feat: layers manipulations

[Xen0Xys]

Fixes #76

[bmatthieu3]

This is just an idea I have that would allow the user to compose its HiPS stack fast with several HiPSes with colormap etc...

It would be nice to have a new traitlet targeting the global aladin image stack (HiPS and FITS) stack. I see it as a python list of object such as:

aladin.stack = [{
    "url": "https://alasky.cds.unistra.fr/SDSS/DR9/band-g/", # HTTPS url pointing towards the HiPS
    "opacity": 0.2,
    "colormap": "rainbow",
    "stretch": "asinh", # apply stretch function to the pixel before the colormap
    "format": "fits", # format of the tiles
    "minCut": 100,
    "maxCut": 2000
  },
  {
   "id": "CDS/P/2MASS/color",
    "opacity": 0.4,
    "format": "png",
  }
]

This list describes a view with 2 HiPS, 2MASS on top of SDSS9/g. For one HiPS, one should only accept "id" or "url" for identifying an HiPS. If both are given, we would take ID and return a warning that the url has not been taken into account for retrieving the HiPS. Eventually, we could only accept id and the user could pass an HTTPS url to the id keyword too. This is better I think.

Something should be done to sync the overlay_survey and overlay_opacity traitlets with this one (i.e. aladin.stack[1].url/id and aladin.stack[1].opacity)

[bmatthieu3]

You could even put some astropy image HDU in that stack list:

aladin.stack = [
  "CDS/P/2MASS/color", # or only the ID
  "https://mastimg.stsci.edu/surveys/hips/DSS/DSSColor/", # only an url pointing to the root of an HiPS
  {
    "url": "https://alasky.cds.unistra.fr/SDSS/DR9/band-g/", # HTTPS url pointing towards the HiPS
    "opacity": 0.2,
    "colormap": "rainbow",
    "stretch": "asinh", # apply stretch function to the pixel before the colormap
    "format": "fits", # format of the tiles
    "minCut": 100,
    "maxCut": 2000
  }, # A HiPS with more optional parameters
  astropy.io.fits.ImageHDU # cf https://docs.astropy.org/en/stable/io/fits/api/images.html#astropy.io.fits.ImageHDU
]

The above list describes a stack of 4 layers:

1. The base is 2MASS/color
2. On top the DSS hosted by MAST
3. On top is a HiPS of SDSS9/g with opacity 20% and a specific colormap
4. On top of all the HiPS, a FITS Image HDU astropy object.

[ManonMarchand]

@bmatthieu3 : ready to merge?

[bmatthieu3]

@ManonMarchand @Xen0Xys - I think we also need to consider sync the overlay_survey and overlay_survey_opacity traitlet. If you add a hips survey with the add_hips method we should update the value of overlay_survey, i.e. a call to add_hips should be sync with overlay_survey (the first overlay should match overlay_survey).

aladin.add_hips('https://alasky.cds.unistra.fr/DES/DR2/CDS_P_DES-DR2_ColorIRG/')
aladin.add_hips('SDSS9/color')
aladin.overlay_survey # should return the id/url from the last overlay added i.e. 'SDSS9/color'

[Xen0Xys]

@ManonMarchand @Xen0Xys - I think we also need to consider sync the overlay_survey and overlay_survey_opacity traitlet. If you add a hips survey with the add_hips method we should update the value of overlay_survey, i.e. a call to add_hips should be sync with overlay_survey (the first overlay should match overlay_survey).

aladin.add_hips('https://alasky.cds.unistra.fr/DES/DR2/CDS_P_DES-DR2_ColorIRG/')
aladin.add_hips('SDSS9/color')
aladin.overlay_survey # should return the id/url from the last overlay added i.e. 'SDSS9/color'

Currently, the HiPS is added as a layer and not an overlay:
this.aladin.addNewImageLayer(hips);
It seems counterintuitive to sync the overlay with a non-overlay layer, doesn't it?

[ManonMarchand]

I also find it strange. If we have two methods then why not leave the users the freedom to manipulate the overlay separately?

[bmatthieu3]

addNewImageLayer adds a new overlay survey so I do not get your points. An overlay has a name, this name is called layer, that is maybe why it is confusing. Internally, addNewImageLayer calls setOverlayImageLayer with your survey and a arbitrary layer name.

[Xen0Xys]

The overlay behavior change between addNewImageLayer and setOverlayImageLayer.
With the first one, the second call to add_hips add a third overlay on top on the second.
With the second one, the second call to add_hips replace the previous added hips.

Which behavior do you want for the implementation?
In addition, how to manage the overlay_survey_opacity traitlets? Cause if we use the first behavior, I don't know if getOverlayImageLayer return something.

[bmatthieu3]

@ManonMarchand @tboch - I guess my point is: what do we want to do with the 'overlay_survey' traitlet since it is now possible to add multiple hips surveys with this method. For example, if a user adds 3 hips by calling 3 times add_hips and then change the overlay_survey traitlet what should it do ? Different possible cases:

• overlay_survey changes the very top layer (the 3rd added hips)
• overlay_survey changes the first added layer, so we may not see the change on screen because there are 2 more on top.
• overlay_survey adds a 4th hips
  What do you think ? I tend to prefer the first case

I also ask myself if it would not be better to have a 'set_hips(hips, name)' instead of add_hips so that the user controls better the stack i.e. it is possible to change any layer like that by specifying its name. If no name is given, the first base layer is taken.

[tboch]

@bmatthieu3 I think overlay_survey should reflect the reality, and become an array, thus overlay_surveys

[Xen0Xys]

@bmatthieu3 I think overlay_survey should reflect the reality, and become an array, thus overlay_surveys

So, we need an event when the survey list is updated on the AL UI, because otherwise, we can't track the overlay list changes corrrectly.