Hello,
I’m not sure what is the correct Render settings-Production for animation, I’ve found here https://docs.chaos.com/display/KB/Render+Settings+Explained+-+Quality+vs.+Render+Time A high settings reference but these are for still images and do not belong to 3ds max.
I aim to render for showreel which question by itself, should be high or medium? The current resolution is 2048x1172.
The image below shows my render settings and frame from a scene with relatively low volumetrics compared to future frames. It took me 1 hour and 37 minutes to render, Are those settings practical or I overkilled it?
It’s looking great…I was curious how this would turn out 
In my opinion, for animation for your showreel that resolution and render time is too much, as a viewer/prospective employer will get the idea from the animation itself; the camera work,
editing etc. (depending on what exactly you’re aiming to sell)
That is supported by hi-res stills to show specifics of shading/lighting or whatever.
Depending on how long any particular segment is, base your frame time on what’s achievable without spending either too much cash at a farm
or if you’re doing it yourself then on what’s achievable in the time you have to do it.
So my advice is to do a few tests at maybe 1280x720 which arguably is big enough.
You’d need to test the sequence in any case to make sure it flowed well.
That resolution plus tweaking the settings is probably able to achieve a time per frame that is
significantly lower than your current times.
Thanks for tips, I did low res test which proved the animation solid. The problem now is the final Res, I might decrease the res as you suggest.
Cool. It’ll be interesting to see how quick you can get it to be…
It could’ve been even faster with Vray cloud, but in my case, it seems impractical, due to low upload speed.
Is there maybe an option to run it through Vantage and comp the volumetrics back in separately?
That would speed things up considerably.
What is Vantage?
I’m not sure I have the needed skills to comp volumetrics in Post, many volumetrics overlap each other and I require the opacity to look right. I’m not saying it can’t be done, I’m not sure I know how.
You can probably raise the noise threshold (0.025, f.e.), and denoise the sequence using our denoiser in multi-frame mode (look for vdenoise.exe in our \bin folder).
Test it on a few representative frames, and you’ll know if it’ll work for you.
For fun (hope you don’t mind), I made a quick scene, similar to what yours seemed to contain, and was able to get very fast times.
Just 6 mins for this one.
Of course I can’t be sure I included all the stuff that you may have used but it had the necessary slow
stuff…mblur, vdb volumes, environment fog…no displacement though, which I’d guess is the most likely
big hit to render times if you used a lot.
In my case, I’ve made two renders at 1280x720, one with Fog and one without. Scene with fog took 01:01:49, the scene without fog took 00:21:39, almost third.
Well that’s something…shame to lose the fog though 
Would you be able to share the fog settings?
@KushKush I think your step size is way too small. Here is the description of the parameter from the docs:
Step size – Determines the size of one step through the volume. Smaller steps produce more accurate results but are slower to render. In general, dense volumes require smaller step sizes than more transparent volumes. In practice, step sizes that are two to three times smaller than the Fog distance parameter work well.
Usually setting it to 1/4th or 1/5th of the distance has worked well for us (250m-200m in your case)
What Svetlin said.
Thanks for pointing out the issue, hopefully, it will reduce render time.
Does step size parameter changes with the unit scale? Is it is, is there a difference in render times when the scene is at different unit scales?
All you need to care for is the ratio between the distance and the step size, regardless of which units they’re expressed in.
You were forcing the fog to be sampled to its max steps (where, ofc, the max distance was visible) per pixel sample (so, multiplicative with the AA), which are clamped (bottom of the dialog) to 1000, by default.
That’s bound to take a long time.
Check the log often for clues as to what’s taking the most rays, you’ll then know where such a problem may lay.
Using a samplerate RE will further narrow down the areas that are sucking up AA samples.
As for the dark line at the edge, try lowering the horizon line (by raising the horizon offset value in your vraySun/vraySky -if it overrides the sun node-).
What is simplerate RE?