LuxCoreRender Forums

Can I save the progress of rendering in order to continue it in the future? Or is such a function planned for at least a year?

S0rda wrote: ↑Thu Feb 28, 2019 8:29 pm Can I save the progress of rendering in order to continue it in the future? Or is such a function planned for at least a year?

It is already possible but I think it is not supported directly by BlendLuxCore: you have to export the scene and than you have to use LuxCoreUI (included in stand alone version) to render the scene. From LuxCoreUI, you can save the rendering/scene to a single file and resume from there.

It's fine. Thank. After me I have another question. I have two video cards on a computer, gtx 1080ti and gtx 970, 11 GB on one, 4 GB on the other. The question is - if my final scene weighs more than 4GB, will the render just not happen, or what? Maybe I somehow misunderstand this process.

S0rda wrote: ↑Thu Feb 28, 2019 10:06 pm It's fine. Thank. After me I have another question. I have two video cards on a computer, 1 and 2, 11 GB on one, 4 GB on the other. The question is - if my final scene weighs more than 4GB, will the render just not happen, or what? Maybe I somehow misunderstand this process.

If you start the rendering on both, you are likely to receive an error while starting on the 4GB card (however some new GPU is able to swap memory to/from CPU memory so it may even start but the rendering is going to be veeeery slow).
However you can select inside Blender the GPU(s) (aka OpenCL devices) to use for the rendering. So just select only the GPU with 11GB if you run out of memory on the small one.

Very nice to receive detailed help. Could you explain something else? Is rendering by layers technically capable of reducing the amount of memory used for each layer that is currently being rendered? Or in any case, the program has to read the entire amount of information, for example, to get a shadow on the second layer from an object that is on the first layer?



==================================================
By the way, what I noticed when I installed the 970 in parallel (naturally, if not) of my 1080th, the rendering speed almost doubled, despite the fact that the number of cores at 970 is two times less than 1080. This is explanation? That is, logically, I expected a speed increase of about 30-50%, and not nearly 100% as it happened now.

S0rda wrote: ↑Thu Feb 28, 2019 10:48 pm Very nice to receive detailed help. Could you explain something else? Is rendering by layers technically capable of reducing the amount of memory used for each layer that is currently being rendered? Or in any case, the program has to read the entire amount of information, for example, to get a shadow on the second layer from an object that is on the first layer?

Yes, any type of AOV is a side product of the complete rendering process so you always need to perform the complete rendering no matter of of the final output type.

S0rda wrote: ↑Thu Feb 28, 2019 10:48 pm By the way, what I noticed when I installed the 970 in parallel (naturally, if not) of my 1080th, the rendering speed almost doubled, despite the fact that the number of cores at 970 is two times less than 1080. This is explanation? That is, logically, I expected a speed increase of about 30-50%, and not nearly 100% as it happened now.

That sounds quite strange, the 1080 GTX should be quite faster than the 970 GTX. Try to do the same rendering with the 1080 alone, the 970 alone and the 1080+970 together and compare the samples/sec you get.