Page 3 of 4

Re: Mass amount of polygons

Posted: Thu Sep 18, 2014 7:59 am
by TBFX
wzxOctane wrote:So what the process with particules? You pop up particules on your ground mesh drive by a greyscale map, and then attach a tree instance on each one? It's as simple as that? :|
Pretty much, yes.
Check out this thread which has some more information. http://render.otoy.com/forum/viewtopic.php?f=28&t=37990

T.

Re: Mass amount of polygons

Posted: Thu Sep 18, 2014 8:04 am
by prodviz
Hi TBFX,

cool, when you get a free moment to test your old scene, let me know.

In the mean time, I'll try placing some of the original high poly trees up front, and then decimate a version of the trees and then test instancing these.

cheers.

Re: Mass amount of polygons

Posted: Thu Sep 18, 2014 8:08 am
by wzxOctane
Thanks, I forget to see it last time. Thanks for all the tips!

Re: Mass amount of polygons

Posted: Thu Sep 18, 2014 10:54 am
by prodviz
With just 1 tree of 800K the render fails on 30 instances (22GB RAM).
(On 50 instances the RAM is maxed at 32 and fails).

Can someone from the Octane render team comment on whether this is an issue with only having 32GB RAM… as the forest is pretty sparse at the mo ;) .

Cheers.

Re: Mass amount of polygons

Posted: Fri Sep 19, 2014 7:58 am
by JimStar
prodviz
What for Octane mesh types are set for trees geometry?

Re: Mass amount of polygons

Posted: Fri Sep 19, 2014 9:29 am
by TBFX
prodviz wrote:With just 1 tree of 800K the render fails on 30 instances (22GB RAM).
(On 50 instances the RAM is maxed at 32 and fails).

Can someone from the Octane render team comment on whether this is an issue with only having 32GB RAM… as the forest is pretty sparse at the mo ;) .

Cheers.
Please see JimsStar's post above, that is most likely your issue. Make sure the original mesh that is being instanced is set to scatter, movable proxy or reshapable proxy depending on if they are static, move or deform, scatter is best if the instances are not animated as then they will only load on the first frame of an animation. If this setting is left at global it will try to load all the trees as part of the global mesh.

I have tested my old scene and apart from what I expected to be broken, Diffuse shaders, env rotation, exposure etc. the scene loaded and started rendering just fine. In fact it took a little less memory, or at least peaked at 12 GB for a much shorter time, and the scene started rendering in just 3 min 30 sec. a full 1 - 2 mins faster than it used to. :)

T.

Re: Mass amount of polygons

Posted: Fri Sep 19, 2014 2:59 pm
by prodviz
Ah, I feel like a dumb ….

Set the original geometry to Scatter and now have a few hundred instances rendering no problem.

Cheers Jim and TBFX.

On another note, can you have animated instanced geometry?

Re: Mass amount of polygons

Posted: Fri Sep 19, 2014 9:48 pm
by JimStar
prodviz wrote:On another note, can you have animated instanced geometry?
I explained this many times across the forum, so to not repeat it one more time - here is the link to the series of my previous posts explaining how it works:
http://render.otoy.com/forum/viewtopic. ... 46#p199046

Re: Mass amount of polygons

Posted: Sat Sep 20, 2014 10:24 am
by prodviz
Cheers for the link.

With a scatter object, which is also moving (say a tree swaying in the wind), would this be best as a scatter object or a moveable proxy?

Re: Mass amount of polygons

Posted: Sat Sep 20, 2014 9:46 pm
by TBFX
prodviz wrote:With a scatter object, which is also moving (say a tree swaying in the wind), would this be best as a scatter object or a moveable proxy?
If the tree is deforming then it would need to be a reshapable proxy, technically this should work but I've never tested it. Of course what ever load time you have for the deforming instances will be required for every frame of an animation. Also depending on how many different animated trees you have and how much they move you may see patterns of movement in the results. I've wanted to test this for a while but definitely do not have the time right now, I'd be really interested in your results when you have some.

T.