I can easily render 60+M triangles with 8GB of GPU ram, do you run out of GPU memory ?
MaxToLux - Autodesk 3ds Max Integration for LuxCoreRender
Re: MaxToLux - 3ds Max Integration for LuxCoreRender
GPU has just 2GB,
I have 16Gb ram. But I will check if it is the GPU.
Re: MaxToLux - 3ds Max Integration for LuxCoreRender
I deselected the GPU, rendering with bidirectional CPU and metropolis.
it renders using almost 14GB ram but it renders. As you guessed it was my GPU the problem.
Re: MaxToLux - 3ds Max Integration for LuxCoreRender
2Gb of GPU ram is a bit low for this scene, Also if you want to render with CPU i think PathCPU could be much faster.
Of course with instancing this scene will need much less GPU memory too.
But as Dade already mention that you can measure how much GPU ram you need by simply counting the scene triangles.
There are many parameters left and I can not promise any date for adding instancing but I'll try my best to make it happen sooner.
As i have already seen a few of your scenes you can clean up your scenes before render too. for example, if a few objects are behind the camera or they should not be supposed to render at all, all un necessarily texture and materials, etc.
Of course with instancing this scene will need much less GPU memory too.
But as Dade already mention that you can measure how much GPU ram you need by simply counting the scene triangles.
There are many parameters left and I can not promise any date for adding instancing but I'll try my best to make it happen sooner.
As i have already seen a few of your scenes you can clean up your scenes before render too. for example, if a few objects are behind the camera or they should not be supposed to render at all, all un necessarily texture and materials, etc.
Omid Ghotbi (TAO)
Last update information
viewtopic.php?p=34390#p34390
My Artstation https://www.artstation.com/omidghotbi
My Instagram https://www.instagram.com/omid.ghotbi
Last update information
viewtopic.php?p=34390#p34390
My Artstation https://www.artstation.com/omidghotbi
My Instagram https://www.instagram.com/omid.ghotbi
Re: MaxToLux - 3ds Max Integration for LuxCoreRender
oh well that's what I have for tersting MaxToLux
Re: MaxToLux - 3ds Max Integration for LuxCoreRender
Add support for object instancing for now it is not multi-thread and in huge scenes, with too many instances objects can get a little bit slower than the old algorithm but will offer a huge benefit in memory usage.
If i have enough time i will work on multi-threading and providing a better and faster solution later this month.
The very first look is like this: All objects are instances and the scene takes almost no memory.
it is a static function yet and i use a loop for it, so it duplicates objects one by one not all at the same time.
If i have enough time i will work on multi-threading and providing a better and faster solution later this month.
The very first look is like this: All objects are instances and the scene takes almost no memory.
it is a static function yet and i use a loop for it, so it duplicates objects one by one not all at the same time.
Last edited by TAO on Wed Mar 03, 2021 8:48 pm, edited 1 time in total.
Omid Ghotbi (TAO)
Last update information
viewtopic.php?p=34390#p34390
My Artstation https://www.artstation.com/omidghotbi
My Instagram https://www.instagram.com/omid.ghotbi
Last update information
viewtopic.php?p=34390#p34390
My Artstation https://www.artstation.com/omidghotbi
My Instagram https://www.instagram.com/omid.ghotbi
Re: MaxToLux - 3ds Max Integration for LuxCoreRender
Add support for instancing and references, 12000 teapot with more than 370 million polygons.
Translate time 17 seconds.
GPU memory usage about 1GB
Renderer PathGPU
Render time 55 seconds for 991 passes
The impressive result compares to parsing different mesh, especially in memory usage but a little bit time-consuming in translation part.
I do not know what is the dark color in middle I have seen that before in another render too and it is not about instancing.
Cpu: i9 9900k
GPU: Nvidia GTX 1080
Ram: 32GB
Translate time 17 seconds.
GPU memory usage about 1GB
Renderer PathGPU
Render time 55 seconds for 991 passes
The impressive result compares to parsing different mesh, especially in memory usage but a little bit time-consuming in translation part.
I do not know what is the dark color in middle I have seen that before in another render too and it is not about instancing.
Cpu: i9 9900k
GPU: Nvidia GTX 1080
Ram: 32GB
Last edited by TAO on Thu Mar 04, 2021 7:11 pm, edited 1 time in total.
Omid Ghotbi (TAO)
Last update information
viewtopic.php?p=34390#p34390
My Artstation https://www.artstation.com/omidghotbi
My Instagram https://www.instagram.com/omid.ghotbi
Last update information
viewtopic.php?p=34390#p34390
My Artstation https://www.artstation.com/omidghotbi
My Instagram https://www.instagram.com/omid.ghotbi
Re: MaxToLux - 3ds Max Integration for LuxCoreRender
This seems very long to me.
The Blender addon exports 12000 particle instances in 0.2 seconds, and 12000 objects with shared meshes (alt+D copies) in 2.8 seconds.
How does your current algorithm for detecting which meshes can be instanced work?
Re: MaxToLux - 3ds Max Integration for LuxCoreRender
You right, i need to do it in a different thread, so i collect all data that i need and then duplicate all objects once the thread is done. for now, i did it in the main thread and it is a bit slow especially that 3dsmax uses a single process to handle UI and object pointer.
Omid Ghotbi (TAO)
Last update information
viewtopic.php?p=34390#p34390
My Artstation https://www.artstation.com/omidghotbi
My Instagram https://www.instagram.com/omid.ghotbi
Last update information
viewtopic.php?p=34390#p34390
My Artstation https://www.artstation.com/omidghotbi
My Instagram https://www.instagram.com/omid.ghotbi
Re: MaxToLux - 3ds Max Integration for LuxCoreRender
The Blender addon does it single threaded too, and in Python on top of that.
It sounds to me like the algorithm to detect instances is very very complicated. How does it work, in detail? I would first check if there's a faster alternative before thinking about multithreading.
It sounds to me like the algorithm to detect instances is very very complicated. How does it work, in detail? I would first check if there's a faster alternative before thinking about multithreading.