Working on a test animation of a kitchen.
There are some chrome appliance handles and table legs
that I just cant seem to clean up. (Sorry no pic at this time)
Went from using Adaptive Subdivision 0 and 3 to fixed rate of 3 with
Random checked. This cleaned it up quite a bit. Question is should
I stick with fixed rate? This is what the manual suggests for large scenes
and it is indeed faster thus far. Anyone run into similar situation?
To compare the Fixed rate method takes 50% less time to render than
Adaptive with less time. I did a test with a smaller scene just a couple
minutes a frame. Fixed rate was faster but I had to use a value of
4 with Rand, at 5 with Rand it took longer that Adaptive.
Any tips would be great. These frames take quite a long time to render so
testing right now is quite painful. The Adaptive version took 4h42m on a
Dual Intel 2.0 and with Fixed it was 2h24m (same box)
I would like to see your settings settings and a image would help too…
Are you using very high interpolations? What look up method are you using? Seems very high render times for a animation :shock:
How many subdivisions are you using for the glossi reflections?
Will try to get you some more complete info this evening, in a bit of
a crunch at the moment. Thanks GoncaloP, Agreed the times
are high, hopefully when I get more info you can show me the way.
This is the first frame of the sequence, very short only a pan across the
room over 240 frames.
The red circled areas show where I am having some problems
The tile backsplash, the grout lines are jumpy and grainy.
Basice VRay mat with a bump (sorry forgot to screen cap the mat before
I left work)
The legs of the table and the salad bowls under it are smooth but over the
animation there is some twitching.
VRay Settings
Stainless material
Mainly from settings found here on the post. Interpolation was just not
working and lower Subs did not work well at all, for the most part it
is clean, the Chrome is the problem.
Chrome material
Fairly basic. Pretty much use the same settings for mirrors (almost)
and there are no problems with flat objects. But the handles and table
legs are a mess, grainy and twitchy.
One more mess to add. Recently I had some unfurnished animations with stainless appliances
in the kitchen. When my box would render the stainless it was like the
smoothing groups would change slightly causing the reflections
to “jump” or warp in a different direction. So I did a clean sweep and MAX
and all related plugs and a re-install. I thought this was resolved. I will
check the bad frames tomorrow to see if my machine rendered any of
them.
All of the slaves are running same versions of everything.
I will VNC tonight and rerender with .25 and .25 for the Clr and Nrml thresh
and turn up the sub-divs on the Chrome and will check the result in
the morning when I get to work. Will post results.
Re-solved with Animation High settings @ every 10 frames
(previous was every 5 - would this cause higher times?!) and changed
the chrome sub-divs to 20 and the Stainless to 40.
The render time has dropped to about 1hour per frame on the single
AMD 2600XP’s so that is nice. But the stainless is back to looking grainy
and the Chrome appears to be the same. It is so weird, it seems that if
you put the chrome mat on any cylindrical or curvy object that is relatively
small it will be grainy, the larger the object the cleaner it will render
though the sink uses the same material and it is clean.
Took the bump off the backsplash, it is cleaner but still grainy in the
grout areas.
No matter what I do to the appliance handles they will not clean up.
Not really what to try now.
Perhaps now I should increase the AA to 4 and see what happens.
If you have lots of glossy effects, it’s best to use Fixed rate. 3 subdivs is probably too low for the fine detail that you have; use 4 or 5. Also try playing with the QMC sampler parameters to control the quality of glossy reflections (for example, increase the Min samples, or lower the noise threshold and/or the early termination amount).
Thanks GoncaloP, Agreed the times
