i really like the new progressive renderer but we have some troubles here…
if we send the scene to the farm… and use time based rendering we get different frames cause our render machines are not the same… some are old and have only 4 cores…
is there a way to get clean frames and have still the control over the time..?
You will have to limit the max subdivs and/or the noise threshold and leave the time limit to 0. F.e. if you know that on your machine you get the desired good quality for 20 subdivs with noise threshold 0.007, you can just use this. V-Ray prints the reached subdivs at the end of a render, so it could be a guideline what value to put.
Like everywhere in V-Ray, the actual number of samples is the square of the subdivs value. (This is because quality tends to improve with the square of the number of samples; it is not a linear relationship. It also help to avoid typing values into the thousands.)
The white color is for pixels that are not complete yet and will be sampled on the next iteration.
thanks Vlado..!
i really like the progressive mode… its good in sampling motionblur..!
i hope one day we get full CPU+GPU support for progressive… id like to see my whole system working…
Well - progressive rendering in vray isn’t supported on the GPU, or am I wrong? RT of course, but not the progressive image sampler.
But this brings up a good question - is the whole GPU thing under active development? I mean - Octane, Redshift & co are creating a lot of buzz these days.
Not sure what you mean. The GPU renders in a progressive way always.
But this brings up a good question - is the whole GPU thing under active development?Very much so. Check my talk at GTC where I talk about some of the things we did in the past year: NVIDIA On-Demand
The GPU renderer is perfectly usable in a large variety of situations… it’s just a matter of people realizing it.
So, when I use the progressive image sampler the rendering happens on the GPU?
Or do I have to use RT? Or is it the same these days?
EDIT: Okay, think I got it. The imagesampler is really just an imagesampler :).
But on RT subD, displacements (and particles) are not available.
But great to hear that it’s going on! I’ll check that talk soon, thanks a lot!
You have a point there - you still have to go through V-Ray RT. So if you wanted to render final frames on the GPU, the only option in V-Ray for Maya is to export a .vrscene and render with V-Ray Standalone, which is inconvenient. However it should not be too difficult to expose the GPU engine as an option for regular frame rendering too.
Subdivision surfaces and displacements actually work in the latest nightly builds.
Partciles, volumes and some other stuff do not work yet, but we’ll get there.
Thanks for the update.
An implementation as a simple switch to choose if you want to render with the gpu or cpu would be really nice. And also good for marketing, I guess.
We’ll check the latest builds soon, sounds really good! Keep it up!
RT CPU is the same as Progressive production render (Vlado can confirm ?).
RT GPU (Cuda or OpenCL) is another code.
If you use CPU + CPU with OpenCL, you will be limited with GPU features.
Pechart wants to use the GPU for final frame renders. Right now, this is not possible from inside Maya - the GPU mode is only available for IPR rendering.
Exactly. But actually I think I’m not the only one who wants to do this ;).
And IMHO VRay really has to support this. Some clients (!) already asked us if we’re using this “new octane renderer” or “that redshift” which other studios are switching to, because they’re “much faster and we can do more turnarounds”. Not my opinion, but we really had such talks by now.