LW2Backburner Script, Backburner repeating frames. All caused by Nodal Motion

lardbros

Not so newbie member
Hey there,

I've spent the entire day figuring this one out, so thought, for the benefit of others, I'd post it here.

Using LW 11.6.1 on my renderfarm is pretty stable most of the time, but we've been encountering errors recently, with one specific scene.
I had an object with Nodal Motion applied, and it was using the raycast node to follow a surface. I've never tried rendering this on our farm before, but it seemed to cause this specific issue:

Frames would render, but as soon as the node had finished it's frames (5 of them) it would then error, and then start rendering the same 5 frames again. All of the nodes would do this same thing, continuously erroring (with a nonsense message) and then start at their first frame again.

I tried everything to fix it that I could think of. Baked all motions into .mdd's etc etc. I had baked the motion into this object, but forgotten to disable/delete the original nodal motion on it. Once I'd done this, everything renders fine.

Very strange, and I'm not sure if this has been sorted in later versions of LW, but something to watch out for! Had me stumped for hours!


It would be cool if there is anyone else out there using the LW2Backburner script, to test this out on their machines! Let me know how it goes :D

Cheers,
Tim
 
Last edited:

ernpchan

Active member
...but forgotten to disable/delete the original nodal motion on it. Once I'd done this, everything renders fine.

Very strange, and I'm not sure if this has been sorted in later versions of LW, but something to watch out for! Had me stumped for hours!

Do you have a dummy scene to try? It'd be interesting to see if this problem happens on a farm controlled by Rush. Then we could see if it's just a general LWSN problem.
 

lardbros

Not so newbie member
I'm really sorry, I don't have a demo scene available... and I use Amleto at home. I'll try setting something up quickly, and see if I get the same thing using Amleto. It could potentially be a LWSN issue?!? I'm not sure though.
 

lardbros

Not so newbie member
Okay Ernest!

I've done a few tests with the content scenes, and it does the same thing here... even up to the very latest version of LW.

Would you mind loading up the 'RayCast_Camshaft.lws' from the LW 11.6 content... and sending it off to your farm?

Make sure you set the rendering to batches of just one frame per node at a time. It seems to error on the bit where it's supposed to go to the next frame.
In batches of 5 frames, it renders all 5 fine and then errors when it's meant to go to the next batch (if that makes sense?). Basically, if you set it to 1 frame batches, it gets to the error problem faster, as it gets to the next batch (of 1 frame) faster. :S
Once it errors here, it then goes back and re-renders the same frame again, rather than continuing with the next one in line.

This appears to be caused by the Raycast node, and I can't see anything else that might cause it. It's definitely a common thing here.
I realise Raycasting should probably be baked before sending to a farm, but it seems to cope fine when not baked AND if:

(F) Frames in Anim
_________________ = (B) Batch of frames to send

(N) Nodes in Farm



Which means that each node has its batch of frames, which in total add up to the entire sequence of frames to be rendered, and the nodes never have to receive a new job once it's finished.


Makes me wonder if it's a backburner issue or something else.

Thanks for helping! :D

Tim
 

lardbros

Not so newbie member
Cool... Thanks a million for confirming!
Kind of glad it's not just our set up.

Will post this in the report system!
 

MSherak

New member
Okay Ernest!

I've done a few tests with the content scenes, and it does the same thing here... even up to the very latest version of LW.

Would you mind loading up the 'RayCast_Camshaft.lws' from the LW 11.6 content... and sending it off to your farm?

Make sure you set the rendering to batches of just one frame per node at a time. It seems to error on the bit where it's supposed to go to the next frame.
In batches of 5 frames, it renders all 5 fine and then errors when it's meant to go to the next batch (if that makes sense?). Basically, if you set it to 1 frame batches, it gets to the error problem faster, as it gets to the next batch (of 1 frame) faster. :S
Once it errors here, it then goes back and re-renders the same frame again, rather than continuing with the next one in line.

This appears to be caused by the Raycast node, and I can't see anything else that might cause it. It's definitely a common thing here.
I realise Raycasting should probably be baked before sending to a farm, but it seems to cope fine when not baked AND if:

(F) Frames in Anim
_________________ = (B) Batch of frames to send

(N) Nodes in Farm



Which means that each node has its batch of frames, which in total add up to the entire sequence of frames to be rendered, and the nodes never have to receive a new job once it's finished.


Makes me wonder if it's a backburner issue or something else.

Thanks for helping! :D

Tim

Reason this is erroring out is Raycast needs the last location to calculate the current if no keyframe is present. So doing a LWSN -3 where only one frame is rendered can cause an error since LWSN do not share information on frame position with each other. (LWSN -3 is what backburner uses) So in cases like this you have to send the whole scene to a single machine or bake it down to send it across multiple. It's the nature of network rendering. And in the case of backburner since it saw the error it try's to send it back out to fix it.
 
Top Bottom