This needs some critical help. Feel free to provide your assistance. ;D
I've never understood why TG2 does that weird oil on water effect on clouds. only suggestion I can make is to change the warp effect in pattern to none.
Cool one!I like it just the way it is.
Thanks jmeyer.
PG, do you mean those dark lines around the clouds? I see what you mean. I'm going to think about this one some more.
Is that not a result of the density, or maybe the fake internal scattering being too low (or too high, can't remember off the top of my head)?
Quote from: JimB on July 22, 2008, 02:37:00 AM
Is that not a result of the density, or maybe the fake internal scattering being too low (or too high, can't remember off the top of my head)?
The fake internal scattering can contribute to this, but the best thing to try is to increase the environmental light.
If you increase the fake internal scattering too much the definition at the edges of the clouds will be lost.
The EXR version (the above was the TIF version)...in all its grimy glory.
I'm working on new values...
Quote from: Tangled-Universe on July 22, 2008, 03:47:23 AM
Quote from: JimB on July 22, 2008, 02:37:00 AM
Is that not a result of the density, or maybe the fake internal scattering being too low (or too high, can't remember off the top of my head)?
The fake internal scattering can contribute to this, but the best thing to try is to increase the environmental light.
If you increase the fake internal scattering too much the definition at the edges of the clouds will be lost.
Maybe that's not a bad thing?
This could make a really hot looking lava flow if you could get it to flow along a channel, change the color, and add glow.
very nice :D:D if u look carefully u can see faces in the clouds lol
Quote from: JimB on July 22, 2008, 06:25:14 PM
Quote from: Tangled-Universe on July 22, 2008, 03:47:23 AM
Quote from: JimB on July 22, 2008, 02:37:00 AM
Is that not a result of the density, or maybe the fake internal scattering being too low (or too high, can't remember off the top of my head)?
The fake internal scattering can contribute to this, but the best thing to try is to increase the environmental light.
If you increase the fake internal scattering too much the definition at the edges of the clouds will be lost.
Maybe that's not a bad thing?
With loss of definition at the edges of the clouds I mean that the smaller whispy details get lost. Though the cloud already lack some of this.
Have you tried putting the exr through Photomatrix? Its a great HDR image manipulation program, better than Photoshop CS2 (dont know about CS3)
http://www.hdrsoft.com/ (http://www.hdrsoft.com/)
Richard
You can achieve better results with EXR indeed, however it's obvious that this image won't benefit that much from EXR because the essential lighting is just off. So you can trick and treat a lot with EXR but the result Calico is after, I think, can't be achieved by just running the EXR through photoshop or photomatix.
Yeah, I used PhotoMatix and went to extremes for fun.
Here is a sample of what I'm working on now. I'm messing with Quality on cloud settings and Environment Light as well as Fake Internal Scattering (in very small amounts). I also increased the buoyancy in the Density Fractal for the clouds closest to the ground. I made other changes I'll explain later. But, tell me what you think -
And I thought it would get bigger when I click on it ;D
Looks promising so far, I'm always keen on your clouds as you know ;)
It looks slightly better now, though the left part of the cloud still has strange light-distribution/shadows.
You might consider working on the light propagation settings.
This is an update, after a few tries and failures. I'd really like some critical opinions about this. I've been messing with the lighting mostly, but have done some small corrections on the geometry. Any ideas will be considered.
Quote from: calico on August 07, 2008, 10:26:07 AM
This is an update, after a few tries and failures. I'd really like some critical opinions about this. I've been messing with the lighting mostly, but have done some small corrections on the geometry. Any ideas will be considered.
I really like the foreground clouds and their flat bases. They'd look great coming over distant mountains.
Looks really massive :o
I would really love to have a look at the tgc as I'm working on a picture right now incorporating these very clouds ;)
I started over again. I'm not happy with that whipped-cream stiff cotton look.
Oh :o
I liked your last approach, it was an improvement at least. Right now in my image I have your last released tgc and would really like to try if it makes a difference to the overall composition :)
Until now your clouds are the best in the picture anyway ;D
Quote from: calico on August 08, 2008, 07:19:49 PM
I started over again. I'm not happy with that whipped-cream stiff cotton look.
I hope you kept those density settings though as they would be just right for an ash plume ejected from a volcano.
That's really good calico. The only thing I can see that would need changing is the odd "Guiness top" to the clouds in the background at the bottom right.
Hiho, I just wondered if you're still working on that clouds or if you've abandoned it ;)
I've been working for going on 6 weeks without satisfaction. I plan to have something useful and will post it back here.
That 2'nd one looks like an eruption.
Quote from: Mr_Lamppost on August 15, 2008, 09:09:49 PM
Quote from: calico on August 08, 2008, 07:19:49 PM
I started over again. I'm not happy with that whipped-cream stiff cotton look.
I hope you kept those density settings though as they would be just right for an ash plume ejected from a volcano.
agree looks like pyroclastic cloud!
Superb shapes. Keep working on it. ;D
This is my final attempt. It has come short of my vision, but I have ignorance abounding. Nevertheless, this is pretty close. The settings are all set pretty low to be able to render in under 4 hours.
Light levels adjusted very slightly in Paint.Net.
This is exactly the same as the immediate image above, but without distance shaders.
Believe it or not calico,
the reason those clouds look unrealistic is,
they are too detailed! Well not the mesh but the way the lighting strikes. You don't see so much bumpy details in real clouds, it might be there but you don't see it.
This pic is a good example: http://hoosierhighways.com/v-web/gallery/albums/Clay-County/thunderhead.jpg the lighting strikes more in wide locations and widely, rather than in small spots, the billowy details are simply overcast because the light penetrates the surface of the clouds. That is the problem why many peoples' clouds look so unrealistic.
Actually the next to the last image looks realistic. I've seen clouds like that. But, it isn't what I want.
I'm not going to continue to make clouds with such fervor in the future. There are too many variables in TG2 clouds that do not mesh with what I understand. It will take time for someone to hopefully document these things, when Planetside gets the necessary time. Or not. It might just be that making TG2 clouds entirely realistic has no particular future.
I understand your position Calico.
I'm very much getting to a 'fuck it' point with my deadlines.
While these certainly aren't your best clouds, the whole debate about 'realistic' or 'not realistic' looking clouds is flawed. In terms of cloud shapes I think we can all agree that nature can come up with far stranger things than we can. In the case of Calico's clouds the debate stems from the lighting and what you can or cannot see, but when you are comparing the render to reality, what sort of reality are you actually making the comparison to? If you are comparing it to a photograph, is the photograph exposed for detail in the clouds or detail on the ground or is it an HDR photograph. All three will give you vastly different lighting and detail representations.
If the argument is that you shouldn't be able to see the details in the clouds (even though they are there) and you want to simulate an overexposed photograph, then simply crank up the light exposure setting in the render camera node until you get to the point where you are getting blown out highlights to match whatever medium you are calling reality.
In regards to your latest clouds, perhaps the problems you are facing do not stem so much from the cloud settings, but from sun and atmosphere settings. Perhaps if you were to closely study a picture of the clouds you are trying to duplicate and make careful note of where exactly the sun is located spatially you would be able to more accurately make the clouds that you are trying to. At the very least it would take a couple more elements out of the equation.
Quote from: RArcher on September 10, 2008, 05:21:21 PM
In regards to your latest clouds, perhaps the problems you are facing do not stem so much from the cloud settings, but from sun and atmosphere settings.
i agree totally... all the clouds (almost all) are rendered with a "basic" atmosphere (so it seems) maybe by tweaking these settings the result will be far better (i am pretty sure of it) ;)
- try at least a 4/2 'haze and bluesky)
- and raise the height to 4000 and 10000 (for atmo and bluesky too)
- what is your contrast setting ?
well... maybe i'm wrong but that's where i would begin :D
tweaking the cloud is great but if the atmo doesn't fit your goal...
I also agree. Most clouds look much better with at least tweaked lighting and also atmosphere. I have basically given up at creating "realistic" clouds for the time being. Right now I am working of figuring out good lighting and atmosphere combinations that will enhance the clouds and the ground of my scenes.
Calico, I actually think your first recent post (not the 2nd without distance shader) was quite nice. It was only the darkness and gray color of the clouds that looked unrealistic to me. If you brightened them up I think they'd be some of the better I've seen posted here.
As for the difficulty of creating truly realistic clouds, this is not just a TG2 issue. There isn't a single application I know of that has successfully solved this problem in an easy to control and consistently reproducible way. The most realistic clouds in Hollywood come from high-end dedicated tools and have taken weeks or even months of tweaking to get right (see Stealth for example of Digital Domain's "Storm" volumetrics renderer in action). So I don't think your efforts are wasted, nor do I think you would have much more success in any other application. In fact from what I have seen TG2 gives you the best starting point right now.
Cloud and atmosphere realism is also an area that I am particularly interested in, and I think Matt shares that interest. Once the foundational work on TG2 is mostly out of the way there will be more time for tweaking of the shaders and controls for maximum realism and ease of use.
Without a doubt we do want to make TG2 easier and more fun to work with, but it's also very challenging to take these systems which are extremely complex and lay on top of them a simple and functional interface. What we have right now is essentially a first pass. Hopefully in the future there will be time for improvements in these areas. In the meantime I encourage you to keep experimenting as time and enthusiasm for your projects allows.
- Oshyan
I think ultimately this would be a worthy goal for cloud realism. Maybe planetside should talk to these university guys?
http://www-evasion.imag.fr/Publications/2008/BNMBC08/
Thanks everyone. I worked on this project for about 2 months. LOL So I'm giddy crazy and burnt toasty a little on the edges. I've watched storm clouds for weeks and blah, blah, blah.
To be clear, TG2 has the coolest clouds I've ever seen in any application ever. It's just I couldn't exactly reproduce something I saw in nature that was so cool. I came close, though. It was from a real cloud that I studied and studied and studied...from my porch. It was outside.
The next thing I plan to do with this is to get a decent render and then I might share the file, if there are interested experimenters. ;D
Calico,
I think its great that you've done all this research. It only brings you closer to your goal. And don't feel all alone. My spare time is usually spent researching and testing concepts like clouds, surfaces, lighting, etc. We wouldn't be where we are today without all that time spent.
It looks like you've made phenomenal progress with the cloud work you've been doing, and I'd love to see more.
Thanks,
Jeff
I have intentionally not done any postwork so I get the clearest criticisms. This is after tuning the lighting. I think something looks too unrealistic and the little blobs here and there look out of place. Any ideas would be nice.
Wow this is getting there though! Maybe increase cloud density just a bit more?
The stills in that research paper look very impressive, though the videos a bit less so. We're always interested in ways to improve the shading systems in TG2 so it's something to look into.
Calico, your latest cloud looks quite nice. I agree with Moodflow that density or edge sharpness increase may help. The edges somehow seem just a little too transparent. Otherwise the shape and shading are very nice. I would brighten the clouds a bit, either through cloud colors, or increasing exposure, and then of course increase samples. You ought to have a very good result on your hands at that point.
- Oshyan
Yes, please let Matt know about the link i posted. :)
You made massive changes in your cloud setup. It looks quite different now from what you originally posted way back in the past. And guess what? I DO use the first *.tgd you posted in the file sharing section in an image that is rendering right now as I write :P :P :P
As I am not that sophisticated with TG2 yet (I still didn't take the time to learn some basic stuff 8)) I just added some more nodes to yours et voilà, the clouds look considerably different now. What I don't know at the moment is if I just overwrote your clouds with another clip file that I used, I'm not quite sure how this node-thing works and how I have to attach them together.
But I wandered from the subject... coming back I have to say that I am very sympathetic towards your efforts! You still may be lacking the technical knowledge of every slider there is to fiddle with but don't underestimate your knowledge concerning this matter. I'll keep my fingers crossed for your work because it takes TG2 to a point where things are actually are beginning to look like they should ;D I'm just speaking of clouds here! Seems they're the most difficult thing to do, since most images are really brilliant but when it comes to clouds they often lack something... and so everybody that puts some heart in the developement of that area should be venerated :)
Where was I going with all this...? Umm, well, I hope you will continue your efforts and post them sometime in the future!!! :P
The thing is, Calico, that your render is excellent with shapes and all, but what you're actually looking for is the complete opposite of what TG2 is doing. Looking at your image, it occured to me that I'm looking at a negative of a real fly-by saturated cumulus. I fly fairly often on my travels, and of course that's a great opportunity to study clouds :-)
I took your render, and created a little mock-up to present my point (conceptually). If I cut out the cloud, invert the image and then up the exposure and gamma, paste it back into the original with the blue sky, I get the following:
[attachimg=#]
What we need TG2 to do is *darken* the cloud egdes, not to make them lighter than the rest.
Regards,
Frank
Frank! Thanks for the insight! You're right. That is what I've been trying to figure out and I'm not sure I knew it until now. Those edges define the "poof" in the cloud.
But, how?
That's "fairly" easy Calico:
Make the cloud color and scattering color very dark, around 0.1 - 0.25 for both. Then increase the fake internal scattering and you might also have to play again with light propagation and mix again. It's difficult to predict which fake internal scattering strength, but it is probably around 1.
Remember to have low amounts of glow for better depth-effects.
If you would like to send your tgd I could give it a go for you. It seems I will have some time for TG2 soon after a long time...
Martin
Thanks TU. This is a good idea. I'll try it, but the lighting is so much more teeter-tottery than just changing the Fake Internal scattering to make it work like this that I think I'll keep at it.
Seems selfish of me, you know, but I've been playing with the lighting for the last two weeks or so and have it about where it want it, except for this.
As soon as it's "road worthy", I'll post the image.
Quote from: calico on September 24, 2008, 03:35:19 PM
Thanks TU. This is a good idea. I'll try it, but the lighting is so much more teeter-tottery than just changing the Fake Internal scattering to make it work like this that I think I'll keep at it.
Seems selfish of me, you know, but I've been playing with the lighting for the last two weeks or so and have it about where it want it, except for this.
As soon as it's "road worthy", I'll post the image.
Well, you've seen my work on it, so I do know what's involved...
Basically, just doing the first steps I described inmediately gives the desired effects, yet not the result.
It's quite difficult to get the edges coloured just dark enough, but not too dark and the same for the lightness of the inner sides of the clouds.
Martin & Calico,
I'm fairly certain the cloud layer in TG2 cannot do that without severe (too many) compromises on other aspects of the cloud. I mean, I'm still open for sugestions, but I also have the impression I have already tried everything possible. I had a few semi-successes, but never to my satisfaction. I think (I hope) that Matt will find some time after the final to give the cloud layer another shot. On my wishlist besides a different (or at least additional) lighting and scattering strategy, would be a modified fractal flavour that gives an out-of-the-box cauliflower look, and another option, maybe an input, to influence the cloud profiles. I think it would also be necessary to take another shortcut on the way the atmosphere and the clouds are sampled, so that speed of rendering would increase.
But I'm drifting off topic. So as for the dark cloud edges, I don't think it can currently be done except with insane render detail settings - maybe.
Regards,
Frank
Thanks Frank. Whew. What a relief that it's not just me nor just my ineptness.
I will spend some more time on this today. And then I might just live with the results until later. Of course, I'm open to clear instructions, but the insane render detail settings is where I've spent my last three weeks.
Quote from: FrankB on September 24, 2008, 04:36:03 PM
Martin & Calico,
I'm fairly certain the cloud layer in TG2 cannot do that without severe (too many) compromises on other aspects of the cloud. I mean, I'm still open for sugestions, but I also have the impression I have already tried everything possible. I had a few semi-successes, but never to my satisfaction. I think (I hope) that Matt will find some time after the final to give the cloud layer another shot. On my wishlist besides a different (or at least additional) lighting and scattering strategy, would be a modified fractal flavour that gives an out-of-the-box cauliflower look, and another option, maybe an input, to influence the cloud profiles. I think it would also be necessary to take another shortcut on the way the atmosphere and the clouds are sampled, so that speed of rendering would increase.
But I'm drifting off topic. So as for the dark cloud edges, I don't think it can currently be done except with insane render detail settings - maybe.
Regards,
Frank