thanks Georg
to clarify, for 0.13.1, the moon and planets appear as textured spheres only
when they exceed a pixel in size. In the startup screen, with a 60deg field of
view, only sun and moon show any image. Their legends are present, as are the
brighter stars and planets. Given enough zoom, the spheres eventually appear.
Jupiter appears as a single pixel at 5deg FOV, but is invisible at 6deg FOV.
Mars does not appear until about 50min FOV. And so on.The stars only appear as
diffuse foggy areas, even given large zoom-in. There is no central bright dot.
Eventually, some stars show (like Antares) but these are drawn as part of a
large rectangular sheet "galaxy image" which suddenly appears at greater zoom.
Contrast this with the startup view of 0.12.4. Even at 60deg FOV, planets
appear as oversized discs (e.g. Jupiter is about 10pixels on-screen, which
represents (say) 35 minutes of subtended arc (on 1920x1080 screen.) But it's
actually way smaller than that, less than 40 seconds arc.) Zooming in starts
to enlarge this once the FOV matches the drawn size. The brighter stars also
show as bright discs of appropriate colour, a few pixels in diameter.
It looks like the earlier code prescribes a minimum drawn disc size so that we
do not lose sight of the stars and planets. Without this minimum size, there
cannot be enough brightness from an on-screen object (say Sirius, Venus,
Jupiter) to show it up as the brightest object in our FOV. Stars do not
subtend any angle, they are point-sources of light, not reflecting surfaces, so
maybe the graphics do not understand drawing them, as they are always sub-pixel
scale objects.
It's only a hypothesis! I've only started looking at it this week, and my
knowledge of shaders and 3d-graphics would fit on a small postage stamp.
cheerGraham
On Wednesday, 19 November 2014, 16:30, Georg Zotti
<georg.zo...@univie.ac.at> wrote:
Hi!
The message on glProgramParameteriEXT may not be completely bad. In the
last half year we are hunting down Qt5-/OpenGL2+-related errors on various
graphics systems, and this was one of the possible issues. However your
logfile is free from relevant messages.
But I don't understand your last mail. Do you see the moon and planets as
textured spheres, and do you see stars, or not? If something is less than
one pixel in size, I would not expect to see it.
If you have seen the stars as one-pixel dots earlier, you may have had a
flag active in config.ini which had been defaulted to false long ago, and
is no longer used in 0.13:
[stars]
flag_point_star = false
I don't know if this makes a difference in your description. As far as I
can see, it applies only to 0.12 and earlier in --safe-mode (OpenGL1
mode).
So, in 0.13 all stars are rendered as small blobs of different brightness
and diameter. Think of seeing disks. If you want pinpoint stars, you can
set absolute and relative scales to make them smaller. In case you only
see the dim star halo, not a whitish center, we have a problem.
G.
On Mi, 19.11.2014, 16:54, Graham Bradley wrote:
> Aha!I may be answering my own question here, but an earlier message
> referred to bug #1383046, and it does look consistent with my issue.
> So I have tried the suggested:
> stellarium --dump-opengl-details
> On the 0.13.1 version with FreeBSD and radeon driver
> and the output file is attached.
> I notice there's a line which sounds like a problem....
> "glProgramParameteriEXT cannot be resolved here. BAD!"
> and that suggests (maybe) something left out of the GL library. What
> would it do to the output?
> cheersGraham
>
>
> On Wednesday, 19 November 2014, 15:22, Graham Bradley
> <gbrad...@rocketmail.com> wrote:
>
>
> thanks for the idea, I have run with FreeBSD-10.1 and latest xorg and vt:
> In each case I switched off atmospherics and horizon from the view.
> 1. stellarium 0.12.4 seems to show everything, and zooms in nicely
> then installed back the later pkg to try that:
> 2. stellarium 0.13.1 runs, can zoom in fine on galaxy images, sun, moons
> and planets, but nothing shows up when it is less than a pixel.Notice that
> Moon and Spica are close at the moment, zooming Moon is ok, as it's always
> at least a pixel in size ~but Spica never shows as a bright dot, but lots
> of zoom gives a 'foggy ball'.
> It is as if the problem is "what does qt show when a bright object is
> smaller than a pixel?"
> As per your suggestion I have attached the two logfiles.
> cheersGraham
>
>
> On Wednesday, 19 November 2014, 14:10, Georg Zotti
> <georg.zo...@univie.ac.at> wrote:
>
>
> So, you have spotted a difference?
>
>
> 1) Moving from Qt4 to Qt5 brought a huge change.
>
> 2) It is our favorite FAQ since 0.13.0 is out:
>
> https://answers.launchpad.net/stellarium/+faq/2570
>
> 3) I am surprised the Radeon driver for this modern iGP does not like the
> GLSL1.3 planet shaders. As always, we could get a bit more info with a
> logfile.
>
> Regards,
> Georg
>
> On Mi, 19.11.2014, 14:51, Graham Bradley wrote:
>> My system is AMD HD7660D processor, integrated graphics processor.
>> This
>> week I updated from FreeBSD 10.0-RELEASE using vesa driver to
>> 10.1-RELEASE using the updated radeon driver, and vt console.
>> Behaviour with stellarium has changed. Has anyone else come across
>> this?
>>
>> When I ran stellarium 0.13.1 on 10.0 with the 'old' graphics xorg
>> packages
>> and vesa driver, the sky was drawn correctly, although a bit slowly at
>> about 2 fps;
>> After upgrading FreeBSD to 10.1, but still using the vesa graphics
>> driver
>> (but all the updated release software packages), the sky is drawn
>> correctly and visibly faster, say 10 fps, then I run stellarium in 10.1
>> with the new radeon driver, the stars and planets in the sky are all
>> dark, though lines and legends are drawn ok (and faster of course.)
>>
>> i.e. identical software, except the graphics driver. Different picture!
>> I tried raising the relative/absolute size of stars, that only gave me a
>> foggy area around where the star should be, but no bright point.
>>
>> So I tried regressing stellarium to 0.12.4. That's the qt4 version.
>> this
>> draws at speed, and zooms in just fine on FreeBSD 10.1. So my
>> workaround
>> is to use the updated BSD and xorg, but the regressed stellarium
>> 0.12.4. I
>> wonder if anyone has spotted a difference between qt4 and qt5
>> behaviour.
>> If there's any test I can do that may help illuminate the issue, just
>> let
>> me know.
>>
>> I am fine with C/C++ programming, but sadly, OpenGL and shader
>> programming
>> is counted amongst the 'dark arts' for me at present.
>> regardsGraham
>>
>>
>
>
>
------------------------------------------------------------------------------
Download BIRT iHub F-Type - The Free Enterprise-Grade BIRT Server
from Actuate! Instantly Supercharge Your Business Reports and Dashboards
with Interactivity, Sharing, Native Excel Exports, App Integration & more
Get technology previously reserved for billion-dollar corporations, FREE
http://pubads.g.doubleclick.net/gampad/clk?id=157005751&iu=/4140/ostg.clktrk
_______________________________________________
Stellarium-pubdevel mailing list
Stellarium-pubdevel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/stellarium-pubdevel
