[Marble-devel] Review Request: Added colors to stars

Tue Dec 25 12:03:08 UTC 2012

> On Dec. 25, 2012, 11:15 a.m., Torsten Rahn wrote:
> > Hi Timothy, awesome work. But there are a few severe issues with this patch:
> > 
> > - I can't find any notice on licenses or copyrights on the vendian.org page. Maybe you can? We need to know whether this work is available in the public domain or whether the author demands attribution (CC by ...). He also needs to be fine with commercial usage. So ideally his work would be provided under the public domain or CC by SA 3.0. Since I don't find anything like that I just assume the worst and consider this work "All rights reserved" which prevents us from shipping it. So we need a nice e-mail from him stating the terms of use. Otherwise it's a show-stopper.
> > 
> > - The current appearance of the stars with the color values is quite clownesque. The reason is that point-like stars have a color appearance that is different from the appearance of discs (which we use to get the huge brightness range across). This issue would also have existed with my colors to some degree (although I chose the colors in a way that makes the colors less apparent and hence more realistic).
> > In order to "mellow" these clownesque colors and in order to make the stars look a bit more natural the next step would be to use pixmaps and colorize those. I did this as part of the preparation for task 
> > 
> > http://www.google-melange.com/gci/task/view/google/gci2012/8122209
> > 
> > See 
> > 
> > http://devel-home.kde.org/~tackat/starmockup.png 
> > 
> > for a mockup with proper sizes and proper color shades (take the current state as a comparison: http://devel-home.kde.org/~tackat/starnow.png )
> > 
> > This colorization is only applied to the glow / newton spikes behind while the star-pinpoint center just has shades of white. In total this provides a more natural color appearance (for non-point depiction. I don't doubt that the vendian.org one works okish for point depictions).
> > Since your approach currently allows for an almost infinite range of colors (which a normal user won't be able to distinguish anyways) I'm not sure how to colorize the glow inside the pixmaps efficiently.
> > 
> > Due to copyright issue I'm not quite sure how to further deal with this task. Nevertheless totally awesome work but we need to resolve these issues somehow :-/ 
> >
> 
> Torsten Rahn wrote:
>     BTW: Regarding the mockup: The "spike" star pixmaps cover only stars down to magnitude 2. The plan would be to also use pixmaps for magnitudes 3-5. For these fainter stars no spikes would be drawn instead there would only be a slight color tinted glow/halo around an otherwise mostly white-shaded disc.
>     What needs to be clear is that the images always consist of two parts combined: a colorized glow/spike part and a white-shaded starry part (colorizing a single pixmap will not result in the same vivid appearance, so this needs to be done separately in layers).

Oh and to make it clear: I'm all fine with going the vendian.org path as long as there is a good solution for resolving both issues mentioned above. 

I still got a question regarding the B-V value based approach: do you know roughly what percentage of B-V data was missing? From what I've seen all the bright clearly colorful stars were available. So I would have suggested to just render the stars for which color information wouldn't have been available just white. So in case the vendian.org path doesn't seem to be viable (or too time-costly for now) we could still go for my original suggestion. 

Oh and if you haven't done it yet and in case you have clear skies over there: just go out and watch Orion. You'll see that the colors of stars are a lot more subtle than what is rendered with your current patch. Most people have a hard time assigning or even seeing the colors. 
It's just some examples where it really becomes totally obvious (like with the double star Albireo seen through a telescope http://de.wikipedia.org/wiki/Albireo in summer, or Alamak http://en.wikipedia.org/wiki/Gamma_Andromedae during autumn ). So I'd really like to have something that is about as subtle as the real appearance and not overly colorful :-) 

- Torsten

-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
http://git.reviewboard.kde.org/r/107906/#review23972
-----------------------------------------------------------

On Dec. 25, 2012, 5:52 a.m., Timothy Lanzi wrote:
> 
> -----------------------------------------------------------
> This is an automatically generated e-mail. To reply, visit:
> http://git.reviewboard.kde.org/r/107906/
> -----------------------------------------------------------
> 
> (Updated Dec. 25, 2012, 5:52 a.m.)
> 
> 
> Review request for Marble.
> 
> 
> Description
> -------
> 
> Colors added to stars according to their spectrum description given in the catalog.
> 
> 
> Diffs
> -----
> 
>   data/stars/stars.dat 40def72 
>   src/plugins/render/stars/StarsPlugin.h d979629 
>   src/plugins/render/stars/StarsPlugin.cpp 4506fd4 
>   tools/stars/starcolor.dat PRE-CREATION 
>   tools/stars/stars.cpp 5cdd997 
> 
> Diff: http://git.reviewboard.kde.org/r/107906/diff/
> 
> 
> Testing
> -------
> 
> Compiled and ran, checked to see if it worked.
> 
> 
> Thanks,
> 
> Timothy Lanzi
> 
>

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.kde.org/pipermail/marble-devel/attachments/20121225/96f49ff6/attachment.html>