Hi all, This has something to do with the way the layer closest to the high-density edge of the transfer function gets treated. Setting the maximum density of the transfer function to be well above the maximum density in the last frame seems to be a quick fix for the problem (e.g. in this new video [made with this updated script] which looks pretty good until the very end). I'm not familiar with parallelism in the camera object so any advice I could get on this would be greatly appreciated. I should be available on IRC tomorrow to try to work this out. -Nathan On Feb 29, 2012, at 11:04 PM, Nathan Goldbaum wrote:

Hi Sam,

It looks like cam.comm.rank = 0 and cam.comm.size = 1 for all of the MPI tasks and at each iteration of the loop.

Nathan Goldbaum Graduate Student Astronomy & Astrophysics, UCSC goldbaum@ucolick.org http://www.ucolick.org/~goldbaum

On Feb 29, 2012, at 10:22 PM, Sam Skillman wrote:

...

Hi Nathan,

This is pretty weird -- Can you print out cam.comm.rank and cam.comm.size just after you create the camera object? The only thing I can think of is that there is some invalid reference to the wrong communicator. When doing this you can get rid of the cam.snapshot so that it runs through the code much faster. I'll be on IRC tomorrow morning as well if you want to catch me there.

Sam

On Wed, Feb 29, 2012 at 8:37 PM, Nathan Goldbaum wrote: Hi all,

I'm running into some weird issues when I try to make a time series volume rendering movie using parallel_objects.

The script I'm running is pasted here: http://paste.yt-project.org/show/2206/

And the resulting movie is located here: http://ucolick.org/~goldbaum/files/volume.avi

The movie is stitched together using ffmpeg but i've confirmed that the artifacts are present in the individual frames. This is not due to my use of the enhance script since the artifacts appear in the raw frames and in fact the movie was stitched together using the raw frames. I don't see any artifacts when I volume render individual frames in serial. I'm doing one processor per file parallelism.

Thanks for your help with this - I know there's a cool looking movie buried in there somewhere.

Nathan Goldbaum Graduate Student Astronomy & Astrophysics, UCSC goldbaum@ucolick.org http://www.ucolick.org/~goldbaum

_______________________________________________ yt-users mailing list yt-users@lists.spacepope.org http://lists.spacepope.org/listinfo.cgi/yt-users-spacepope.org

_______________________________________________ yt-users mailing list yt-users@lists.spacepope.org http://lists.spacepope.org/listinfo.cgi/yt-users-spacepope.org

!DSPAM:10175,4f4f175630302723031603!

!DSPAM:10175,4f4f1f9142002844514455! _______________________________________________ yt-users mailing list yt-users@lists.spacepope.org http://lists.spacepope.org/listinfo.cgi/yt-users-spacepope.org

!DSPAM:10175,4f4f1f9142002844514455!

Hi all, This was fixed by using the col_bounds keyword to add_layers(). The issue was that the transfer function was nonzero at the high density ceiling. -Nathan On Mar 1, 2012, at 6:22 AM, Stella Offner wrote:

Hi Nathan,

I've seen the exact same behavior making yt volume renderings not using any parallelization, so I'm surprised if that's the problem. I think I've fixed it in the past by setting the max value appropriately (which it sounds like you've discovered).

Stella

On Mar 1, 2012, at 4:20 AM, Nathan Goldbaum wrote:

...

Hi all,

This has something to do with the way the layer closest to the high-density edge of the transfer function gets treated. Setting the maximum density of the transfer function to be well above the maximum density in the last frame seems to be a quick fix for the problem (e.g. in this new video [made with this updated script] which looks pretty good until the very end). I'm not familiar with parallelism in the camera object so any advice I could get on this would be greatly appreciated.

I should be available on IRC tomorrow to try to work this out.

-Nathan

On Feb 29, 2012, at 11:04 PM, Nathan Goldbaum wrote:

...

Hi Sam,

It looks like cam.comm.rank = 0 and cam.comm.size = 1 for all of the MPI tasks and at each iteration of the loop.

Nathan Goldbaum Graduate Student Astronomy & Astrophysics, UCSC goldbaum@ucolick.org http://www.ucolick.org/~goldbaum

On Feb 29, 2012, at 10:22 PM, Sam Skillman wrote:

...

Hi Nathan,

This is pretty weird -- Can you print out cam.comm.rank and cam.comm.size just after you create the camera object? The only thing I can think of is that there is some invalid reference to the wrong communicator. When doing this you can get rid of the cam.snapshot so that it runs through the code much faster. I'll be on IRC tomorrow morning as well if you want to catch me there.

Sam

On Wed, Feb 29, 2012 at 8:37 PM, Nathan Goldbaum wrote: Hi all,

I'm running into some weird issues when I try to make a time series volume rendering movie using parallel_objects.

The script I'm running is pasted here: http://paste.yt-project.org/show/2206/

And the resulting movie is located here: http://ucolick.org/~goldbaum/files/volume.avi

The movie is stitched together using ffmpeg but i've confirmed that the artifacts are present in the individual frames. This is not due to my use of the enhance script since the artifacts appear in the raw frames and in fact the movie was stitched together using the raw frames. I don't see any artifacts when I volume render individual frames in serial. I'm doing one processor per file parallelism.

Thanks for your help with this - I know there's a cool looking movie buried in there somewhere.

Nathan Goldbaum Graduate Student Astronomy & Astrophysics, UCSC goldbaum@ucolick.org http://www.ucolick.org/~goldbaum

_______________________________________________ yt-users mailing list yt-users@lists.spacepope.org http://lists.spacepope.org/listinfo.cgi/yt-users-spacepope.org

_______________________________________________ yt-users mailing list yt-users@lists.spacepope.org http://lists.spacepope.org/listinfo.cgi/yt-users-spacepope.org

!DSPAM:10175,4f4f1f9142002844514455! _______________________________________________ yt-users mailing list yt-users@lists.spacepope.org http://lists.spacepope.org/listinfo.cgi/yt-users-spacepope.org

!DSPAM:10175,4f4f1f9142002844514455!

_______________________________________________ yt-users mailing list yt-users@lists.spacepope.org http://lists.spacepope.org/listinfo.cgi/yt-users-spacepope.org

-------------------------------------------------------------- Stella Offner NSF Postdoctoral Fellow Harvard-Smithsonian Center for Astrophysics http://www.cfa.harvard.edu/~soffner

!DSPAM:10175,4f4f862620186797501! _______________________________________________ yt-users mailing list yt-users@lists.spacepope.org http://lists.spacepope.org/listinfo.cgi/yt-users-spacepope.org

!DSPAM:10175,4f4f862620186797501!