maxpayne
March 19, 2014 09:32:10
Hello all. I'm trying to do smoke effect. I've got some result and I noticed that the smoke has bugs. There are “lines” in the smoke. It seems that the smoke move in voxels only to horizontal direction. Does anybody know what it is ?
Thanks.
eetu
March 19, 2014 09:48:32
I've seen that stuff too. I haven't had the time to methodically chase the culprit, but my gut feeling is that it's some of the shaping operators in pyro, particularly I have my eyes on sharpening.
Sadjad Rabiee
March 19, 2014 13:26:06
I should check your project ,I should know about your parameters and setup ,Sometimes that's because Pyro parameters and sometimes that's because Render setting ,for example sometimes we have some noise and unwanted event on the Pyro that they solved with Raytracing Engine !
So upload a simple example about your project if you can :?
Constantine Petrov
March 19, 2014 18:51:11
To me it looks like an effect of Disturbance.
maxpayne
March 20, 2014 09:02:37
Here is a settings of the solver. I found a video with comparing parameters.
https://www.youtube.com/watch?v=qwGshtSQuxY [
youtube.com]. There are a lot of the same bugs when he is comparing disturnace parameters. But disturbance is off in my case.
Constantine Petrov
March 20, 2014 09:13:08
Huh, so it's neither Disturbance nor Sharpening. Maybe Confinement then?
maxpayne
March 20, 2014 09:20:29
Probably. I should check this out.
eetu
March 20, 2014 10:09:58
Please do, we'd all appreciate it if you found out the culprit
Btw, are you solving with OpenCL active?
maxpayne
March 20, 2014 10:59:24
Open Cl is disabled. I think that the reason in the fluid engine.
maxpayne
March 21, 2014 03:36:46
I tried to disable Confinement parameter. It didn't help.
eetu
March 24, 2014 05:13:25
Sharpening is using the Gas Reduce Local DOP, which works on a cubical subvolume, and that could introduce axis-aligned artifacts - I presume.
“ The neighbourhood regions are always cubic in shape, despite the term radius being used. ”http://www.sidefx.com/docs/houdini13.0/nodes/dop/gasreducelocal [
sidefx.com]
maxpayne
March 24, 2014 06:35:23
Yeah but sharpening is off
xukca
March 27, 2014 07:25:11
Last summer, when I was exploring pyro2.0 I encoutered the same issue. There was a fix suggestions in THIS
http://www.sidefx.com/index.php?option=com_forum&Itemid=172&page=viewtopic&p=133337&highlight=#133337 [
sidefx.com] thread, but I still couldnt resolve the issue.
I must try again turning off the shaping tab, when I have some free time…
maxpayne
March 27, 2014 07:52:14
Hi xukca. It seem that you had problem with a banding. It's a bit another thing.
Skybar
March 27, 2014 08:21:30
Hmm I seem to have the same problem as well. No idea what is causing it. Everything except Dissipation under the Shape tab is turned off.
Neil78
March 27, 2014 14:05:50
Me too - never worked out the problem…. - although I found with motion blur, DoF, and a lens convolve in Nuke, most of that stuff dissapears.
xukca
March 31, 2014 05:01:14
maxpayne
Hi xukca. It seem that you had problem with a banding. It's a bit another thing.
Thanks for the replay maxpayne! If this is really a banding issue, what would be a propper solution? I could try playing with the color gradients a bit, but I somehow doubt this alone will fix the problem… Is the only viable option to introduce noise/dither and blur to the image in post as Neil78 suggested?
eetu
March 31, 2014 07:34:07
xukca
maxpayne
Hi xukca. It seem that you had problem with a banding. It's a bit another thing.
Thanks for the replay maxpayne! If this is really a banding issue, what would be a propper solution? I could try playing with the color gradients a bit, but I somehow doubt this alone will fix the problem… Is the only viable option to introduce noise/dither and blur to the image in post as Neil78 suggested?
Did you try changing your volume filter from Box to Gaussian, and maybe increasing the size a little bit from 1.0?
(OBJnode ->Render->Shading->Volume Filter)
johner
March 31, 2014 09:31:14
Sometimes MacCormack advection can cause minor artifacts, so it might be worth trying BFECC advection for density and velocity. It's a bit more expensive but less prone to artifacts. (Advanced | Advection tab)
maxpayne
April 1, 2014 01:46:09
I tried this. Everything the same. no differents.