Outerra forum
Outerra Engine => Development screen shots and videos => Topic started by: cameni on October 07, 2010, 03:46:27 pm
-
A few preliminary screenshots for comparison of enhanced atmospheric model.
The new model is more precise than the old one, resulting in better colors and less artifacts. With the old one the sky is usually darker than it should be, and in some conditions it failed to capture the colors.
The new model is also slightly faster but more importantly it allows additional optimizations on other parts of the pipeline.
New one on the left, the old one on the right.
(http://www.outerra.com/shots/atmo/s1b.jpg) (http://www.outerra.com/shots/atmo/s1b.jpg)
(http://www.outerra.com/shots/atmo/s2b.jpg) (http://www.outerra.com/shots/atmo/s2b.jpg)
(http://www.outerra.com/shots/atmo/s3b.jpg) (http://www.outerra.com/shots/atmo/s3b.jpg)
(http://www.outerra.com/shots/atmo/s4b.jpg) (http://www.outerra.com/shots/atmo/s4b.jpg)
-
I do like how the effects are more consistent now. Well done. :)
-
Nicely done :) Like Raiko said, looks a lot more consistent and realistic.
-
Wow. I was so busy with the newly launced ISI forums I forgot to check here.
http://www.outerra.com/shots/atmo/s3b.jpg Literally just said.. "jelah--ah-ah.. Gel AH.. ahh.. ahh"
Probably a good side effect.
-
http://www.outerra.com/shots/atmo/s3b.jpg Literally just said.. "jelah--ah-ah.. Gel AH.. ahh.. ahh"
Yeah that was obligatory zeos-ish screenshot :)
-
Is there Groundfog integrated? That would be nice. ;)
Like this: (http://joergmartinek.de/Die%20Churfirsten%20im%20Hintergrund.jpg) (http://joergmartinek.de/Die%20Churfirsten%20im%20Hintergrund.jpg)
-
While this part of the engine wasn't specifically meant to handle it, it can be simulated to an extent.
Here, a simple hack that boosts the aerosol density below a given height:
(http://www.outerra.com/shots/s-fog.jpg) (http://www.outerra.com/shots/s-fog.jpg)
There are some disturbing effects on the boundary during the movement though, and it would probably have to be implemented in a different way because fog particles have slightly different properties than normal non-condensing aerosol particles in the air.
-
The fog looks amazing!
-
Yes, the fog does look very nice. But as Brano said, more work will have to be done. And welcome to the Outerra community yoonsikp. :)
-
The fog does look great- No flightsim i've played has ever got it 'right' (i.e. wispy, variable density etc).....but that looks fantastic!
-
Wow, very nice. :)
-
For a quick hack that fog looks incredible.
-
For a quick hack that fog looks incredible.
I cannot help but agree.
-
Well done ! :D
-
Simply brilliant, if this is the result of a "quick hack" then I cant wait to see how the atmospheric generator part of it will look. Creating realistic atmospherics has never been done by any current flight sim, they all have settled for static balls of cotton wool and fixed backdrop imagery.
Given that weather computation can only be performed on "super computers" I would be most interested to know what approach you will take to generate a spectacular cloudscape, the dynamics of the atmosphere and "reading clouds" is fudamental to a great many aviation activities, weather effects and so on all play a part.
-
Well I have looked at these comparrison images carefully over a couple of days of re-visiting them, it seems to me that with the possible exception of the planetry horizon scaled shots (where the after is much better) there is not so much, an improvement as a change.... both 'before' and 'after' images represent possible realities because the atmosphere is so dynamic in reality. The main difference between before and after is the degree of haze (visible range) and once again this is a normal variable for a plenetry atmosphere that has thermo and flud dynamics in motion all of the time... Weather. I note that the contrast in terrain detail is lost or significantly reduced in the 'after' images which in my opinion is less attractive.
BOTH SETS have great reality impact but if in acheving the 'afters' you must lose the possibility of the 'before' then this might be a backward step where the main benefit is in the accessability mentioned in your remarks. The final judgement will most likely require compromise.
All in all this graphic still represents substancial improvements over any other simulator I have ever looked at.
-
Yeah the comparison isn't saying enough about the change. The new atmospheric model mainly solved some corner cases that weren't rendered correctly (the night side, twilight transition) and also the perception of depth in some cases. It's also a bit faster and allowed for further optimizations as well as for a better further tuning of atmospheric parameters.
Additionally, the parameters weren't the same in both cases - their meaning changed and it was harder to match to the old ones. In the end there will be a UI window with sliders for adjusting the various atmospheric parameters.
-
Maybe,with Direct X 11 one can do this even better ,so you better incorporate it now ,if possible.I came to think about it ,can Vue 7 program maybe help in making a better reprensentation of season and maybe add to realism by adding parameters to make it simulate different road conditions-add a certain colour and it adds in a slippery road or something.
Better find out what it can simulate without dragging down fps now,before we go too complex on weather model and everything else we can think of.
Make sure of that it can manage the new whitepaper on the new MP Dis and HLA future systems,so that we can integrate the weather model into it.Hopefully we canuse Vue 7 system ,where one can `paint`their own clouds and overlays into the weather model,without the need of scripting or C++ programing .
That will make it a whole lit easy for a lot of people. :D
-
Hi Johan. Outerra isn't rendered with the DirectX libraries. It uses OpenGL which is why it looks as good as it does and also why there isn't a demo out.
Glass is half full.. of poop.
-
Hi Johan. Outerra isn't rendered with the DirectX libraries. It uses OpenGL which is why it looks as good as it does and also why there isn't a demo out.
Glass is half full.. of poop.
OpenGL is not the reason it looks that good. One can do the same with DX. It are just wrappers around the hardware...
-
OpenGL could do DX11 when version 2 was out.
-
OpenGL could do DX11 when version 2 was out.
?
I think you do not understand what OpenGL and DirectX are.
They are API's for rendering.
You can do the same with both api´s. But DX will be faster in one and OpenGL in the other.
There is no such thing as OpenGL can do DX11...
-
I think he wanted to say that OpenGL can do what DX11 can since version 2.
But it's not true regardless.
OpenGL 4 + OpenCL can probably do all the things DX11 can.
-
You would not post your shaders here, would you? :D
But for all of us, trying to get something similar working in our simulation software, a small technical blog post would be very appreciated! :cool:
Because it looks just gorgeous...
-
I actually wanted to write a post about it, for the same reason I've been blogging about other technical things - to save myself and others the trouble of rediscovering it all again. But I missed that period when it has been fresh in my mind, so I have to go through it .. again :(
But I've been doing some fixes to the atmospheric code when toying with the water - I've encoded atmospheric parameters of reflected ocean ray into unused portion of precomputed textures. I've fixed some bugs during that encounter, but it's not yet all finished.
Btw the current incarnation of my atmospheric code was inspired by Eric Bruneton's (http://evasion.inrialpes.fr/~Eric.Bruneton/) atmospheric scattering algos, did you look at that? I've just simplified it a bit, reduced the inscatter table to be smaller and 3D instead of 4D, at the cost of having to recompute it from time to time. Basically doing some tradeoffs here and there.
-
Yepp, the work of Eric Bruneton is also very impressive. I read the papers but annoyingly the work on my visual software is only a side project at my office so I have to focus on getting everything in there but not in full detail :(
Last thing was volumetric clouds. But combining them with a complex atmospheric shader is some tough work. (At least for me... :/ )
So I am still looking (and hoping) for some ideas for a less complex atmospheric code...
-
Yes, the paper is not easy to understand without having a special training in the way these people write them :)
I hope I'll be able to write it in a more intelligible manner, but that will require even bigger effort than just dumping my notes and equations. Atmospheric code was one of the more annoying parts because of the amount of issues and tweaks required to get it right at all scales, and I still don't have it bug free.
-
Hell ya, thats right. I had this problem also. If the radians arent correct the sunset was too early or too late and it was looking unnatural. But i took nvidias gpu gem model. Thats physically "correct" but slower. So i'm also curious about your notes :P