Hi, so I have recently switched from Vray to Corona and although I love it, Im still confused about a few things.
Firstly, My interactive render takes a lot of time to render. For ex; it took about 3min for 5 passes and the output is still extremely noisy.
I want to know how I can make the render settings to provide me with quick , denoised renders during IR. I tried to use “fast preview denoise” option but it my image so smooth that even the edges got smoothed out and the geometry made no sense.
IR will only be fast if A. your PC is adequately fast, B. the scene is relatively simple and C. there are no fundamental issues with the scene (e.g. the scene fits to RAM, there’s unreasonably complex materials, etc.). Fast denoiser will only helps to achieve clean result faster, but it does not help much in very early stages, when render is still very noisy. You need to provide more information in your post - tell us your setup, show some screenshots with VFB and stats panel during rendering.
Sorry for the insufficient information.
I am attaching a few screenshots
screenshot 1: aften enabling “fast preview denoise during render”. Notice how so much geometry misses and is distorted.
next couple of screenshots are : my render settings
IR without “fast preview” enabled ; and
my scene
Also, I had previously rendered the scene in Vray and it was comparatively fast and showed no error of maxed out RAM.
And since I have read that Corona Renderer uses solely the CPU, the interactive render maxes the RAM. But shouldn’t there be a way to optimize usage and also get clearer results?
My system specs are:
intel core i7-6700HQ
NVIDIA GeForce GTX 950M with 4gb dedicated RAM
16GB DDR4 Memory
Can you show screenshot of task manager during the rendering? Since it’s laptop and it shows unusally low rays/sec, it might be that your machine is thermal throtling due to insufficient cooling.
I also noticed that you have upscaling enabled, while this is not making rendering slower (quite contrary), but it may add to overall feeling of blurriness and degraded image quality.
Thank you for replying
I have attached a screenshot of :
a) The Render setup’s System settings.
The image upscaling factor was set to 3.0 by default and I changed that to 1.0
b) Screenshot of task manager during IR. Also please look at the stats of the VFB and do let me know if the IR is slow as someone pointed out that the rays/sec are very less.
Yep, it’s either thermal throttling, or judging from disk activity it could be out of RAM situation. Anyway, it means there’s issue with your machine and not with Corona.
Ohhhh okay. So is it possible to stress the system less and still use this render engine ? Or how do I optimise the RAM ? I hope there’s some way somehow this problem never came up with VRay during IR and also during final renders of 1080px
That’s very scene dependent, but usually biggest RAM eaters are displacement and huge textures. I noticed in your previous screenshots that your displacement is set to 1 px value. That’s pretty high subdivision for 16 GB RAM, especially in Corona 4, which doesn’t have new improved displacement. Try to reset it to default (2 or 3 px can’t recall exactly) and see if that helps. If you still see low RAM warnings, then try to lower resolution of your textures. Smart texture optimization can save tons of RAM, without sacrificing image quality too much.
Yes okay I will change the displacement back to default 3.0.
And since I started using corona just a few days ago and this is my first space designed using this , I have used almost all materials from corona material library. I also optimised and instanced most of the polygons so that it doesn’t stress the system much.
I just watched a video on YouTube that suggested going to the Laptop’s power settings and lowering the maximum processor state from 100% to 90%. Any such suggestions / advice ? I’ll be quite grateful.
6700HQ is a mobile CPU. Mobile CPUs are never as efficient as their desktop counterparts. The GPU is not the fastest either, which may impact the way Nvidia denoiser works. I would suggest: