Hi Sam,

Thanks for the suggestion, I think it could work if I choose an appropriate size for the embedding sphere.

For not using yt ver 2.5 it is because when I use yt 2.5, I got this error:

AttributeError: 'AMRSmoothedCoveringGrid' object has no attribute 'NumberOfParticles'

so John switched back to yt2.4 with the transpose=False fix, which worked most of the time.


On Jan 14, 2013, at 10:27 AM, Sam Skillman wrote:

Hi Renyue, 

If you can describe the galaxy using a data object (sphere, disk, etc), you could then do something like galaxy_object.quantities['BulkVelocity']() to get the motion of just that piece of the data.  I'm not sure if that is helpful for you or not.

I would say that it would be best to get on a more recent version of yt, where you will be able to use the tranpose, and the north vector will make more sense with option 1 or 3.


On Mon, Jan 14, 2013 at 7:25 AM, Renyue Cen <cen@astro.princeton.edu> wrote:
Hi Sam,

I am using option (2) below without transpose (per your suggestion to fix the symmetric image problem)
and it does seem to point to the right (I have a bit trouble to decide exactly at which direction the galaxy 
is traveling even though I have its peculiar velocity, because I do not have information about 
the velocity of the volume it is in).


On Jan 14, 2013, at 10:17 AM, Sam Skillman wrote:

Hi Renyue,

It depends a bit on how you are saving the image, unfortunately.  If you are using a recent changeset of yt (after 2.4 where the transpose causes symmetric images on some compilers/machines), and you save with:

Then image1.png will have north pointing up.

If you did:
im = cam.snapshot()
write_bitmap(im, 'image2.png', tranpose=False)
Then image2.png will have north pointing right.

If you did 
im = cam.snapshot()
write_bitmap(im, 'image3.png', transpose=True)
Then image3.png will have north pointing up.

I am working on ways to make sure that it always points up by using the new ImageArray class, but have not come up with the full solution yet.  I hope that description helps.


On Sat, Jan 12, 2013 at 8:12 PM, Renyue Cen <cen@astro.princeton.edu> wrote:

I think I am unclear about the orientation in pf.h.camera(c,W,L, ..., north=vec, ...).
I  got some nice looking pictures but don't know what north means in this case,
because when I switch north=vec to north=reverse of vec, the image rotates along the vertical direction,
so it seems like north is pointing to either left or right.  It is very desirable for me to be able
to make sure which direction vec points to (vec is the vec in north=vec) in the final rendered plot.

Thanks in advance for help,