My (completely biased) opinion:

> *2.6.0
> In 2.6.0 we had only the choice between default sky or Zan's sky shader.
> In both variants shaders on aircraft and on terrain worked, but there  
> was the problem with horizon.
>
> It gave us pretty cool looking skies, as you can see on the 2.6.0  
> gallery.

I think the correct way to phrase this is: Under certain conditions (correct 
view angle, correct weather, correct time of the day) you could make pretty 
plausible screenshots. Under most real flight conditions, it gave visual 
artefacts ranging from mildly annoying to outright wrong.

> *2.8.0
> Now in 2.8.0 we have the choice between the default sky, or your  
> Lightfield shader, also named as Atmospheric scattering.
> I disregard Rembrandt, as most users won't use it yet for different  
> reason.

Rembrandt has precisely the same issues - not all shaders work with it, so 
using Rembrandt you will not have some shaders available.

I've explained that a (subjective) 1000 times, but let me try again: There is 
no such thing as 'the skydome having just a problem with the horizon' that can 
be fixed independently of aircraft effect shaders.

If you want to match the horizon, you have to modify the terrain shader because 
terrain needs to blend into skydome always under all conditions. If you modify 
the terrain shader, you have to modify the default model shader, otherwise 
distant models will have different light and fog and stick out. If you modify 
the default model shader, effect shaders of your aircraft will break unless you 
take care to modify them as well. 

> With Lightfield shader also named as Atmospheric Sky Scattering enabled:
> we get pretty cool looking skies, a correct horizon but terrain colors  
> which looks somehow pale

Yes. That's realistic physical fogging ~ exp(-x/d) as opposed to non-physical 
fogging ~exp(-x^2/d^2) as done in the default scheme - complain to nature, I 
didn't invent the real fog function :-) In the atmospheric light scattering 
scheme (as in reality) the near zone is a lot more fogged than in default 
Flightgear for the same visibility range (defined as 'range out to which you 
can recognize an object). As a result, the overall impression is a lot more 
pale colors. Plus, currently in Basic Weather there is no way to control the 
visibility in the ground layer, so you're stuck with the default 16 km - I've 
asked a few times here if anyone is interested in modifying Basic Weather to 
support the details of the atmospheric scattering scheme, so far no response. 
Thus, currently only Advanced Weather makes full use of the features of the 
scheme.

> Limitations are there because some cool shaders like reflection,  
> lightmap (important for dusk/dawn/night flights) and transition aren't  
> working.
> If we want to show cool images or videos we have to decide if we want a  
> cool looking sky, or a cool looking everything else.
> Both together isn't possible as it was in 2.6.0.

That http://imgbin.org/index.php?page=image&id=6391 is how cool it looked in 
2.6 to have the skydome rendered in one scheme and the terrain in a different 
scheme when you don't cherry-pick conditions for the screenshot. That 
http://www.phy.duke.edu/~trenk/pics/skydome1.jpg is how beautiful the horizon 
looked when there was no mountain range to hide it. I rest my case. You never 
had a cool looking sky matching everything else. You carefully had to design 
the situations in which this wasn't problematic.

> The big problem again is, that there isn't a simple developement line  
> visible anymore.
>  It is difficult to explain users why some cool features advertised in  
> 2.6.0 will be "broken" again in the next stable release 2.8.0.
> Or with other words: things that wasn't finished yet, had been changed  
> with new things which aren't finished yet again, but behaves now  
> different again.

To say this again: The only way to 'finish' the skydome shader is to continue 
what I have started to do, i.e. to change fogging and lighting in every other 
shader. That is the development line that is happening. Rembrandt adds to the 
confusion, because making the skydome work with Rembrandt requires to change 
every shader again to support Rembrandt as well.

The manpower to do that is pretty limited - that's FredB, Emilian, Vivian and 
myself currently working on the shader zoo. So you don't get to combine all 
cool features at the same time yet, because there are like 30 shaders waiting 
in line to be processed. But then again, most people start complaining when all 
the cool features result in ~5 fps, so it's not enough to write the shaders, 
they also have to be optimized. I manage about one shader per month, which 
isn't so bad if you think about it.

> I'm not sure about. From users point of view we have a skydome shader,  
> and the lightfield shader which makes use of this skydome shader and  
> adds some further features but with the side effect of a lot non-working  
> shaders.

I give up. 

I wish the skydome shader in its inconsistent state had never been included. 
I've spent a year coding to finally get it consistent without creating 
artefacts all over the place to read this.  You're kidding yourself if you 
claim the skydome ever worked properly in 2.6. Just create some poor visibility 
and rain and show us just how nice the skydome shader alone looks with the rest 
of the scene... then you may appreciate just how big the issues to be fixed 
really are. 

I don't think it's a good idea to put cool features which work sometimes if you 
cherry-pick conditions into a release. Doing so in 2.6 with the skydome shader 
turned out to be a mistake. Going further into the direction of the mistake 
makes the problem worse, not better. 

Cheers,

* Thorsten
------------------------------------------------------------------------------
Live Security Virtual Conference
Exclusive live event will cover all the ways today's security and 
threat landscape has changed and how IT managers can respond. Discussions 
will include endpoint security, mobile security and the latest in malware 
threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/
_______________________________________________
Flightgear-devel mailing list
Flightgear-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/flightgear-devel

Reply via email to