Mercator areas not shown and used correctly

[ameraner]

Problem

It seems that areas in the mercator projection are not shown and used correctly.
For example, when using the area

CE_1160x800:
  description:   Central Europe 1160x800
  projection:
    proj: merc
    lat_0: 50
    lon_0: 14.5
    ellps: WGS84
  shape:
    width: 1160
    height: 800
  area_extent:
    lower_left_xy: [-1805152.04,  5172213.35]
    upper_right_xy: [1811322.74,  7660679.05]

just a black screen appears on the main window:

If I try to resample a LEO granule that is over France to this area, and then switch to plate carree, I see that the data has been resampled to Mongolia:

For reference, this is the plain Satpy result:

Additional Information

• Not sure if it's related, but there is a code comment mentioning mercator issues workarounds here:
  

  sift/uwsift/view/visuals.py

  Lines 941 to 942 in f5e584c

  	# Clip lats to +/- 89.9 otherwise PROJ.4 on mercator projection will fail
  	np.clip(vertex_buffer[:, 1], -89.9, 89.9, out=vertex_buffer[:, 1])

• logs do not show anything particular, apart from the usual

2023-06-12 16:22:06 DEBUG document:change_projection:L125 Changing projection (area definition) from 'MSG SEVIRI FES 3km' to 'CE_1160x800'
/tcenas/proj/optcalimg/OFTs/SIFT/conda-packaged_v2.0.0b0/SIFT_2.0.0b0/lib/python3.11/site-packages/pyproj/crs/crs.py:1286: UserWarning: You will likely lose important projection information when converting to a PROJ string from another format. See: https://proj.org/faq.html#what-is-the-best-format-for-describing-coordinate-reference-systems
  proj = self._crs.to_proj4(version=version)

[djhoese]

Off the top of my head, I don't think a mercator projection with a non-equator lat_0 is valid in PROJ. Right?

[djhoese]

https://proj.org/en/9.2/operations/projections/merc.html

No mention of lat_0. Not that that explains it, but yeah, maybe affecting it.

[arcanerr]

Is mercator implemented in SIFT or exactly speaking VisPy at all? If I remember correctly, VisPy has an own implementation of the projections and at least equidistant cylindrical (eqc) is missing.

[djhoese]

Vispy does not implement any geographic projects. I wrote them into SIFT in shader language by modifying a OpenCL (not OpenGL) version of PROJ if I remember correctly. The main point here is that mercator with a lat_0 other than 0 is not very standard and is not implemented most likely.