i was trying to render my animation with multiple rendernodes, but at the frame 172 of 350, this is what i get:
some rendernodes didn’t load the scene well maybe ? or there is something missing ? only this frame is like this… why ???
Either something is missing or there are some discrepancies in the version/service pack/plugins/scripts etc.
1.Please first make sure that all the machines have absolutely the same versions/service packs of V-Ray/3dsMax and any additional plugins/scripts etc.
Even small variations between versions or might cause difference in the buckets.
2.Make sure that all the files are accessible for the render host machines, Transfer Missing Assets usually transfer what’s missing but it’s still good idea to ensure that the files are reachable for the hosts.
3.If V-Ray Spawner on the host machine is started as a service, make sure that it uses Windows Account that has proper rights to access files, default Local System Account doesn’t have those rights.
4.Try to convert Bitmap nodes to VRayHDRI ones using the converter script that ships with V-Ray:
5.Check which machines render wrong and if they always render wrong or occasionally.
This happens to me too. It usually occours on the subsequent renders if I cancel a DBR job. The precautions Svetlozar mentions are all okay. I usually use static scene geometry, and I made sure there is enough memory set aside in the VrayHDR settings. The thing is I can work a whole day with rendering and rerendering, no problem. But if I cancel a DBR job at some point, it usually goes haywire after that. I have to restart all the spawners (not the machines itself), and then it works fine again …
yes to me it happens also when i add new node to render a animation.
when render of animation is running and i add any of extra node i get the same buckets. it is not map or anything 100 % check too many time. if i save the scene and open it again and start with all nodes from the very beginning then it works. just can’t add or cancel any during rendering. It looks weird because it appears as the bucket is having different exposure and higher noise.
all dr nodes have access to the assets and have the same version of max and the plugins.
When I check the render element “VRayDRBucket” I can see, that the node loses the Information during the rendering. The node begins correct, then after a few heavier regions comes the first corrupted buckets.
It helps to restart the spawner on all nodes, but by the next rendering a other node loses the GI information. I write Gi information, becouse it is mostly visible in the render element VrayglobalIllumination.
Forgive the whining, but at this moment , it is so pity, that DR using progressive mode is unusable. Yes the option theoreticaly exists, but no matter what number for “Ray bundle size” I type the load of the nodes jumps by a few %, only for a few seconds they reach the 100% utilisation.
When I start the same hi-res job using buckets, then the nodes runs all constantly by 100%. But then I get the buckets with the lost GI.
In my opinion, using DR with progressive mode, it would by much better, when the node would send their results only one time per minute, maybe with a small delay between the nodes, or only after a number of passes the DR node has finished.
I have written about the DR using progressive mode, because maybe when a single node produces a false solution, the other passes from the other nodes covers the failure and the over night rendered hi-res is still full usable at the next morning when you come to the office and your psychopatic client ist already on telephone and asks for renderings he could show his even more psychopatic client (investor).
Excuse my dark humor, but it is a bit the reality out there.
Ruined frames are a really problem here.
I think we still didnt find the fix. to me this is still occurring, now even after i restart everything. i have all the assets correctly tracked, nothing missing just this for some reason different buckets are happening.
I know it is suspicious that now it is on the Vrayfur, but this happens randomly throughout the scene.
In any file explorer window right click and select
Select the dropdown arrow on [Services and applications] on the left column
select
in the middle column scroll down to and right click on it and select
goto the tab and select the radio button and input a name/pass for a user that would have access to the files needed to render.
select and you should be good to go.
I run a unique user profile for the render slaves on my farms that has access to the proper folders on my networks.
yes that would. In my case what fixed it, was that i after render finished, i switched to selected mode and than there was no problem. But this is fix in still image, not animation.
This might be a bug, we got a scene whit specific setup with Override Material which randomly gives different buckets on the render hosts.
From the images posted so far I don’t think that it’s the same thing and would be a good idea to send us over some scenes for further investigation.
You are welcome to send the scene as well, more scenes will definitely give more information about the issue.
It’s a good idea to check however if all the software is identical between the machines since any discrepancy in the versions could cause that issue.