Page 1 of 1
strange isosurface behavior
Posted: Wed Sep 16, 2009 2:46 pm
by kbedka1
See attached example, I have a 3-D gridded field, 69 levels, 400 elems, 300 lines. I have data in the upper levels set to a value of 1.5 and the data in the lower levels set to .5. When I do an isosurface and set my isosurface value to .5, why do I see both levels of data? I should only see the levels encasing the values of .5. Also, Mc-V is picking isosurface colors out of this air...that grey value is not in the colorbar. It was doing the same behavior before but was choosing a magenta color, which DEFINITELY was not in my color bar.
Re: strange isosurface behavior
Posted: Wed Sep 16, 2009 5:33 pm
by tdrink
Hi Kris,
I'll assume you have only data field in the display. The iso code has been
around awhile and thus pretty rigorously tested, so it's likely that your
3D field is multi-valued in the sense that values around 0.5 exist
in more than one location.
As to the color question, this could be an artifact of strong asymmetry in
the field gradient. I think the color are computed via interpolation
resampling - I'd have to look closer.
Tom
Re: strange isosurface behavior
Posted: Fri Sep 18, 2009 7:37 pm
by kbedka1
there are only values of .5 and 1.5 in the field. The 1.5 values are the upper level cloud and the .5 are the lower level cloud. This is the reason for the differing color between the two layers. What I'm not understanding is why both appear when only an isosurface value of .5 is selected in the layer controls.
Re: strange isosurface behavior
Posted: Tue Sep 22, 2009 4:03 pm
by tdrink
Can you provide the dataset? If there are only (2) discrete values on
the grid, maybe iso-surface is not the best choice. What are you
hoping to see?
Re: strange isosurface behavior
Posted: Tue Sep 22, 2009 6:50 pm
by kbedka1
I was expecting to be able to see one layer or the other based upon the isosurface value I choose. I thought that this is the essence of isosurface, when you pick a value, you only see points in your 3-D grid that equal that value. It appears that all values are being shown no matter what the isosurface value is in the layer controls. My data file is at:
ftp://ftp.ssec.wisc.edu/pub/incoming/ma ... ness.nc.gz
Re: strange isosurface behavior
Posted: Wed Sep 23, 2009 6:31 pm
by tdrink
Thanks for the data. The idea is that the grid samples a continuous field, and
interpolation is done to determine the location of the iso-value. So believe
it or not, the algorithm is working as intended. To see what I think you want
to see, your file could contain NaNs where there are clearly no clouds
since a cloud thickness approaching zero, might just be a really thin
cloud. If you can do this, I think you would see what you expect. This
would also help with the color interpolation problem.