- Tue Jul 15, 2008 12:19 pm
#275862
thanks guys,
To make things proper: The chandelier was not modeled by me but by Markus H. one of our freelance artistst. Credits for that. It clocks in at around 4Million polys and consists of roughly 9500 seperate objects.
It has been created as a centerpiece for an animation we did here but which I cannot show to you atm.
I did this heroshot actually for personal convenience and to bring MXW to the test. Of course it was also interesting to see how close MXW could get to the real thing.
The task was: Render 4 million polys, consisting of 10000 multifacetted frosted (!) glass parts with spectral coating in 4k resolution along with depth of field and 140 arealight sources together with multilight. It rendered overnight on 24 quadcores and reached SP level 17.5 .MXS was 995 MB and MXI was 1.6 GB large, all rendered in coop mode without a hassle. Rendertimes are huge of course but the task was insane as well, so I dont think it to be slow.
Btw. instancing was not really on option here since many instances are intersecting each others bounding boxes which decreased the benchmark roughly 10fold. It needed around 5.5 GB of RAM without instancing so everything was fine.
BB
(I try to go easy on the abberation next time Tom
)
To make things proper: The chandelier was not modeled by me but by Markus H. one of our freelance artistst. Credits for that. It clocks in at around 4Million polys and consists of roughly 9500 seperate objects.
It has been created as a centerpiece for an animation we did here but which I cannot show to you atm.
I did this heroshot actually for personal convenience and to bring MXW to the test. Of course it was also interesting to see how close MXW could get to the real thing.
The task was: Render 4 million polys, consisting of 10000 multifacetted frosted (!) glass parts with spectral coating in 4k resolution along with depth of field and 140 arealight sources together with multilight. It rendered overnight on 24 quadcores and reached SP level 17.5 .MXS was 995 MB and MXI was 1.6 GB large, all rendered in coop mode without a hassle. Rendertimes are huge of course but the task was insane as well, so I dont think it to be slow.
Btw. instancing was not really on option here since many instances are intersecting each others bounding boxes which decreased the benchmark roughly 10fold. It needed around 5.5 GB of RAM without instancing so everything was fine.
BB
(I try to go easy on the abberation next time Tom

Last edited by Tora_2097 on Tue Jul 15, 2008 4:24 pm, edited 1 time in total.