search

Quasars / orange-spectroscopy

Other
52 stars 58 forks source link

Multiple files in Hyperspectra: banded #567

Open markotoplak opened 3 years ago

markotoplak commented 3 years ago

In #562 @GioB81 reported that multiple hyperspectral images appear banded, like this.

gioscreen

This definitely happens when a common grid for all the images, which Quasar tries to infer, does not exist. Even more terrible things would happen if one of the images was rotated.

At this stage, Hyperspectra does know which points belong to which images. A possible solution would be to have an algorithm that find a set of perfect X-Y grids along a set of points and use that for drawing.

stuart-cls commented 3 years ago

If we are re-thinking how we plot away from a strict grid (which is only really valid for array detector data), I'd like to also consider data like this where there may be arbitrary grids, line scans, and points in a single dataset. I can provide the file as an example.

linescan_vis_overlay linescan_vis_overlay_zoom

Note for this data type the measured spot size is not included in the datafile, so it is just as correct to plot something visible as it is to plot this super thin line.

borondics commented 3 years ago

I think it would be awesome to go beyond just simple 2D grid plots. One thing is the uniqueness of the plotting and the other is exploiting new measurement approaches that don't use square grids.

@markotoplak did you mention Voronoi diagrams as 2D plotting at some point somewhere? I would be really curious how these maps (not the FPA and the square grids) would look like. Could we make some "demo" plots if there is data? I can dig up something... Or maybe @stuart-cls also has non-uniform maps?