tried to access null node

[jwfvr]

I am working on a file that forces me to use OOC - but I am getting the error - "tried to access null node" even even before I exceed my vram - it also seems related to the mesh that is made visible - I think - to avoid one big heavy lift, I start rendering with about 2/3rds of the scene visible, then enable additional items one at a time - I will enable a mesh, it triggers the error, it keeps rendering, I then enable another item that is even more dense and it does not trigger the error. if I load the scene completely, I get to about 2500mb in OOC and then at some point it just fully crashes modo instantly. (working on the same file in previous versions of Modo and Octane I could go even higher in OOC usage and while slow, it continued to work and not give this error) what does this error refer to exactly?

the other weird thing is if I enable (make visible) a mesh that triggers this error, I can turn it off, then turn it on again and it does not give the error.

thank you

amd 5950x
rtx 4090
rtx 3090 x2
128 ram
win 10
modo 16.1.3
octane 2022.1.1.190 Studio+

[face_off]

Hi

The general rule is that if a mesh starts off invisible, when you then make it visible part way through the animation it is going to cause issues. When exporting an animation to ORBX format, Octane is expecting the same number of vertices every frame in order to calculate vertex motion blur, so the plugin assume this. If you check the Modo Event Log you might see an error to this effect from the plugin.

Thanks

Paul

[jwfvr]

forgot to mention, this file makes use of a lot of proxies - maybe this is nothing but it seems like the error is triggered when I unhide a proxy (or many proxies at once) it also seems consistent in that the error shows up when I unhide the first time only. if I hide and unhide the same exact group of proxies the error does not repeat.

[jwfvr]

just saw you reply thank you- the only reason I hide items is me being worried about system ram - I believe that to max out 24 gigs of vram I need to use about 128 gigs of system ram (including win) so I was thinking if I ease my way in, I prevent a crash. is that not the case? I can see when I load most of it at once, my system ram usage climbs right to 125-127 (according to the task manager)

[jwfvr]

btw - no animations here - just still renderings

[face_off]

I believe that to max out 24 gigs of vram I need to use about 128 gigs of system ram

I'm not sure that is the case. If all meshes are live, then you could have a relatively small RAM size and still max out your 24GB of VRAM (since each mesh would be loaded separately into Octane). The Maximum Vertices tooltip on the plugin says "10 million polygons takes approximately 2.3GB of graphics card memory to store."

forgot to mention, this file makes use of a lot of proxies - maybe this is nothing but it seems like the error is triggered when I unhide a proxy (or many proxies at once) it also seems consistent in that the error shows up when I unhide the first time only. if I hide and unhide the same exact group of proxies the error does not repeat.

If you have a scene (which will fit into my 32GB RAM PC) which produces this error, can you send it to me please, and I can take a closer look.

Thanks

Paul

[jwfvr]

I would love to send you the file, its about 2.8 gigs thou. when its fully opened in modo its using 60gigs of ram. (not open in octane, just modo) the gl count is 908 million. if I try to load it into octane all at once, I can watch the system memory climb till it hits 128 and then it crashes. if I load it bit by bit, I can get most of it open but cannot get past about 2.5 gigs of OOC before it crashes - but I have the OOC settings to 24gigs. if I bring the "trigger" items into a empty scene, they render fine so it seems very much related to maxing out memory. I think the system memory needs are much higher to get it loaded, but after its loaded the memory usage drops back down. is there any documentation as to what the error messages refer to exactly? I am getting different ones when I try different amounts of geo when loading. I am guessing there is no clear answer/solution but I really would love to know if the errors are specific to the gpus themselves, or the file, or a corrupt mess or material.

[face_off]

I would love to send you the file, its about 2.8 gigs thou. when its fully opened in modo its using 60gigs of ram. (not open in octane, just modo) the gl count is 908 million. if I try to load it into octane all at once, I can watch the system memory climb till it hits 128 and then it crashes. if I load it bit by bit, I can get most of it open but cannot get past about 2.5 gigs of OOC before it crashes - but I have the OOC settings to 24gigs. if I bring the "trigger" items into a empty scene, they render fine so it seems very much related to maxing out memory. I think the system memory needs are much higher to get it loaded, but after its loaded the memory usage drops back down. is there any documentation as to what the error messages refer to exactly? I am getting different ones when I try different amounts of geo when loading. I am guessing there is no clear answer/solution but I really would love to know if the errors are specific to the gpus themselves, or the file, or a corrupt mess or material.

900 million polygons? That is 10 times higher than the previous highest polygon count I have seen! I think you need to reduce the poly count.

Paul

[jwfvr]

900 million is the gl count, which is each poly triangulated (if I have that right) so I don't know if you would cut that number in half? also there are a ton of replicators and instances, but those seem to add to the count regardless. I have the max vertices set to 50 million.

[face_off]

900 million is the gl count, which is each poly triangulated (if I have that right) so I don't know if you would cut that number in half? also there are a ton of replicators and instances, but those seem to add to the count regardless. I have the max vertices set to 50 million.

Max Vertices is the maximum vertices for any individual Live mesh, or for the combined non-Live meshes. It sounds like potentially as Octane is using more OutOfCore, it's running out of RAM. I would suggest limiting your max OutOfCore to ensure you still have free RAM. And you should be able to reduce your Max Vertices to the size of your largest mesh - just make sure all meshes are "Live".

I hope that helps.

Paul