TPF viewer #81

kecnry · 2024-01-19T19:03:57Z

This PR implements a dynamically-created TPF viewer.

dynamic viewer to visualize cube (with appropriate data filtering in menu)
mouseover support for cube viewer(s)
fix viewer without any data (fails when attempting to set attributes)
- follow-up: re-enable clone viewer tool for TPF viewers 🐱
filter/exclude TPF entries from lc viewer data menus (~~might need logic or refactoring upstream~~ - see lcviz cube support jdaviz#2678)

Note that in order to get correct mouseover and stretch histogram, spacetelescope/jdaviz#2678 is required (which is targeted for jdaviz 3.9). Even with that, the default stretch percentile often drives the whole range, requiring using a custom stretch 🐱.

Example case:

from lightkurve import search_lightcurve, search_targetpixelfile
from lcviz import LCviz

lc = search_lightcurve("HAT-P-11", mission="Kepler", cadence="long", quarter=10).download() #.flatten()
tpf = search_targetpixelfile("HAT-P-11", mission="Kepler", cadence="long", quarter=10).download()


lcviz = LCviz()
lcviz.load_data(lc)
lcviz.load_data(tpf)
lcviz.show()

Manually setting stretch and colormap:

po = lcviz.plugins['Plot Options']
po.viewer = 'image'
po.axes_visible = True
po.image_colormap = 'Viridis'
po.stretch_vmin = 0
po.stretch_vmax = 20000
po.open_in_tray()

Until the slice plugin is integrated, slicing can be done manually:

lcviz.viewers['image']._obj.state.slices = (slice_number, 0, 0)

Next up:

slice plugin with indicator in all time viewers (try to generalize and inherit from upstream)
- read-only slice indicators in phase viewers (???)
- slice plugin to be able to revive viewer if deleted (in a reusable way that can later be also used in the extraction plugin or any other plugin that acts on cubes)
extraction plugin with live-preview and pipeline defaults

codecov · 2024-01-19T19:15:27Z

Codecov Report

All modified and coverable lines are covered by tests ✅

❗ No coverage uploaded for pull request base (feature-tpf@8c680ca). Click here to learn what that means.

❗ Current head fb656fa differs from pull request most recent head 35de766. Consider uploading reports for the commit 35de766 to get more accurate results

Additional details and impacted files

@@              Coverage Diff               @@
##             feature-tpf      #81   +/-   ##
==============================================
  Coverage               ?   93.36%           
==============================================
  Files                  ?       35           
  Lines                  ?     1734           
  Branches               ?        0           
==============================================
  Hits                   ?     1619           
  Misses                 ?      115           
  Partials               ?        0

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

bmorris3 · 2024-01-29T14:53:17Z

lcviz/viewers.py

+        # Hide axes by default
+        self.state.show_axes = False


Just commenting that I like show_axes = False as the default. I see you toggle axes back on in the example code – handy to have the option, but takes up a lot of space.

ah, yes, that was mostly for debugging and confirming the mouseover information, I don't think we'd want that in any public-facing example.

bmorris3

This looks great! I've put some suggestions below, which we can copy/paste into tickets if you like.

Expose slice number somewhere in the image viewer

I know we will implement the slice plugin in a different PR, but I wanted to propose (somehow) that we annotate the image viewer with the slice number/time/cadence number. If/when we export the image in the image viewer to PNG, for example, it's only useful if you know which slice it came from. lightkurve gives you the TPF slice plots with a cadence number label in the title, but I'm not sure where we'd put it.

You could argue that an annotation is redundant if you can see the slice marker in the light curve viewer or if you have the slice plugin open. But if you're zoomed elsewhere in the LC viewer, or don't have the slice plugin open, that info isn't shown.

If we don't want to put the annotation over the data in the viewer somewhere, one compromise could be in mouseover info for the image viewer.

Image viewer WCS

I confirmed that in your example, tpf.wcs returns WCS for the image in the image viewer. It might be straightforward to use this WCS in the cube viewer (with padding for the time axis for glue via PaddedTimeWCS, already implemented in this branch).

bmorris3 · 2024-01-29T14:56:45Z

lcviz/parsers.py

+        # TODO: move this to an event listener on add_data so that we can also remove when empty?
+        from jdaviz.core.events import NewViewerMessage
+        from lcviz.viewers import CubeView
+        viewer_reference_name = 'image'


Good name. 😊

bmorris3 · 2024-01-29T14:58:35Z

lcviz/viewers.py

+    # TODO: can we vary this default_class based on Kepler vs TESS, etc?
+    default_class = KeplerTargetPixelFile


Curious to check in on this. Does the mission specificity matter in the viewer class?

yeah, I wasn't sure what to do with this 🤔 . Ultimately this is just for the default and the user can override, but maybe its better to just have as None and require the user to pass a class?

I like that.

done and linked a comment back here if we ever want to revisit.

bmorris3 · 2024-01-29T14:59:30Z

lcviz/viewers.py

+                    ['jdaviz:homezoom', 'jdaviz:prevzoom'],
+                    ['jdaviz:boxzoom'],
+                    ['jdaviz:panzoom'],
+                    ['bqplot:rectangle'],


Is there anything stopping us from allowing other ROI shapes at this point?

not really, but we'll have to see what we want to support for actual photometric extraction (and if we want sub-pixel support or not).

Agreed, but even rectangle ROIs aren't required to have 100% overlap with the pixel grid. I don't see a difference with other shapes in that sense.

right, we might decide to just do pixel-by-pixel masking if we don't want sub-pixel at all. Let's consider this when we write the extraction plugin - as of now the subsets don't do anything but I just wanted at least something to show up in the menu to test that they are created succesfully.

kecnry · 2024-01-29T15:37:17Z

Expose slice number somewhere in the image viewer

I agree this would be useful in some way, but think we might want to consider this as a general feature upstream in cubeviz?

* Adding TPF translator and parser (#75) * TPF viewer (#81) * make use of upstream refactor to override indices in lcviz (#83) * fix creating phase-viewer when TPF is loaded (#86) * Time Selector (adapted version of cubeviz's slice) plugin (#85) * enable clone viewer for image/TPF viewer (#101) --------- Co-authored-by: Brett M. Morris <[email protected]>

kecnry force-pushed the tpf-viewer branch 2 times, most recently from 565583a to 2e23a90 Compare January 19, 2024 19:12

kecnry force-pushed the tpf-viewer branch from 2e23a90 to 3d575f3 Compare January 23, 2024 17:57

WIP: TPF cube viewer

408b89b

kecnry force-pushed the tpf-viewer branch from 3d575f3 to 408b89b Compare January 23, 2024 18:40

kecnry changed the base branch from main to feature-tpf January 23, 2024 18:40

kecnry added 2 commits January 23, 2024 14:30

set correct attributes for axes in cube viewer

5391845

enable mouseover display for cube viewer

761961f

kecnry force-pushed the tpf-viewer branch from 5e94fe8 to 761961f Compare January 23, 2024 19:30

fix bug with subsets

e2bc30a

kecnry mentioned this pull request Jan 26, 2024

lcviz cube support spacetelescope/jdaviz#2678

Merged

9 tasks

kecnry added 3 commits January 26, 2024 12:34

set cube viewer default_class

0df19b2

add comment to explain initial_x_axis

4711c6e

fix tracebacks when unloading all data from viewer

8a703ae

kecnry marked this pull request as ready for review January 26, 2024 18:30

kecnry requested a review from bmorris3 January 26, 2024 18:30

rename TPF viewer to "image"

fb656fa

bmorris3 reviewed Jan 29, 2024

View reviewed changes

don't set default class and leave up to user

35de766

kecnry merged commit c6dbb63 into spacetelescope:feature-tpf Jan 29, 2024
9 checks passed

kecnry deleted the tpf-viewer branch January 29, 2024 15:55

kecnry mentioned this pull request Jan 30, 2024

make use of upstream refactor to override indices in lcviz #83

Merged

kecnry mentioned this pull request Feb 14, 2024

Feature: TPF support #82

Merged

1 task

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

TPF viewer #81

TPF viewer #81

kecnry commented Jan 19, 2024 •

edited

Loading

codecov bot commented Jan 19, 2024 •

edited

Loading

bmorris3 Jan 29, 2024

kecnry Jan 29, 2024

bmorris3 left a comment •

edited

Loading

bmorris3 Jan 29, 2024

bmorris3 Jan 29, 2024

kecnry Jan 29, 2024

bmorris3 Jan 29, 2024

kecnry Jan 29, 2024

bmorris3 Jan 29, 2024

kecnry Jan 29, 2024

bmorris3 Jan 29, 2024

kecnry Jan 29, 2024

kecnry commented Jan 29, 2024

		# TODO: can we vary this default_class based on Kepler vs TESS, etc?
		default_class = KeplerTargetPixelFile

TPF viewer #81

TPF viewer #81

Conversation

kecnry commented Jan 19, 2024 • edited Loading

codecov bot commented Jan 19, 2024 • edited Loading

Codecov Report

Choose a reason for hiding this comment

Choose a reason for hiding this comment

bmorris3 left a comment • edited Loading

Choose a reason for hiding this comment

Expose slice number somewhere in the image viewer

Image viewer WCS

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

kecnry commented Jan 29, 2024

kecnry commented Jan 19, 2024 •

edited

Loading

codecov bot commented Jan 19, 2024 •

edited

Loading

bmorris3 left a comment •

edited

Loading