of avatars it just started to clearly have issues rendering them, as parts of avatars were simply never rendered (probably because alot of the avatars simply do not have LOD models), caused by my frames droping to 45 fps and the auto LOD kicking in.
Comparatively, I can keep quite stable 60 fps in Playa without the headset, but with it, I still begin to stutter, and have rendering issues as more Avatars come into the scene or when objects are loading.
Yes, I have a sub-minimum requirement GPU, a 780 TI, but it has still been overclocked (@ 1102 Mhz Core) to the match the performance of a 980 TI. If we would compare hardware specs when overclocked, its is as fast, if not faster than a 980 TI, except with the memory of a 970. I wouldnt be suprised if others that meet the min spec will have the same issues when using the HMD.
So here are my questions:
- What happened to the cardboardification of avatars we used to have last year?
- Auto Detail Decimation There needs to be away to set LOD levels for entities, and make avatars auto-decimate over distance instead of removing body parts (This seems like a partial, but unfortunate method, as most of the avatars seem to start with the highest density component: aka the face. I see quite a few headless people if the LOD adjustments kick in), eventually making them dissapear completely at range.
- Allow us to disable HMD mirror mode for the client (I feel like this isnt actually just mirroring, instead it feels like it is rendering it again, cutting my rendering rate like crazy)
One of the improvements that could be done is allow us to disable to mirror mode display