BUG #1811: Show point information for plots using geographic projecti… #1878
Conversation
|
@aashish24 @doutriaux1 Please review. Picking works for boxfill and meshfill. We can make it work for all plots after #1881 |
|
@danlipsa why would this baseline changed? test_vcs_settings_color_name_rgba_meshfill.png |
| I = d.array[0].getAxis(-1).mapInterval((X, X, 'cob'))[0] | ||
| # for non-linear projection or for meshfill. Meshfill is wrapped at | ||
| # VTK level, so vcs calculations do not work. | ||
| if gm.projection != "linear" or gm.g_name == 'Gfm': |
There was a problem hiding this comment.
@danlipsa why not use the same code (under if) for both cases?
There was a problem hiding this comment.
Because this code does not work now for isofill, isoline and vectors. When we fix dataset creation we can get rid of the vcs code.
There was a problem hiding this comment.
@danlipsa thanks. And I guess we have an issue for it?
|
@aashish24 I think has to do with the fact that now I use one renderer for all mappers. Meshfill and boxfill use several mappers one for each range of scalar values. I assume using only one renderer changed the order in which those mappers are rendered. We can file an issue fix this many mappers problem. |
| else: | ||
| vg = grid | ||
| # Add a GlobalIds array to keep track of cell ids throughout the pipeline | ||
| globalIds = numpy_to_vtk_wrapper(numpy.arange(0, vg.GetNumberOfCells()), deep=True) |
There was a problem hiding this comment.
@danlipsa why we have to do this? Meaning why we have to explicitly append the global ids?
There was a problem hiding this comment.
Because the hardware selector renders and gets the cell ids of the geometry which are not usually the same as the cell ids of the dataset.
There was a problem hiding this comment.
Because the hardware selector renders and gets the cell ids of the geometry which are not usually the same as the cell ids of the dataset.
this is interesting. But by default should they be same?
There was a problem hiding this comment.
What do you mean 'by default'? In the uvcdat case we have threshold + different mapper for different scalar ranges so the ids are not the dataset ids. If you have only the dataset in the pipeline I think only the polydata will not change ids. Maybe there are others but I would not bet on it :-)
There was a problem hiding this comment.
If you have only the dataset in the pipeline I think only the polydata will not change ids. Maybe there are others but I would not bet on it :-)
I see. yeah, I won't either.
Sure, if you can create an issue for it (if it does not exists already), that would be great. |
|
@danlipsa other than some of the questions I have, this branch looks good to me. I did manually compiled the branch and use the GUI for testing. Looks good to me 👍 |
|
@danlipsa the only thing I am not clear about is point data scalars other than that I think you covered it pretty well. |
|
So, for point data, the hardware selector renders the points using pointids instead of color. So we can get the pointid and from it we can get the point scalar. The only question I have is how close do you have to click on the point to get the point id and how user friendly is that. We might be able to set some sort of point size/range so that you don't have to click exactly on the point to get its value. |
|
Sure. I guess we should create an issue for it as well. |
|
@doutriaux1 I think this is ready to go. :-) |
|
Ping @charlesdoutriaux |
|
@doutriaux1 please approve. |
|
just finished building. Testing the click and approving |
|
@danlipsa so it does not work on isofill. Correct |
|
@doutriaux1 right, it does not workon isofill (#1885) |
BUG #1811: Show point information for plots using geographic projecti…
|
Ok approving this for now then. |
|
@doutriaux1 Make sure you merge the data branch as well: CDAT/uvcdat-testdata#117 |
…ons.