When I render this scene it takes 13 hours to calculate lightcache, but when I render just trees it takes 12 minutes and if I render everything except trees it renders 3 minutes. And thats just Lightcache.
12 + 3 equals 15 minutes and it takes 13 hours to render it all together.
Rendering uses 1.43 GB of ram and I have 2 GB with 3GB switch.
Where is the problem.
Vray 1.5 RC2
kpalov,
I’ve had the same issue with one of my scenes. I ended up using all geometry in the scene and no proxies and it rendered at normal speed.
Anybody else having the same problem with Irr + LC?
kpalov, are you “showing calc phase” in your LC?
You can try increasing your dynamic memory limit and see if that helps.
The problem is most likely that when the GI rays fly between the trees and the buildings, most of the geometry needs to be flushed and reloaded all the time.
Best regards,
Vlado
are your trees purely geometry or are there alot of opacity maps being used?
Thanks Vlado, the memory trick helped me.
I think this is very important issue so everybody has to know about it.
Where were you when I had 800 Onyx trees at my last project. I had to wait 10 hours for Lightcache to calculate and like this it doesnt take more than 20 min. I thought the problem was 500 milion poligon scene but now I see its peace of cake for Vray.
What’s a safe value to raise the Dynamic Memory limit? Is it limited to system page, hard drive size, etc? What value ‘fixed’ your scene?
800
You think it should be set at this value by default? Is there any penalty for doing so?
Just let it be noted that using dynamic memory will in most cases cause your actual render to be about 50% longer.. i.e. a 10min render on static memory becomes a 15mins render on dynamic.
daforce, that depends on how much dynamic memory you give it…right?
Depends on the scene, but i think as soon as you switch to dynamic its going to take alot longer. Changing the memory limits can just make it longer still if you have them too low.
Did a few tests to confirm.
In a plain scene with about 50 poly trees, a vray plane and a direct light.
These are the rendertimes
Basic IR/LC setup
11m50s static
17m09s dynamic 400mb
17m15s dynamic 800mb
Its basically like proxying the whole scene.. if you proxy a tree for instance its just like choosing dynamic memory for that object. If you turn on dynamic memory if obviously affects the whole scene.
It will always be longer… but uses less memory overhead which is all it was designed for.
do a test, try dynamic 2000
oh and drop the face/level co-ef to .6
ok.. will test now with dynamic at 2000 and face co-eff at 0.6
ok 16m33s slightly quicker, but i would put that down to the change in teh co-eff.
So either way.. its about 50% slower 
cool, wait for 64bit max/vray and static is all you need
Thanks for the info Vlado, kpavlov and DaForce. I haven’t had time to test my scene again as I have had another job to complete.
chris.
Hi all, just a note, we can set the memeory value to a high value and maintain the geometry at static and it will speedup the scene (that contains proxies/fur/displace). Since the limit is for dynamic geometry like proxies/fur/displace. Only when setting the geometry also to dynamic will it slowdown since everthing is “converted” to dynamic geometry and it will do lots of loading/unloading of all the scene geometry.
Best regards,
Daniel
I didnt changed static to dinamic, I used static memory but I changet 400 mb as default to 800 mb in slot for dynamic memory limit and it solved the problem. Dont ask me why, it works.