Hey everybody! It's been a long time since I've posted an image. This one has been in the making for quite a while now, and it's finally to a point where I'm ready to open it up for critiques :). This image took so long because some of the techniques here are new (at least to me). The dust storm was definitely a challenge to get looking proper (and it's still not *quite* there yet, in my opinion). The arch is all procedural displacement, which allows me to put some nice textures and bumps on it, and will allow me to get the sand and rocks to lead up to its bases when I get around to working on that bit. I also want to share an example of a technique that I will be using to get the sand to push over the rocks to simulate wind. I've attached a test render of the technique but due to reasons explained in this (http://forums.planetside.co.uk/index.php?topic=7477.0) topic I haven't implemented it into the main scene yet.
Anyways, all kinds of comments and critiques are appreciated, as I think that this image can still go quite a ways and become quite good :).
The sandstorm looks mean and fairly good to be honest. Can you make the arch more prominent? For example make it more a hero object. I'd also make some parts of the basin slightly elevated just to give a feel of more loose parts of dirt. Maybe clamping certain areas where hard meets soft ground?
OB, I wish you could be here in the Phoenix area when we have our sandstorms during July through August. Google Sandstorms/Phoenix and you'll see what I mean. You've done excellent work here to simulate the wall of dust ...it really looks like this (and, later a real mess for us to clean up). I agree with CCC regarding the main arch. Making it larger and moving it a bit to the left so that it is about 1/3 to the left from the right edge of the scene. This should balance the scene a little better. Love the colors here too. Looking forward to seeing the final.
Bob
looking really nice OB....cant wait to see this finished!
The sandstorm is very impressive. Something is bothering me about the lighting, but I'm not sure whether it's the angle or the strength (most likely a combination). Great scene, and I look forward to further developments.
The sand 'cloud' is too dense. I would put the sun so it shines through the dust. It could make a really nice light effect.
The arch is also interesting. If it were larger, the dust could seem to blow through it.
I agree with domdib about the lighting, it looks like two stitched renders ... quite flat. Only in the middle area left from the arch (which is wonderful) is an area where I can feel the depth of the storm.
The sandstrom itself is very good ... I would go for more grain as a next step (do not ask me whether cloud or atmosphere samples).
Keep on going!!!!!
Thanks for the input, everybody. I absolutely agree that the lighting isn't where it needs to be yet, and I really appreciate all of the rest of the advice. Luckily, this doesn't take too long to render on my machine (about 2 hours for this), but the trick will be finding time to sit down and work on it a little, especially considering that school starts for me next week.
very good sandstorm :)
i agree with Volker though, you will need more grainy look.
Quote from: old_blaggard on September 09, 2009, 01:12:49 AM
I've attached a test render of the technique but due to reasons explained in this (http://forums.planetside.co.uk/index.php?topic=7477.0) topic I haven't implemented it into the main scene yet.
would this technique get you started? http://en.tgblog.de/?p=40#more-40
Regards;
Frank
After a very, *very* long pause, I've revisited this scene. I'm much happier with it now, but there are still a couple of things that don't quite work for me:
- The image maps are being stretched oddly on the underside of the arch
- Some of the fake stones (bottom left for example) have weird shadows on them
- The plant at the base of the arch is just a placeholder; does anyone know of good small cacti (or other more desert-like plants) that aren't too saturated?
Anyways, any ideas about these issues or other comments about the image are much appreciated.
Superb work! I find the hardlower edge of the arch a bit off putting The texture streaking under the arch is as I'm sure you know a mapping problem. If the archis procedural you are going to have to find a way of wrapping the texture.
Mick
Thanks. Yeah, the image mapping is a little weird. Because the arch is, indeed, procedural, TG2 doesn't seem to like the way that the coordinates are stretched in some places. I've tried through-camera projection, but that just looks plain bad (although it does help to soften that edge you mentioned).
If your arch is an object, give each face direction a different material. You can then plug the various procedural node structures into each material. This should make the mapping of the various areas look right.
Awesome work! Would like to see the arch much larger. Linda
If you are going to try to image map it via "through camera" projection, I've found that it works best to create three or more different camera locations at angles to the main render camera and project your image (or slightly varying images) through these cameras. This way you can set the break-up of each individual layer which really helps avoid the flat plane look that happens if you project the image directly through the main render camera.
Thanks for the tips.
njeneb - Good ideas, but the arch is a set of shaders built within TG2, so I can't apply that directly
Ryan - I started thinking along similar lines after I made that post. The problem is that I don't like the texture of the through camera projection on most of the rock, so I'm currently using a mask to get my Object UV projection on the entire arch except for the underside, where it switches to through camera. I've attached a crop render of what that looks like.
Linda - Do you mean that you would like the camera to get closer, or simply a higher-resolution render? I'll certainly be rendering the final at high resolution, but I'm not sure if I want to bring the camera a whole lot closer.
Any ideas for the other issues?
When setting up for your final render, I'd like to see you use a vertical format; like 960 x 1280 ...that would emphasize the the rock structure.
Good idea. I usually default to landscape formats because they're more traditional, but something more vertical could be interesting. I'll look into it.
To be perfectly honest, I hadn't seen the larger arch when I commented. I really like the size of the arch in your most recent render- no need to make it larger. Very cool work. :)
Quote from: old_blaggard on December 24, 2009, 12:59:55 PM
Linda - Do you mean that you would like the camera to get closer, or simply a higher-resolution render? I'll certainly be rendering the final at high resolution, but I'm not sure if I want to bring the camera a whole lot closer.
Well what could I say? I up to your level of skill is to grow and grow!
This is a very interesting technique, old_blaggard. It makes my brain spin again in trying to figure it out. I would be interested how you managed this procedurally. As for the image itself, in the first there is a slight haze near the horizon, which I would get rid of. I would also bump the plain a bit. The sandstorm is great and overpowering, in that overpowering sense it would indeed be good to make this a vertical. Horizontal images tend to soothe, vertical to get you on your toes.
I have made a tall cactus, if you're interested (even a petrified cactus).
---Dune
Gets better all the time - very impressive work.
Mick
Quote from: RArcher on December 24, 2009, 12:54:00 PM
If you are going to try to image map it via "through camera" projection, I've found that it works best to create three or more different camera locations at angles to the main render camera and project your image (or slightly varying images) through these cameras. This way you can set the break-up of each individual layer which really helps avoid the flat plane look that happens if you project the image directly through the main render camera.
Now that's very tricky of you Ryan!! That opens a whole new set of possibilities for texturing, does it not?
Sorry for the brief delay. There were a couple of extra issues that I hadn't noticed in the first go-round that I had to work out.
I think the vertical composition came out very nicely. I'm pretty satisfied with it now, but I'm still open to any and all suggestions and criticisms. Thanks for all of the input so far!
That's such a cool pic - great use of atmo and clouds - lovely lighting and textures too :)
Excellent. The texuring is perfect although I still don't like the hard edge!
This has progressed into an excellent quality render, perhaps the sand is a touch too grainy (no pleasing some people). What would complete this thread is a clip file of the basic arch construction before texturing (that would be very instructive).
John
Cool picture Paul :D
I think this merely a showcase rather than something where you tried to perfect all the elements, so I won't deal with the rest.
I finally figured out how this Arch thing you created works and I'm making a version of my own now, quite differently :)
Two questions: if you scale the values for the arch size etc. then you loose the thickness of the arch easily, resulting in a thin piece of folded paper if you know what I mean.
Is this because the displacement's origin is also from a small surface area? Could you give me directions of how to do this? I now have a nice arch, but I have to trick and cheat quite a lot to make it look a hundred of metres while it is a dozen of metres in TG2.
Second, do you plan to share this technique over here? I might help you out with a cleaned up and organized version with renamed nodes and such. Just let me know :)
Cheers,
Martin
Very nice render, Paul! The arch is great and I especially like the texturing!
Cheers,
Frank
The only thing I find is the hard arch edge. Aside from that, awesome pic!
That is looking real fine Paul. I would like to suggest that youkeep the camera pointed where it is at; but, zoom backwards until the entire structure is visible. Also, you might consider changing the color of the dust to a gray (with just a tint of redness) - this way, there would be a better visual emphasis on the rock structure. Damn; I know, you just can't please everybody!
Thanks for the input and comments, everyone. I, too, am still not entirely satisfied with the hard edge, but I think it looks plausible now instead of just wrong ;). I also understand how the stones might be too sharp for some people, but I think I'm going to leave them the way they are. As you've all said - can't make it perfect for everybody :P.
It's true, Martin, that the big technical focuses here were the arch and the dust storm, but I also had to fight a bit with the fake stones to get something that was at least realistic enough to blend with the rest of the scene. Relatedly, perhaps someone can chime in with a word of advice on how to fix my problem: at one point I decided to use some intersect underlying between the larger fake stones and the terrain. However, even after I did that the shape of the smaller fake stones that I added later were always somewhat distorted and didn't have the same shape as the other ones (I disabled smoothing effect in the surface layer, which helped, but didn't completely fix it). Does anybody have ideas about how to get intersect underlying to have minimal impact after the surface layer where it's performed?
Also, I haven't tried scaling my arches up to that level yet, although it certainly should be possible. A couple of things that come to mind:
- The displacement along x in the redirect shader is not applied evenly across the arch: there is a mask that in fact varies it over the course of the base, otherwise it would always become the little paper thing you mention. Try experimenting with slightly different values of Constant scalar 01 and Constant scalar 01_1.
- Increasing the base surface area also helps, so you can certainly expand that.
- If you're getting that folding effect, you can always get rid of it by displacing everything outwards by a few meters.
Finally, if you want to clean up and rearrange nodes to post it for people, that would be great. I was going to just post my project file, but it would be good to have someone with experience in this kind of thing go through and make it more legible.
Quote from: old_blaggard on December 27, 2009, 12:27:55 PM
Thanks for the input and comments, everyone. I, too, am still not entirely satisfied with the hard edge, but I think it looks plausible now instead of just wrong ;). I also understand how the stones might be too sharp for some people, but I think I'm going to leave them the way they are. As you've all said - can't make it perfect for everybody :P.
It's true, Martin, that the big technical focuses here were the arch and the dust storm, but I also had to fight a bit with the fake stones to get something that was at least realistic enough to blend with the rest of the scene. Relatedly, perhaps someone can chime in with a word of advice on how to fix my problem: at one point I decided to use some intersect underlying between the larger fake stones and the terrain. However, even after I did that the shape of the smaller fake stones that I added later were always somewhat distorted and didn't have the same shape as the other ones (I disabled smoothing effect in the surface layer, which helped, but didn't completely fix it). Does anybody have ideas about how to get intersect underlying to have minimal impact after the surface layer where it's performed?
Also, I haven't tried scaling my arches up to that level yet, although it certainly should be possible. A couple of things that come to mind:
- The displacement along x in the redirect shader is not applied evenly across the arch: there is a mask that in fact varies it over the course of the base, otherwise it would always become the little paper thing you mention. Try experimenting with slightly different values of Constant scalar 01 and Constant scalar 01_1.
- Increasing the base surface area also helps, so you can certainly expand that.
- If you're getting that folding effect, you can always get rid of it by displacing everything outwards by a few meters.
Finally, if you want to clean up and rearrange nodes to post it for people, that would be great. I was going to just post my project file, but it would be good to have someone with experience in this kind of thing go through and make it more legible.
Thanks for explaining Paul :)
As I said I've "organized" your network and have renamed the displacement-shaders to name which I "think" they do like controlling width, thickness, height and length.
Further, there are a couple of root-scalars and the like which if you disable sometimes gently influence the shape of the arch, but I have NO idea what they actually do.
Honestly, I have barely any idea how you pulled it off ;D
I just reversed engineered how it worked by changing lots of numbers and things, that's it. So the renaming might not make sense :)
Anyhow, I now can manipulate it to 90% of my likings, besides I still don't get how to avoid the thin paper look.
I will try your suggestion to change those 2 scalars, thanks :)
However, I don't know how the expand the base surface area, which part of the network defines that?
I attached a doodle I made yesterday which also has snow and fake stones. Is this kind of what you're looking for?
Cheers,
Martin
Old_blaggard this is a nice and interesting technique :)
Kadri.
Just an image of 'Delicate Arch' to view for comparison; I've been here several times over the years. The beautiful rock structure is seen at the 'Arches National Park' in east-central Utah. The hike to the place is uphill and is worth the effort. This place reminds me of the Roadrunner and Coyote cartoons where you must walk along ledges of red sandstone. Worth the visit which I wish all of you could make.
When i see such photos and 3d images side by side i see how far we are gone in software and artistic ability :)
Kadri
Thanks, Kadri :).
Bob - That was actually one of my original reference pictures (although I decided on a different milieu and rock style in the end ;P)
Martin - first of all, good-looking test! Increasing the base area of the arch couldn't be simpler. Right underneath the Subtract scalar 01 and Subtract scalar 01_1 nodes, simply create two divide shaders. Toss a couple of constant scalars into input 2, and then use larger numbers to get a larger surface area. If you look closely, the first few nodes in the arch are actually the formula for defining the unit circle, and adding the divide scalars turns this into the formula for creating an arbitrary ellipse, so if you want, you can create a more stretched-out arch. The root and square scalars that you're talking about are simply there to provide different interpolations of getting from zero to one. As you said, they influence the shape gently, but it's easier to just experiment with them than to trace out their logical effects (trust me :P). The constants I mentioned before should have slightly more dramatic impact on the scene, especially if you pump in extreme values, but I think that the best results are to be had within the range from zero to two.
Ok, I'm on it! Thanks!
I really had a laugh about this
QuoteIncreasing the base area of the arch couldn't be simpler.
;D lol
btw. I'm really having problems populating this, do you have any ideas?
Quote from: old_blaggard on December 27, 2009, 01:09:32 PM
...
Martin - first of all, good-looking test! Increasing the base area of the arch couldn't be simpler. Right underneath the Subtract scalar 01 and Subtract scalar 01_1 nodes, simply create two divide shaders. Toss a couple of constant scalars into input 2, and then use larger numbers to get a larger surface area. If you look closely, the first few nodes in the arch are actually the formula for defining the unit circle, and adding the divide scalars turns this into the formula for creating an arbitrary ellipse, so if you want, you can create a more stretched-out arch. The root and square scalars that you're talking about are simply there to provide different interpolations of getting from zero to one. As you said, they influence the shape gently, but it's easier to just experiment with them than to trace out their logical effects (trust me :P). The constants I mentioned before should have slightly more dramatic impact on the scene, especially if you pump in extreme values, but I think that the best results are to be had within the range from zero to two.
When i read such instructions i say to me that English is not my mother language . In this way it feels less depressing to me ;D
Cheers.
Kadri.
On the Renderosity site, there are two files available call 'Unique Rocks' (do a search) - I ordered one of them a few years ago. There are about 12 different shapes in the file about which there are a number of arches. They can be opened up in TG2. Worth having a look.
Quote from: Tangled-Universe on December 27, 2009, 01:13:24 PM
I really had a laugh about this QuoteIncreasing the base area of the arch couldn't be simpler.
;D lol
btw. I'm really having problems populating this, do you have any ideas?
Heheh, it looks like I've fallen into using one of my professor's favorite sentences :P.
To populate, you can use a bunch of nodes to distort coordinate space on the function level. Or you can use the "Shader array" node that was released twenty-four hours after I'd figured that first one out :P. However, all of the arches will be going in the same direction. I haven't tried this yet, but you should be able to use carefully coordinated perlin noise functions as multipliers of the displacement shader inputs.
Bob - Those look like good packs. I knew from the outset that I could probably find a pretty good arch model out there, but part of the fun of this project was learning how to grow it procedurally using shaders.
I've decided to hunt for a slightly wider composition, (so the word "Finished" in the title is now a lie :P), but I think I am done tweaking shaders.
If you're curious, here is an early test of populating arches:
Quote from: old_blaggard on December 27, 2009, 01:51:40 PM
If you're curious, here is an early test of populating arches:
Oops, sorry...by populating I meant putting populations on them, like shrubs etc.
My bad (english)!
Cool concept nevertheless!
Very interesting; thanks Paul.
Ah, I see ;). I never really played with that. I did a quick test with some minuscule birch trees, and I didn't have huge troubles with it - I'm treating the arch just like a normal bit of terrain and things are populating as expected. What kind of issues are you running into, specifically?
Well I treated it exactly the same way as you, but if I want to populate the arch I have to set the density very very low to get a couple to appear on the arch while tons are on the flat surface. Somehow TG doesn't recognise the surface correctly it seems?
Ah, yes, that happened to me, too. I just used a distribution shader to restrict it based on altitude. Here's what I think is happening: when it's calculating the distribution of the plants, the populator is looking at the original coordinates, which means that the arch is just a unit circle (in the distributor's eyes). The populator matches the objects to the terrain, though, after computing the arch's displacement. This means that the arch, whose surface area has grown several times over, only has one unit circle's worth of plants on it. The solution would be to force the distribution function of the populator to look at the final terrain output for its coordinates, but there's no way to do that right now.
Hi,
nice arch!
You said earlier:"I was going to just post my project file, ..." and i wonder
if you would still do so,because i for one would like to see the unaltered
file rather than the cleaned up version Martin does.(to me it's less confusing)
Regards,J.
I have made a small (but long!) render showing the arch in a slightly different way, will post the link within an hour...
Quote from: j meyer on December 28, 2009, 10:20:34 AM
Hi,
nice arch!
You said earlier:"I was going to just post my project file, ..." and i wonder
if you would still do so,because i for one would like to see the unaltered
file rather than the cleaned up version Martin does.(to me it's less confusing)
Regards,J.
I can assure you, it will be more clear ;)
Nodes which haven't been used or lead to a dead unused end are removed and I have retopologized things from bottom to top, because some connections were going all over the place. But if you insist ;)
I do ;) ;D,but of course it's up to ob if and what he likes to share.
Here's the plan - I'll probably post the basic .tgd tonight or tomorrow morning, and then I'm planning on releasing the cleaned-up version and an in-depth explanation of how to use it and the logic behind the nodes through NWDA.
Thanks again for all of your comments, folks!