Projection: plate carree #2856

gzotti · 2022-11-18T15:27:55Z

Description

Ruslan's shader implementation of Milky Way and ZL makes this finally interesting: A window-filling all-sky map. I can imagine an application using Spout where the result goes into another program, as live texture for some object, or just a printed all-sky map.

Fixes #1120

What currently still looks bad at the seam are of course all meshes which are projected before being sent to the GPU:

landscapes
constellation artwork
DSO textures
DSS
HiPS
sky images (from scripting)
An axis-aligned grid has flickering parts on the window edges.
Points like NCP flicker as they should be plotted on the edge but may be projected to the other side by rounding.

Fixing the meshes is not part of this PR. For the time being the application is limited to mostly MW/ZL all-sky maps.

Screenshots (if appropriate):

Type of change

Bug fix (non-breaking change which fixes an issue)
New feature (non-breaking change which adds functionality)
Breaking change (fix or feature that would cause existing functionality to change)
This change requires a documentation update

How Has This Been Tested?

Test Configuration:

Operating system: Windows 10
Graphics Card: Geforce 960M

Checklist:

My code follows the code style of this project.
I have performed a self-review of my own code
I have commented my code, particularly in hard-to-understand areas
I have made corresponding changes to the documentation (header file)
I have updated the respective chapter in the Stellarium User Guide
My changes generate no new warnings
I have added tests that prove my fix is effective or that my feature works
New and existing unit tests pass locally with my changes
Any dependent changes have been merged and published in downstream modules

github-actions · 2022-11-18T15:28:52Z

Great PR! Please pay attention to the following items before merging:

Files matching src/**/*.cpp:

Are possibly unused includes removed?

This is an automatically generated QA checklist based on modified files

src/core/StelProjector.cpp

src/core/StelProjectorClasses.cpp

gzotti · 2022-11-20T15:00:15Z

I think it's ready for discussion. What I see:

great for All-sky maps including just MilkyWay/ZL. Will become also great for horizontal maps when @10110111 (?) can implement shader solutions for the landscapes and other large textured regions. (Not part of this PR.)
With the respective grids activated, grids flicker at the screen edge. This could be remedied by plotting all 4 specific straight lines if this projection is active. Necessary?

Draw extra gridlines along screen edges
Ignore

NCP flickers at the edge in alt-azimuthal mode. Solutions:

Repeat drawing with position set manually to the left edge.
Ignore

10110111 · 2022-11-21T11:00:28Z

Current state of your branch looks like below on my machine. Screen scaling factor is 1.5, if it matters, Qt6 (manually compiled) on Ubuntu 20.04.

gzotti · 2022-11-21T11:56:03Z

OK, yes, scale obviously matters...
Uh, did landscape drop below horizon as well?

10110111 · 2022-11-21T12:20:44Z

Uh, did landscape drop below horizon as well?

They appear to be at different scales: as I pan around, the horizon of the landscape moves as if with a parallax relative to the horizon of the sky dome.

10110111 · 2022-11-21T12:22:16Z

Actually, no, it's the atmosphere horizon that moves. The stars move synchronously with the landscape during panning.

gzotti · 2022-11-21T13:29:43Z

Grmpf... I have now set 125% and 150% scaling on Windows (works on the fly only with built-in screen), and see no issues (apart from obvious lack-of-screen space issues). For whatever reasons, it crashes with 175% when calling the view settings dialog.

10110111 · 2022-11-21T14:52:57Z

works on the fly only with built-in screen

Change on the fly is done purely by the OS. Qt doesn't get a chance to adapt for some reason. Restart of the app should make Qt aware of the change and in charge of it.

gzotti · 2022-11-21T16:47:04Z

Yes, of course I quit Stellarium, changed resolution, and restarted Stellarium. It comes up with the new resolution/bigger GUI but still fills the window. So this is an issue visible only if running on Linux. Getting this screen scaling right on all platforms is a nightmare.

- ProjectionCylinderFill (derived from ProjectionCylinder) - No functional changes yet - also clarified declaration of virtual methods

- Follow Qt's recommendation to code style (Q_DECL_OVERRIDE -> override) - implementation of 3 methods in base class - Removed corresponding overrides in several derived classes

- restore customized screen stretch when going to another projection - better specific map layout

…subclasses.

gzotti · 2022-12-09T12:12:57Z

@10110111 just tried on Ubuntu 22.04.1/Geforce M 580. I set scaling to 125%, and now I see scaling issues in fullscreen, but for all projections. Our view frame is confined to the upper left quadrant. These issues are also present in master!

Is the problem that you see above limited to the new projection?

10110111 · 2022-12-09T12:52:29Z

Master works perfectly fine for me (aside from failure to compile due to GregorianCalendar includes being commented out in JKV calendars).

How do you set the scaling factor? I just use the following environment variables:

QT_FONT_DPI=96
QT_SCREEN_SCALE_FACTORS=1.25

Similarly fine it is when I don't use any variables, while the screen DPI is set to 167 (in this case Qt sets scale factor 167/96, i.e. 1.73958...).

I'm testing with Qt 6.4.0 on Ubuntu 20.04 and Intel UHD Graphics 620.

gzotti · 2022-12-09T13:07:28Z

Ah no, I set scaling in the OS's Settings GUI. (Ubuntu 22.04.1) First, enable Fractional Scaling, then set to 125%, then (to be sure all programs get it) logout/login. U22.04 includes Qt6.2.4.
I can try with the QT env. variables as keep the OS scaling at 100%, but it is probably not what most users would do.

10110111 · 2022-12-09T13:53:47Z

Ah no, I set scaling in the OS's Settings GUI.

Oh well, I don't know what tricks they use. Could you post output of xrandr --verbose after having set fractional scaling?

gzotti · 2022-12-09T14:14:41Z

xrandr_100pct.txt
xrandr_125pct.txt

This is without the QT... variables set.

10110111 · 2022-12-09T15:00:58Z

AFAICT, what you are setting is a scaling, but not the high-DPI scaling. Your setting only enlarges the desktop, downsampling the resulting raster to fit on the screen. Globally, regardless of app awareness, in a raster-downsample mode! This is definitely not something we can support well.

Still, I have tried this on my Intel-based machine, and it works without confining the viewport.

From the output names I figure that you are using nvidia driver, right? Generally this is the correct choice, but historically, nvidia's implementation of XRandR was not the best, so the viewport confinement issue you're seeing might be due to a bug there.

Finally, you should be able to reproduce the problem of this branch being messed up while master looks OK by using the environment variable approach (resetting XRandR scaling factor to 100%).

Is the problem that you see above limited to the new projection?

Just retested, the problem equally affects all projections in this branch.

gzotti · 2022-12-09T15:13:15Z

AFAICT, what you are setting is a scaling, but not the high-DPI scaling. Your setting only enlarges the desktop, downsampling the resulting raster to fit on the screen. Globally, regardless of app awareness, in a raster-downsample mode! This is definitely not something we can support well.

OK, but I have no HiDPI device. This is what the user sees.

Still, I have tried this on my Intel-based machine, and it works without confining the viewport.

From the output names I figure that you are using nvidia driver, right? Generally this is the correct choice, but historically, nvidia's implementation of XRandR was not the best, so the viewport confinement issue you're seeing might be due to a bug there.

Ah, OK. Yes, this is the NVidia driver for Ubuntu.

Finally, you should be able to reproduce the problem of this branch being messed up while master looks OK by using the environment variable approach (resetting XRandR scaling factor to 100%).

Is the problem that you see above limited to the new projection?

Just retested, the problem equally affects all projections in this branch.

Interesting. Changes other than the new projection (variant) were not so many. Let's see...

gzotti · 2022-12-09T23:11:13Z

Interesting. With these variables and without setting scaling via Windows system settings, I see the issue on Windows. Is this what users are supposed to configure, or is this a weird hack which should look broken by design?

10110111 · 2022-12-10T06:29:08Z

Is this what users are supposed to configure, or is this a weird hack which should look broken by design?

These variables have two use cases:

for the developers to test the GUI on different scaling factors,
for the users to override wrong system configuration (e.g. wrong EDID reported by a monitor).

They definitely aren't supposed to look broken.

10110111 · 2022-12-10T09:17:27Z

e624ccd indeed makes things work better.

To make this simplification work, the Vector4<T>::operator*(T) should have a custom template parameter to take right-hand operand as is, without coercing it to T. (Ideally, all these overloaded operators should not force T for scalars.)

gzotti · 2022-12-10T09:55:24Z

Yes, I have concluded the same. But it's not needed for this PR. So, I think this should be ready to merge.

The next big thing would be moving some drawing code for the Landscapes into shader code to avoid the visible gaps. For now, this new mode works best for all-sky maps without labels or lines along the edge.

About HiDPI, as they are becoming more common, we should add a section to the User Guide, e.g. 2.5.1.

10110111 · 2022-12-10T10:10:40Z

The next big thing would be moving some drawing code for the Landscapes into shader code to avoid the visible gaps.

I have this in progress (spherical is done, fisheye almost done, old style not started yet), but I suppose it won't get into 1.2, since I'm getting a new job next week and won't have as much time then.

gzotti · 2022-12-10T10:19:38Z

Great to read, and of course this can wait until 23.1 or later. Congrats on the new job!
@alex-w any comments/objections? Or can we merge this?

github-actions · 2022-12-12T14:44:09Z

Hello @gzotti!

Please check the fresh version (development snapshot) of Stellarium:
https://github.com/Stellarium/stellarium-data/releases/tag/weekly-snapshot

github-actions · 2022-12-25T16:05:13Z

Hello @gzotti!

Please check the latest stable version of Stellarium:
https://github.com/Stellarium/stellarium/releases/latest

gzotti added the feature Entirely new feature label Nov 18, 2022

gzotti added this to the 1.2 milestone Nov 18, 2022

gzotti requested review from alex-w and 10110111 November 18, 2022 15:27

gzotti self-assigned this Nov 18, 2022

10110111 reviewed Nov 18, 2022

View reviewed changes

src/core/StelProjector.cpp Outdated Show resolved Hide resolved

10110111 reviewed Nov 18, 2022

View reviewed changes

src/core/StelProjectorClasses.cpp Outdated Show resolved Hide resolved

gzotti marked this pull request as ready for review November 20, 2022 15:00

gzotti added 13 commits December 9, 2022 11:09

First changes towards frame-filling image

58e7b4e

Simplicications

d3240a6

Added new projection

23690bc

- ProjectionCylinderFill (derived from ProjectionCylinder) - No functional changes yet - also clarified declaration of virtual methods

Inhibit FoV and panning for the new projection.

8d714e8

Minor cleanup

abd21ae

- Follow Qt's recommendation to code style (Q_DECL_OVERRIDE -> override) - implementation of 3 methods in base class - Removed corresponding overrides in several derived classes

Fix indentation

64d9cc3

Equirectangular projection with automated windows stretching

7d7728b

Slight restructuring

c2f5c11

Avoid scaling problem

2c9a29b

Remove inherited overrides

d5b2f4e

Minor improvements

7c9ba31

- restore customized screen stretch when going to another projection - better specific map layout

Cleanup and fix tests

0405b2b

Use default implementation in base class and avoid overrides in many …

c1f7d38

…subclasses.

Better description

e327468

gzotti force-pushed the projection_plate_carree branch from eb42335 to e327468 Compare December 9, 2022 10:11

Revert a supposed code simplification

e624ccd

alex-w approved these changes Dec 10, 2022

View reviewed changes

gzotti merged commit fd8cf36 into master Dec 10, 2022

gzotti deleted the projection_plate_carree branch December 10, 2022 11:50

alex-w added the state: published The fix has been published for testing in weekly binary package label Dec 12, 2022

alex-w removed the state: published The fix has been published for testing in weekly binary package label Dec 25, 2022

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Projection: plate carree #2856

Projection: plate carree #2856

gzotti commented Nov 18, 2022

github-actions bot commented Nov 18, 2022

gzotti commented Nov 20, 2022

10110111 commented Nov 21, 2022 •

edited

Loading

gzotti commented Nov 21, 2022

10110111 commented Nov 21, 2022 •

edited

Loading

10110111 commented Nov 21, 2022 •

edited

Loading

gzotti commented Nov 21, 2022 •

edited

Loading

10110111 commented Nov 21, 2022

gzotti commented Nov 21, 2022

gzotti commented Dec 9, 2022

10110111 commented Dec 9, 2022

gzotti commented Dec 9, 2022

10110111 commented Dec 9, 2022

gzotti commented Dec 9, 2022

10110111 commented Dec 9, 2022 •

edited

Loading

gzotti commented Dec 9, 2022

gzotti commented Dec 9, 2022 •

edited

Loading

10110111 commented Dec 10, 2022

10110111 commented Dec 10, 2022

gzotti commented Dec 10, 2022

10110111 commented Dec 10, 2022

gzotti commented Dec 10, 2022

github-actions bot commented Dec 12, 2022

github-actions bot commented Dec 25, 2022

Projection: plate carree #2856

Projection: plate carree #2856

Conversation

gzotti commented Nov 18, 2022

Description

Screenshots (if appropriate):

Type of change

How Has This Been Tested?

Checklist:

github-actions bot commented Nov 18, 2022

gzotti commented Nov 20, 2022

10110111 commented Nov 21, 2022 • edited Loading

gzotti commented Nov 21, 2022

10110111 commented Nov 21, 2022 • edited Loading

10110111 commented Nov 21, 2022 • edited Loading

gzotti commented Nov 21, 2022 • edited Loading

10110111 commented Nov 21, 2022

gzotti commented Nov 21, 2022

gzotti commented Dec 9, 2022

10110111 commented Dec 9, 2022

gzotti commented Dec 9, 2022

10110111 commented Dec 9, 2022

gzotti commented Dec 9, 2022

10110111 commented Dec 9, 2022 • edited Loading

gzotti commented Dec 9, 2022

gzotti commented Dec 9, 2022 • edited Loading

10110111 commented Dec 10, 2022

10110111 commented Dec 10, 2022

gzotti commented Dec 10, 2022

10110111 commented Dec 10, 2022

gzotti commented Dec 10, 2022

github-actions bot commented Dec 12, 2022

github-actions bot commented Dec 25, 2022

10110111 commented Nov 21, 2022 •

edited

Loading

10110111 commented Nov 21, 2022 •

edited

Loading

10110111 commented Nov 21, 2022 •

edited

Loading

gzotti commented Nov 21, 2022 •

edited

Loading

10110111 commented Dec 9, 2022 •

edited

Loading

gzotti commented Dec 9, 2022 •

edited

Loading