Surveys: diagonal stripe at specific scale

[axd1967]

Expected Behaviour

Actual Behaviour

Describe or maybe attach a screenshot?

Zoom in: observe the diagonal stripe through M42

Zoom out: Diagonal gone

Steps to reproduce

1. make sure no survey is active
2. select M42
3. zoom to FOV around 1 deg (CTL-ALT-9)
4. in Surveys, select "AKARI FIS color WideL"
5. observe diagonal stripe (see sshot above)
6. zoom out to FOV 10 deg (CTL-ALT-6)
7. observe the stripe is filled normally
8. as from now, zooming in (eg CTL-ALT-9) no longer shows this artifact

System

• Stellarium version: 0.20.4
• Operating system: KUbuntu 20.04
• Graphics Card: <Manufacturer (likely Intel, NVidia, AMD?), Model (HD, Geforce, Radeon..., with model number), driver version?>
• Screen type (if applicable): Resolution, HighDPI, scaling

Logfile

If possible, attach the logfile log.txt from your user data directory. Look into the Guide for its location.
log.txt

[axd1967]

Independent of the chosen Survey - happens with other surveys too

[alex-w]

Alex, did you compare visualization of HiPS with Aladin for example?

[axd1967]

I tried a few surveys with the online version (probably not good enough, I should try the desktop app):

It looks like maybe there is a small bug in Stellarium on handling the sides of what probably are tiles. The "stripes" are aligned to an internal coordinate system.

When I switch between az/eq mode, the display follows:

(look for the faint green az/eq indication in the center of the shot)

[guillaumechereau]

The black and red lines seem to be from the original survey, but there is also clearly a rendering bug, probable in the mesh generation of the HiPS during rendering. I cannot reproduce it on my machine though. It might be something that happens while the survey is still loading, or maybe only with some specific screen resolution / density.

[guillaumechereau]

@axd1967: are you using the default Stereographic projection ?

[axd1967]

The black lines are not an issue, those are clearly data artifacts.
But the transparent band draws the attention, even if the workaround is to zoom out, then in.

I am using stereo, yes; but it also happens in perspective.

If you feel for proposing a patch, I can compile it and post the logs/results.

[alex-w]

@axd1967 please check the current master - @10110111 made refactoring for the the rendering of HiPS...

[gzotti]

I no longer see the glitches. However, in high zoom this is coming to a crawl now on my Geforce 3070. No idea why, system load is about 15%. (OK, hints at single-thread issue...) The HiPS (Akari color) takes a long time to finish loading, then crawls at 1.7fps. Without HiPS, I see 127fps.

...Uhm, weird. The 2MASS is much faster. Must lie in the AKARI provider itself.