Hi,
Thank you all for answers. I tried and tested the scene. I used İtoo with Octane Proxy. Apparently the problem is Vram. With 980 ti I have 6 gb of Vram. I reach the limit at approximately 13-14 million polys. It looks low ,if 8.5 gb can handle 40 millions. But after 15 millions it become very unstable, sometimes works some times fails to render or even crashes.
I must say that 15 millions is not much, just some grass areas with 5-6 mid quality trees. I rendered the scene in several parts, to avoid crashes. I will post it later, to give an idea about grass trees objects amount.
Is the polycount affected by displacement, is it counted also as geometry? In the end I also removed all displacements.
Help on large and dense model strategies
Forum rules
Please post only in English in this subforum. For alternate language discussion please go here http://render.otoy.com/forum/viewforum.php?f=18
Please post only in English in this subforum. For alternate language discussion please go here http://render.otoy.com/forum/viewforum.php?f=18
tamerkseniya wrote:Hi,
Thank you all for answers. I tried and tested the scene. I used İtoo with Octane Proxy. Apparently the problem is Vram. With 980 ti I have 6 gb of Vram. I reach the limit at approximately 13-14 million polys. It looks low ,if 8.5 gb can handle 40 millions. But after 15 millions it become very unstable, sometimes works some times fails to render or even crashes.
I must say that 15 millions is not much, just some grass areas with 5-6 mid quality trees. I rendered the scene in several parts, to avoid crashes. I will post it later, to give an idea about grass trees objects amount.
Is the polycount affected by displacement, is it counted also as geometry? In the end I also removed all displacements.
here is an example where we have over 78000 trees in the area . This tree is 600000 polys. So it is over 46 billion of polygons of itoo forest trees and we are only using 0.2 mb of vram. So you can easily scatter 5-6 trees like this and have a nice forest and only use 1 gb of vram. The rest can be used for other things.
Make sure original trees etc are hidden as a hidden layer when rendering. Only itoo forest enabled. and i would not use octane proxy with itoo forest. just bring 5-6 trees to the scene, let itoo forest add them to custom objects and then hide those original trees as a layer
i would not recommend scattering trees with poly count over 1 mil.
- tamerkseniya
- Posts: 66
- Joined: Fri Apr 15, 2016 5:30 pm
This is the view, I was working on. It is around 13 million, several high poly trees, grass areas. I had to render buildings+closer grass, then further grass and then trees + surfaces, so in 3 times and then combine them.
Otherwise, when trying render all at once, it just says rendering but nothing comes to screen. I will try as coilbook suggested, scattering grass directly, not converting to octane proxy and see if it helps.
One more question, why when model totally converted to movable proxy, it is transferred much faster, then if I dont convert. Lets say I wait for 5 minutes without convertation, but if I select everything and convert to m.proxies, it loads in 2 minutes.
Otherwise, when trying render all at once, it just says rendering but nothing comes to screen. I will try as coilbook suggested, scattering grass directly, not converting to octane proxy and see if it helps.
One more question, why when model totally converted to movable proxy, it is transferred much faster, then if I dont convert. Lets say I wait for 5 minutes without convertation, but if I select everything and convert to m.proxies, it loads in 2 minutes.
2 x GTX 980 Ti, 32 GB ram, i7 5820K
tamerkseniya wrote:This is the view, I was working on. It is around 13 million, several high poly trees, grass areas. I had to render buildings+closer grass, then further grass and then trees + surfaces, so in 3 times and then combine them.
Otherwise, when trying render all at once, it just says rendering but nothing comes to screen. I will try as coilbook suggested, scattering grass directly, not converting to octane proxy and see if it helps.
One more question, why when model totally converted to movable proxy, it is transferred much faster, then if I dont convert. Lets say I wait for 5 minutes without convertation, but if I select everything and convert to m.proxies, it loads in 2 minutes.
looks nice! by the way you can scatter buildings, people like trees using itoo forest
when it is movable proxy it is loaded to vram once and then reused every frame without reloading .
- paride4331
- Posts: 3808
- Joined: Fri Sep 18, 2015 7:19 am
Hi tamerkseniya,
Because your movable proxy compiled file is smaller and therefore loads faster.
Regards
Paride
Because your movable proxy compiled file is smaller and therefore loads faster.
Regards
Paride
2 x Evga Titan X Hybrid / 3 x Evga RTX 2070 super Hybrid
- tamerkseniya
- Posts: 66
- Joined: Fri Apr 15, 2016 5:30 pm
In this case, it seems that the most logical is to convert everything to movable proxy by default for speeding things, from the very beginning. Is this correct way of working, is there any drawbacks of doing this?
2 x GTX 980 Ti, 32 GB ram, i7 5820K
- paride4331
- Posts: 3808
- Joined: Fri Sep 18, 2015 7:19 am
Hi tamerkseniya,
The right choice depends on whether you are making a still render or an animation, in the case of animation what moves and what is not on the scene.
In an extreme case:
If you have only a camera animation, putting all non-movable proxies, the first frame will take longer, but the subsequent frames will be faster to load.
If you have a still render, then all movable proxy can shorten a single frame time.
Regards
Paride
The right choice depends on whether you are making a still render or an animation, in the case of animation what moves and what is not on the scene.
In an extreme case:
If you have only a camera animation, putting all non-movable proxies, the first frame will take longer, but the subsequent frames will be faster to load.
If you have a still render, then all movable proxy can shorten a single frame time.
Regards
Paride
2 x Evga Titan X Hybrid / 3 x Evga RTX 2070 super Hybrid