Fprime rendertime testing

[duke]

As much as I'd like to show an image, I can't. Anyway, it's an underwater scene, water takes up the top half, there are floaties (HV), moss (clip mapped polys), and rocks (displaced a bit). My results/rendertimes are as follows (rendertimes with approx % gains):

1:50:9
1:23:6 (19%) - baked AO
1:13:5 (9%) - no HV
0:33.5 (70%) - no water
0:12:9 (62%) - no rocks
0:09:6 (25%) - no moss

I found the AO and water really amazing. I thought baked AO would give me a better speed gain and never thought the water would be such a huge hit! Time to dive into the shader settings!

Also note that the things i've turned off are accumulative, so no HV also has baked AO, etc.

[duke]

Ok here's the water. Original scene used but with bakedAO/no HV. These results aren't accumulative and i'll update the original ones above because it was a silly idea.

water tests: (baked AO, no HV)
1:23:8
1:22:5 - no trans
1:21.3 - no spec
1:21:8 - no reflection
1:16:8 - no bump
1:22:2 - no displacement AND subD level changed from 24 to 2

[Sensei]

Isn't water transparent reflective and refractive thing? Reflection - another ray fired to scene, refraction - yet another ray.. That is 2.. And these rays must have calculated shadows - so at least two another rays fired per light (with area light probably even 16 times more).. And then mixed..

[duke]

Seeing as I can't edit my first post here are the revised times that aren't accumulative - they're pretty much the same except the rocks (I suck at math). Yeah Sensei, I'm not using area lights but with the water it is effectively rendering the entire scene a second time.

1:57:9
1:30:6 (18%) - baked AO
1:44:9 (8%) - no HV
0:43.4 (73%) - no water
1:08:9 (31%) - no rocks
1:29:0 (18%) - no moss

[duke]

What's so awesome about the new fprime is with the new cameras. I've just thrown in one to render out a spherical map for the water, let it sit for a while, saved it out and loaded it, and have shaved 60% off my rendertime.