Loading scene takes longer than rendering Tweet

[coilbook]

We have a 4.5 GB scene (city with moving cars and itoo forest trees.) By the time the scene is loaded to all slaves it is almost done rendering making rendering slow. (master with 3 GPUs does 80% of rendering) It takes 1 min 20 sec to render a frame where 1 min is just to load scene to all slaves (so really if we did not have to wait for scene loading it would only take 20 sec to render)

I do not understand why octane reloads scene to GPU v ram every frame i thought stationary geometry should just stay in GPU ram so should movable proxies like moving cars.

Will otoy have any solution for this in the future especially for slaves

P.S this is Rico_UK post from two years ago. He is right when I scrub a slider octane instantly updates everything. So why during actual rendering we have to wait 1 min every frame to reload all 4 Gigs of data



" was wondering if there was any detailed documentation on instancing and movable proxies in Octane, especially with regard to optimisation and also the loading of a scene at every frame when it comes to render time. I've found some info online, but all very broken bits of advice, I'm not actually sure what's happening under the hood. Main points to raise from my finding are:

If I have a lot of instances of the same object it is slow as hell to load the scene, and it uses a lot of vram.

If I select all of those objects and check their movable proxy checkbox then they load quickly, use next to no vram and I can move them about in the viewport.
If this is true why isn't this 'magic' checkbox always off? There must be some tradeoff surely.

When it comes to actually rendering the final sequence / not using the Octane preview window, Max reloads the geometry for every single frame. This seems crazy as I can scrub through the timeline with the Octane frame buffer open and it doesn't need to update which is awesome. And with movable proxy checked the objects that are moving also update. So why when I need to render the final sequence does it need to reload all of the geometry every time?

Also, there's the Autoupdate Movable Proxy checkbox in the render settings, what does this do? It seems movable proxies by their very nature automatically update with or without that checkbox."

viewtopic.php?f=27&t=39750&p=183043&hilit=slow+loading+to+vram#p183043

[coilbook]

here is more
Why reload all this static mesh like instance tree from itoo forest, building if the do not move?

the lightweight ball animation will load nearly immediately on the next frame, but the 10 different flavors of trees will take you the 2 gb to reload into your cards every frame - even if they do not move - and THAT IS A WEAKNESS of the movable proxy approach.

[Elvissuperstar007]

viewtopic.php?f=80&t=57342
about it for a long time already all write very slowly! I open 2.25, and get real time and octane 3 I get a long load, it is not real time, waiting for a long download

[Jolbertoquini]

Strange for me is fine...

If I do it similar scene as you have and I set the static tree as scatter they load one time on the first frame only if I have set as reshapable they load for each frame. and my car is set at movable proxy they are load only on the first frame and they Octane load only the transforms.

If I have animated instance so animated tree and I add reshapable they are load for each frame.

and my render setting are set as "animation mode" "Movable proxies" to recognize the settings. I use the Maya plug.

Of course doesn't work with our farm with deadline cause deadline load the always the scenes but this is not Octane problem.

but as cmd is working fine to me.

Best,
JO

[roeland]

here is more
Why reload all this static mesh like instance tree from itoo forest, building if the do not move?

the lightweight ball animation will load nearly immediately on the next frame, but the 10 different flavors of trees will take you the 2 gb to reload into your cards every frame - even if they do not move - and THAT IS A WEAKNESS of the movable proxy approach.

If you don't get a wait when scrubbing, that probably means the moveable proxy setting is set up correctly, and Octane is updating only the transforms for the meshes. In that case it will also only upload the new transforms to the slaves. Uploading data to the GPU is a fast operation, even for big chunks of data (it happens on your local computer all the time while scrubbing) so it definitely won't affect network render speed. The slow part is sending data over the network.

I think this depends on how the batch rendering is set up. If the batch rendering completely reinitializes the render engine for every frame, it will have no way to do partial updates and the entire scene will have to be sent over the network for every frame.

--
Roeland

[coilbook]

here is more
Why reload all this static mesh like instance tree from itoo forest, building if the do not move?

the lightweight ball animation will load nearly immediately on the next frame, but the 10 different flavors of trees will take you the 2 gb to reload into your cards every frame - even if they do not move - and THAT IS A WEAKNESS of the movable proxy approach.

If you don't get a wait when scrubbing, that probably means the moveable proxy setting is set up correctly, and Octane is updating only the transforms for the meshes. In that case it will also only upload the new transforms to the slaves. Uploading data to the GPU is a fast operation, even for big chunks of data (it happens on your local computer all the time while scrubbing) so it definitely won't affect network render speed. The slow part is sending data over the network.

I think this depends on how the batch rendering is set up. If the batch rendering completely reinitializes the render engine for every frame, it will have no way to do partial updates and the entire scene will have to be sent over the network for every frame.

--
Roeland

Thank you

Can we adjust batch rendering . (our set up is simple we have a gigabit switch with cat6 cables and all slaves and master connected to the switch, Cables are 10-20 feet long) When we use scrubbing during octane preview slaves update within 5 seconds (especially during simple camera fly through) but during rendering each new frame first 30 seconds just to reload slaves. Do max tile sample play any role?

[abstrax]

Thank you

Can we adjust batch rendering . (our set up is simple we have a gigabit switch with cat6 cables and all slaves and master connected to the switch, Cables are 10-20 feet long) When we use scrubbing during octane preview slaves update within 5 seconds (especially during simple camera fly through) but during rendering each new frame first 30 seconds just to reload slaves.

This sounds as if the update system behaves differently between using the live viewer and final rendering. I'm assuming you are using the 3ds max plugin If yes, which version? I don't know much about 3ds max but it sounds as if using batch rendering will reset the render engine between frames. If that's the case, the ability to do differential updates gets lost.

Do max tile sample play any role?

No, they don't. This is only relevant for integration itself (i.e. when all the data has been transmitted already).

[JimStar]

Strange for me is fine...

If I do it similar scene as you have and I set the static tree as scatter they load one time on the first frame only if I have set as reshapable they load for each frame. and my car is set at movable proxy they are load only on the first frame and they Octane load only the transforms.

If I have animated instance so animated tree and I add reshapable they are load for each frame.

and my render setting are set as "animation mode" "Movable proxies" to recognize the settings. I use the Maya plug.

Of course doesn't work with our farm with deadline cause deadline load the always the scenes but this is not Octane problem.

This feature you describe is only implemented in Maya and Blender plugins. It is just not implemented in 3ds Max plugin. So, there is no way in current versions of 3ds Max plugin to fine-tune which meshes and how exactly they are reloaded/not-reloaded between frames during batch animation rendering. This is the root of coilbook's problem...
Here is this feature explained for Maya plugin: https://docs.otoy.com/manuals/products/ ... try-types/
Here is for Blender: https://docs.otoy.com/manuals/products/ ... try-types/

[coilbook]

thank you all for explaining. I hope 3ds max will get this fixed one day.

[coilbook]

Thank you

Can we adjust batch rendering . (our set up is simple we have a gigabit switch with cat6 cables and all slaves and master connected to the switch, Cables are 10-20 feet long) When we use scrubbing during octane preview slaves update within 5 seconds (especially during simple camera fly through) but during rendering each new frame first 30 seconds just to reload slaves.

This sounds as if the update system behaves differently between using the live viewer and final rendering. I'm assuming you are using the 3ds max plugin If yes, which version? I don't know much about 3ds max but it sounds as if using batch rendering will reset the render engine between frames. If that's the case, the ability to do differential updates gets lost.

Do max tile sample play any role?

No, they don't. This is only relevant for integration itself (i.e. when all the data has been transmitted already).

we are using the latest octane plugin and 3ds max plugin But looking at jimstar comment it is not implemented in 3ds max. I hope you guys can fix it soon thank you