rectangular FOV wrong for other than very small angles

[axd1967]

Expected Behaviour

The angle measurement tool should measure the exact sizes that have been set in the rectangle FOV.

Unless I'm mistaken, this is a known issue.

Actual Behaviour

The rectangular FOV does not take projection into account. (Exception: in cylindrical projection, the vertical dimension is respected)

[gzotti]

it's a rectangle drawn with IIRC the screen center scale. Using such a big square must lead to errors. We should probably limit this field to 5°.

[axd1967]

it's a rectangle drawn with IIRC the screen center scale. Using such a big square must lead to errors. We should probably limit this field to 5°.

No. Because

1. Stellarium should correctly display angularly defined topocentric rectangles of any size (yes, indeed, also of 180x180!)
2. artificially limiting the user where there is no fundamental issue is not a good idea
3. 7-10 degrees is a typical binocular FOV

Note that this can also serve as a solution for #2882

[gzotti]

I knew you would say this. Some limits must be defined to avoid errors. As always, you are free to develop a PR that overcomes this trivial simplification that is not so trivial to solve. (All-1) seem to be able to live with a quick estimate. To simulate instruments, use the Oculars plugin which was designed for this. (It also just shows a rectangle, but with info.) Please send a link to a bino with rectangular fov.

[axd1967]

I'm aware this is not an easy issue. This is a good issue for an advanced developer, so please don't close it otherwise it will slip under the radar of keen developers.

And it's not because "all-1" can live with it, that it can't be mentioned. It is often precisely that single voice that might point out a problem. Remember "the Emperor has no clothes"?

I don't want to simulate instruments; see #2882 for a very real use case.

Remember: if Stellarium claims to be accurate, this is part of the deal...

And Georg, can we drop the silly remarks on rectangular FOV of bino's? It is hurting, you want to make me look stupid. You should know better by now... because it should be obvious that I'm referring to the rectangle that spans the FOV of a bino. Let's engage some brains before entering the room, please!

[alex-w]

The #2882 has bad description and this is very untypical use case, at least I don’t know people who want see the sky through tonnel. For correct solving #2882 you need to introduce the new type of landscape.

[gzotti]

The button next to the rectangular fov controls a circular fov. Now who engages brains? You can surely report 600 more tiny issues, from which we must still sort out those worth solving. This is a waste of our time. Thanks, we are still busy.

[axd1967]

Discussions end where we get personal.

@xalioth

[alex-w]

@10110111 would you look at this issue? Probably the solution will be similar to your solution for Oculars plugin.

[10110111]

What should this rectangle actually represent? Possible interpretations:

• the field of view of a rectilinear lens whose central horizontal and vertical FoV is specified by the settings;
• a quadrilateral with the sides having the specified FoV (this is more in line with the screenshot in the OP);
• something else?

What is the use case of this marking?

[axd1967]

the FOV box should follow the same mathematical behaviour as the angular tool. No lenses involved here, that belongs in the Oculars module.

[10110111]

You see, it's a vague description: there's no such spherical rectangle that would have all "horizontal" great arcs equal and all vertical great arcs equal. The best one can hope for is that the sides have the required FoV, or a pair of central lines have the required FoV, or something else, but not all the lines of the same orientation. So I ask once again: what is the use case?

[alex-w]

By the fact, this tool was added as very simple visualizer for rectangular FoV of some optical setup with CCD, where we have known only value of field of view and neither other.

[axd1967]

the primary issue is the discrepancy between angular tool and RFIV. The angular tool seems to define the correct limits of the rectangle.

Here the bug is visible for a 90x20 RFOV in stereographic projection

• the angular tool correctly reports 90d between N and E
• the RFOV tool (red) seems to cut off short (so the demanded 90 FOV is not actually met)

use case: see #2882. (which is NOT to define yet another landscape). (plus: Stellarium should be accurate.)

btw - it makes no sense to limit circular FOV (CFOV) to 28 while limiting rectangular FOV (RFOV) to 180x180.

[alex-w]

use case: see #2882. (which is NOT to define yet another landscape). (plus: Stellarium should be accurate.)

This use case is not use case by the fact, because it is not related to rectangular FOV of optical setup with CCD.

btw - it makes no sense to limit circular FOV (CFOV) to 28 while limiting rectangular FOV (RFOV) to 180x180.

As we said before, these limits have objective reasons.

[gzotti]

What's the point of a 180° rectangle then? Limit to ~20° to avoid the obvious distortion-caused error to keep the "simple" solution, or copy/adapt "sensor" code from recent work by @10110111 .

[10110111]

I have an adaptation of the sensor code, but we need to make the decision whether it's OK to use the same kind of rectangle (judging by what @alex-w says, it's OK). It won't have the same angular sizes on the borders as it does in the middle between the borders (and it can't in principle).