Planetside Software Forums

General => Terragen Discussion => Topic started by: sjefen on August 27, 2007, 01:49:11 PM

Title: Feature Request for Clouds
Post by: sjefen on August 27, 2007, 01:49:11 PM
Maby I'm totally wrong here, but I have a little idea for the clouds.
It would be nice if they have a options like: Edge Darkness. I think that it would make it a lot easyer to make clouds like this:

http://xpda.com/junkmail/junk157/pict5107.jpg (http://xpda.com/junkmail/junk157/pict5107.jpg)

http://www.systemj25.net/SUNnFUNPICS2002/KSAT_KNEW/04_16_Big_Clouds.jpg (http://www.systemj25.net/SUNnFUNPICS2002/KSAT_KNEW/04_16_Big_Clouds.jpg)

I have been trying to make this for a long time, but with no success. When I studyed them i noticed that the edges was darker.
Title: Re: Feature Request for Clouds
Post by: Tangled-Universe on August 27, 2007, 02:18:43 PM
Well I think it's not the case that the edges are darker, they're just thinner/softer and against a relatively (relative to the clouds) dark background the edges seem too look darker.
And for that we already have the edge sharpness setting.
It's possible to almost get the exact same result with low edge sharpness + high density, or with high edge sharpness and low density.
Which approach is most suitable depends on things like the size of your clouds, their distance from the camera and probably a lot more.
It surely can be done, but it would take a lot of fiddling, patience and test-renders.

Check out these 2 from Christianfly:

http://www.renderosity.com/mod/gallery/index.php?image_id=1403189&member
http://www.renderosity.com/mod/gallery/index.php?image_id=1399545&member

Regards,
Martin
Title: Re: Feature Request for Clouds
Post by: sjefen on August 27, 2007, 03:26:16 PM
Those clouds from Christianfly are very nice and I don't wanna be really annoying her, but I still think there are a difference in those from Christianfly and those I have postet. At least in the first one from me.
Christianfly's clouds are brighter/white in the edges and if you look at mine you will se they are darker in the edges. They are brighter inside.
I may be wrong. But it sure looks like they are.
Title: Re: Feature Request for Clouds
Post by: old_blaggard on August 27, 2007, 03:49:23 PM
They aren't identical, but his clouds are the closest I've seen at this point.
Title: Re: Feature Request for Clouds
Post by: bigben on August 27, 2007, 07:13:38 PM
I agree with Sjefen...

I've sat and stared at clouds for a long time wondering how to get the same effect in TG2.  I don't think a simple edge darkness setting would help directly as it is a lighting effect that is dependent on the angle between the camera orientation and the sun (although I guess linking the effect to this angle might make a suitable hack?). The key difference with Christianfly's clouds is that the camera is pointing towards the sun, while this edge darkening happens more as you turn the camera away from the sun. It's not related to the background behind the cloud, rather the way light gets bounced around by the cloud particles. 

I haven't played around with cloud settings enough to know how well TG does this (or if it's possible), but the lighting in clouds is certainly very complex (GI is essential for very realistic results) and TG2 certainly does most of it extremely well.... but something like this either with existing settings or a later addition would really put the icing on the cake.

While we're asking for cloud stuff what about the circular rainbows you get when you look down on a cloud with the sun behind you? ;)
Title: Re: Feature Request for Clouds
Post by: Cyber-Angel on August 27, 2007, 11:11:48 PM
The apparent darkness at cloud edges as in the provided links is due to the internal scattering of light inside the cloud (Multiple forward and backward Rayleigh scattering) the darkness at the cloud edges is due to the proportional fall off of light photons from a single light ray inside the cloud. As you know a single light photon may only travel a certain distance before it runs out of energy and the fall off of the energy is directly proportional to the distance traveled, this effect can be seen on a ceiling in your home near a window you will observe that the part of the ceiling directly above the window is brightest and the further away you are the darker things become.

The same phenomena is observed in clouds.

Regards to you

Cyber-Angel       
Title: Re: Feature Request for Clouds
Post by: Volker Harun on August 28, 2007, 05:42:07 AM
Try the following: Set cloud colour to 0.1 - the scattering colour to 0.35.
Edge Sharpness should be around 0.001 to 0.005 while the density is about 3 to 12.
In the lighting tab, turn the fake internal scattering up to 0.5

Maybe this helps

Edit: And then increase the cloud fractal's contrast - maybe
Title: Re: Feature Request for Clouds
Post by: Volker Harun on August 28, 2007, 06:28:28 AM
@Sjefen: And least to mention, in the cloud's settings, look at the Tweak-Tab.
Title: Re: Feature Request for Clouds
Post by: sjefen on August 28, 2007, 07:21:14 AM
Thanks Volker. I'm gonna look in to it later.

ta ta....
Title: Re: Feature Request for Clouds
Post by: Matt on August 28, 2007, 02:17:04 PM
This particular visual aspect of clouds is something I want to put more effort into in future. To the best of my knowledge it is caused predominantly by multiple scattering rather than differences in particle size or any other change towards the edge of the cloud. Multiple scattering in clouds is simulated by GI in Terragen, although it is very difficult to compute GI to the level of detail needed to capture these very distinct edges.

Because GI cannot capture these details very well, TG's cloud has a parameter called "Fake internal scattering", which attempts to simulate this change in brightness away from the edges of clouds. Really it is brightening the interior rather than darkening the edges. To get the best out of Fake internal scattering you need very large, dense clouds, and the basic cloud colour needs to be darkened enough to give dark edges.

Careful use of Fake internal scattering in combination with GI is the best tactic I can suggest to achieve this effect.

Matt
Title: Re: Feature Request for Clouds
Post by: sjefen on August 28, 2007, 03:10:00 PM
Thank you very much Matt ;)
Title: Re: Feature Request for Clouds
Post by: bigben on August 28, 2007, 08:06:33 PM
Had a bit more of a play with the cloud in my tree animation test and noticed that the clouds did exhibit some of this lighting effect.  I ended up inthe same area as Matt suggested (and Volker). I think a detailed explanantion of what each of the cloud settings is actually intended for would help, but as has been suggested by others it's also a case of working out what you're actually trying to simulate. This is a lighting effect (so the lighting/tweaks sections will be a start) which is partly dependent on the cloud type (all of the other settings)

One of the main problems I find when experimenting is that when you make one adjustment to improve one thing, you end up making something else worse. To get around this, I usually try to look for relationships between two variables and use this as a base for making large tweaks. It's not an entirely scientific approach but it can provide some decent starting points.

In the case of clouds I'm currently making some progress with these two:
Tweaks: Fake Dark power = Lighting: Fake Internal Scattering
Lighting: Light Propagation = 2x Lighting: Light Propagation Mix

A lot, however depends on the settings Main Cloud settings, the cloud's Fractal Scale and Density settings.  That's 3 sets of 3 settings which can make the job relatively complex. One thing that has confused me regarding people reporting how many cloud samples they are using is that the number of samples for a given quality level will vary according to the values of these settings. 

This makes it very difficult to actually determine whether someone is using a very high quality setting or not and it would seem to me to make more sense to actually report the quality level.  If you're using a quality level of 1 and it's stating an equivalent number of samples over 120 then there's a good chance the image will be noisy and require substantial additional tweaks to the atmosphere and rendering setttings to fix.  Most of my clouds only have 60 - 90 samples for a quality of 1 and I never have much trouble with noise. 

I'm not saying that there is an optimum number of samples that we should aim for, as this will probably vary depending on the cloud type... e.g. requiring higher samples with thicker/denser cloud settings... but if you find that the number of samples is really high when you set the quality to 1 you may well find that you can make some small adjustments to the 3 x 3 settings mentioned above to lower the amount of noise in the image without affecting the cloud shape too much. 

Here's a rough test I did on the train this morning. GI was off for speed, but I'd recommend having it on some for some really cool lighting... 
Title: Re: Feature Request for Clouds
Post by: bigben on August 29, 2007, 01:03:32 AM
Not the greatest of samples but these renders use similar settings to the previous TGD I posted. The first includes GI, and while the lighting and clouds need a lot of tweaking you can see evidence of the lighting effect referred to above.  GI was set to 2,4 for this although I'd take out the colour next time and finish the render so I can get a decent HDR image to adjust...

The second has no GI (my fill light setup) and while the clouds aren't the billowing cumulus type the lighting of the clouds seems better to me.. the width of the darkened edges is a little wide for my liking so I'd need to tone it back a bit.. 90° fov, sun off to the left.

slooooowly getting there  ;)
Title: Re: Feature Request for Clouds
Post by: rcallicotte on August 30, 2007, 10:36:26 AM
Ben, I like the second one and this shows us, as with nature, there are lots of varied types of clouds with various anomalies even in what is considered standard. 
Title: Re: Feature Request for Clouds
Post by: Volker Harun on August 30, 2007, 11:13:02 AM
@Calico ... right! Everytime I think I managed to get decent clouds that match nature, I spot something completely different in the sky. Always!

@Ben ... I like the second one more. You could go for the following:
Decrease your clouds density by 50%, duplicate the layer. Plug a transform shader between clouds and fractal and move the noise some 10 to 20 meters up- or downwards.
This should break up those uniform grey areas.
It is just an idea ... maybe I upload an image today that shows such clouds. (I bet I will upload it, too much work for the bin ,-))
Title: Re: Feature Request for Clouds
Post by: Volker Harun on August 30, 2007, 05:30:08 PM
Here comes the image-link:

(http://forums.planetside.co.uk/index.php?action=dlattach;topic=1960.0;attach=5691;image) (http://forums.planetside.co.uk/index.php?topic=1960.msg21078#msg21078)

Volker
Title: Re: Feature Request for Clouds
Post by: bigben on August 30, 2007, 08:09:16 PM
I think the duplicate cloud layer has some merit to it.  I also experimented with a duplicate layer with an inverted profile but I didn't take it too far... 

I think the clouds in TG are a bit like nature sometimes too... occasionally I'll see a TG render where I think the clouds are unrealistic only to see a real cloud formation that looks very similar a few days later.

One of the biggest problems I have at the moment is that I do most of my TG work on a laptop and the sky preview is patchy (large patches of black)  My other systems that preview properly are either slower than the laptop or are at work...  I think I'll just have to set a few animated settings for a cloud and let it run while I'm away.

ahhh... nice image... your post arrived just as I hit reply...  the precipitation looks pretty good too.
Title: Re: Feature Request for Clouds
Post by: jo on August 31, 2007, 12:49:43 AM
Hi Ben,

Quote from: bigben on August 30, 2007, 08:09:16 PM
I think the clouds in TG are a bit like nature sometimes too... occasionally I'll see a TG render where I think the clouds are unrealistic only to see a real cloud formation that looks very similar a few days later.

I've found the same thing. I've given up on the idea of saying whether I think a rendered cloud looks realistic or not, because as if I say something isn't I end up seeing something similar shortly after, as you've found. There are certainly particular types of cloud I see in renders which I don't like from an aesthetic point of view however.

Quote
One of the biggest problems I have at the moment is that I do most of my TG work on a laptop and the sky preview is patchy (large patches of black)  My other systems that preview properly are either slower than the laptop or are at work...  I think I'll just have to set a few animated settings for a cloud and let it run while I'm away.

What sort of graphics card has your laptop got? I wonder if it's something to with depth buffer precision or something. Have you tried messing with the clip distance? You can use the "[" and "]" keys for this ( decrease and increase respectively ). I'm guessing you've already tried that though :-).

Regards,

Jo
Title: Re: Feature Request for Clouds
Post by: bigben on September 03, 2007, 04:50:33 PM
Quote from: jo on August 31, 2007, 12:49:43 AM
Hi Ben,

....
What sort of graphics card has your laptop got? I wonder if it's something to with depth buffer precision or something. Have you tried messing with the clip distance? You can use the "[" and "]" keys for this ( decrease and increase respectively ). I'm guessing you've already tried that though :-).

Regards,

Jo

ThinkPad T60 (Mobile Intel 945GM Express) Everything displays fine except for the sky. Hardware acceleration is turned off. Clip distance??
Title: Re: Feature Request for Clouds
Post by: NWsenior07 on September 03, 2007, 08:46:24 PM
I've had the same issue. I've got a ATI Raedon 1150 express in a dell 1501 laptop, but old, and comparatively slow home PC does the sky just fine.