After a quick check it seems that the celestial background sphere is still using the old pre-2016 texture mechanism. This limits the maximum resolution a bit, but should still support significantly higher than is currently provided.

I had a look at the web site you referenced, but I didn't see an option to download the entire global source image. It appears to be an interactive tool that dynamically generates an image from source data at a specific location request. Did I miss something?

This seems to be the place to download imagery, although there doesn't seem to be an option to download the entire source image, but it should be doable by selecting a very high resolution and wide field of view on the right side Extended parameters: toolbar. I'll try to dig around and find if there's a more streamlined option out there.

I noticed the previous link wouldn't work when changing the RA and declination, this one should do the trick.

This form doesn't seem to work very well with large fields of view. In particular, the "tangent" projection (the only one offered) causes serious distortions at the image borders, and it also produces artefacts (blank areas) in the image. I assume "tangent" simply projects the celestial sphere onto a flat tangent plane at the central point.

Here's the deal: if you manage to extract a global image at higher resolution in cylindrical projection (galactic longitude and latitude), I'd be happy to process it into a celestial background texture and include it in the next beta commit. Alternatively, you could do the processing yourself - it should just be a matter of running pltex and making sure that the result is correctly aligned when loaded into Orbiter. You could then let me have the tex file to include it in the beta, or upload it yourself as an addon to orbithangar.

I wasn't able to find any other type of projection for Wikisky so I looked for some alternatives and found this rather interesting which is 16K, and the second best so far is this 6K panorama (the latter was already made into an addon I believe).

How did Dr. Martin do it the first time?

Hello martin,
out of curiosity I would like to know (as Abdallah), how to process an image to a "csphere.tex".
Can you post a step-by-step guide to convert the example file[1]?


[1] "Indicatrices_of_Distortion.png"
It's a .png, I know. Converting to BMP (with or without alpha?) might be necessary

Here is the recipe:

• convert to BMP (say 24bit, there shouldn't be an alpha channel)
• Run Utils\pltex to generate an old-style tex file, like for any planetary surface (global level 1-8, optionally higher-res patches)
• Copy tex file to Textures\csphere
• Edit Config\CSphere\bkgimage.cfg to add a new entry in the same format as he existing ones.
• Run Orbiter, and select the new texture in Visual effects\Celestial sphere\background

Above I said that the image should be mapped to the ecliptic frame, but I just realised that this doesn't seem correct. I guess it is in fact the galactic frame (i e. plane of the milky way horizontal along the centre of the image), but I need to check the sources tonight to confirm. In any case, it's probably the same frame in which these images are usually published (check for example http://blogs.cardiff.ac.uk/chromoscope/download)

Thanks (a lot)

While Utils\pltex only supports detail level up to 8
and assuming 256 x 256 tiles,
it doesn't seem to make any sense to download / convert images larger than 8192 x 4096,
right?

While playing around with this I've made myself a Test-Pattern (checkers)
...and couldn't resist to post it.

Although I tend to call it "Amiga" for some reason ;-)

Hm, yes, you are right. I checked the code, and the bit responsible for loading higher resolution tiles consists of a big //TODO!
I probably won't implement that - if anything, I would update the code to use the new quadtree format.

Btw the source image for generating the celestial background should indeed be aligned with the galactic frame.

Good to know! So if it looks like this:

it's fine.

Should be, but it's always a good idea to make sure that the texture lines up with the star database (i.e. the stars rendered as individual pixels). If I remember correctly, I hand-tweaked the rotation matrix to optimise the fit. If you find a small discrepancy, I can try to improve the transformation. If it's a big discrepancy, then probably something went wrong with the texture generation

No worries!
I believe the current dds2.tex is already "encoded to the max" (level 8 from a 8192 x 4096 source)...
...so digging out a bigger source image has to wait until it can be transformed to higher level .tex anyway.

Thanks for the insights, by the way.