How much ram denoise takes? yesterday ive try a 6000 x 12400 pixels image and the render did ok and only took like 40 gigs of ram but when denoise its apllied it crashes, is there a maximum resolution for the denoise to work? also try the dr and happend the same as other people. intel xeon 3.66ghz 64 gigs fo ram on system.
I think that is not the problem with the denoiser's ram usage itself but with the additional render elements needed for it to work - there are at least 3 (from that I can tell) of those and at that resolution those elements themselves can be quite a problem - just try to render that scene with denoiser off yet with some additional elements (no less than 3) enabled, I think it doesn't matter which ones those are but I would expect out of ram crash in that case too. Happened to me too and not once, and I've got 64gigs of ram too, it's just that scenes like this
https://www.flickr.com/photos/119850875@N05/18340080264/ where it happened are really ram hungry and if some additional render elements are enabled they crash miserably while without them this particular scene chewed only 48gigs out of 64 and rendered fine or at least did so on 4k resolution - on the other hand your scene's render resolution is much, much bigger\higher and so are the additional ram requirements for those passes to be stored in memory, and that sad limitation, as I can understand and as far as my experience goes, is renderer agnostic and that rule applies to all of them not Corona only (
Maybe some sort of solution could be done on developer's side to avoid such and similar problems\limitations, maybe something like rendering those passes to hdd and then loading them to vfb only after render itself has stopped or something like that, but honestly I have no idea if that is possible at all or if it is, what performance hit it would produce.
At least it could be a good idea to put some warnings of that behavior in help bubbles for denoising selector\toggle (same goes for render elements too but don't know where I would put it there) for users to understand what can happen if it is enabled and to more predictably avoid such unpleasant surprises in the future.
