This is something I've been working on for a while now. It's the result of what was a much larger project originally where I built my Terragen world completely out of scale and had to then rework the whole thing because there's no easy way to just scale down the entire scene. I thought of scaling up the planet but decided it would be better to just bite the bullet and start over with the correct scale settings.
I'm posting this as a WIP because I'm struggling with the foreground terrain, and after seeing Fleetwood's Fake Stones images he just posted I thought maybe I might be able to glean some insights. I do plan on adding a bunch of Joshua Trees to the scene that I've exported from Speedtree. Initial renders I did with them were ok but not great as I'm having trouble getting the UVs of the main trunk of the tree to render without all the apparent tiling. I tired bringing the model into Maya and adjusting the UVs but when I brought the obj back into Terragen I found I could no longer attach any of the shaders. It seems the mtl somehow got lost in the translation between Terragen and Maya I'm guessing?
The plants that are in here now are called Snakeweed. Living more north and east of Mojave, It's not really indigenous to the Joshua Tree area, but it looks a lot like the kinds of shrub one sees in the area. I plan on adding more variants of them as well as older dead ones from past seasons. I'm also planning on adding various grasses, but so far haven't had much luck finding what I'm looking for. If anyone has a link to where I might find savanna type grasses I'd appreciate it.
This a crop of a larger image, the part that I'm most happy with up to this point. I mostly only get to work on this late at night, but with a huge (and much needed) storm about to hit the Socal area through the weekend I hope to be able to spend more time on this.
Man, it sucks that you had to start over. But this is looking like a great 2nd try. Do you have a reference photo for the foreground area your trying to achieve?
I think that the lighting and the dramatic color palette are spectacular. The composition lends itself to space and distance, but certainly the foreground is going to ultimately determine the image. I'm looking forward to your Speedtree objects ( :)). Nature always surprises me with it's diverse complexity of the little things in the foreground; rocks, twigs, dried twisted branches. This image already rivets my attention, looking forward to the next installment.
Looks already very nice!Especially as you said the upper part.
@TheBadger,
I was looking online for some inspiration and found this image. I think maybe something like this for the extreme foreground. I've dabbled in this type of displacements before so I'll give it a try tonight as I'm sipping scotch and listening to the rain fall. Should make for a relaxing evening. Sun angle works too.
Quote from: gregtee on February 26, 2014, 05:07:13 PM
Initial renders I did with them were ok but not great as I'm having trouble getting the UVs of the main trunk of the tree to render without all the apparent tiling. I tired bringing the model into Maya and adjusting the UVs but when I brought the obj back into Terragen I found I could no longer attach any of the shaders. It seems the mtl somehow got lost in the translation between Terragen and Maya I'm guessing?
obj's & UV's coming out of Maya do get funky w/ TG, I can only Speak on Bobby Starr's behalf and recommend running it through Poseray ;D Then check your maps and re-save it. Terragen will take that in nicely
https://sites.google.com/site/poseray/ (https://sites.google.com/site/poseray/)
Good work...........I'll be watching this. I'm afraid you can keep your scotch............I'll take a good aged bourbon. ;)
Oh I'll take that too! I think I've got some Eagle Rare hiding somewhere under the bar.
Sounds good.................. ;D
On the sand in the reference image. There was a file share here with some of the best powdery looking sand I ever saw in TG.
You would probably have to mix it in with more rocky-ish sand. But the effect should be really close to what you are after.
Ill see if I can find the post.
Thanks Badger! I'd love to see it.
found it.
You will probably need something slightly different. But as a starting point, I really don't think I have seen a more powdery like sand shared before. Most everything else I have seen has more grain/rock particles. But this should help with the more 'fine dust' aspect of a desert scene:
http://www.planetside.co.uk/forums/index.php/topic,16464.0.html
If you have not done sand ripples before, or even if you have, this thread is filled with in-depth discussion:
http://forums.planetside.co.uk/index.php?topic=843.0
I've had a few tiling issues with Speedtree using the sample trees with the provided textures and UV settings. With my own assets not so much, so without really knowing your maps and settings it's hard to guess what's up. My own experience is that ST brings all in pretty well without need of pov-ray as an intermediary (or thru Maya). I tend to create pretty large textures for the trunk surfaces, which is not a 'thrifty' way to spend resources, but TG eats up big poly and large map objects with great ease. The lack of fbx or alembic suport for the imported meshes is a limit for sure, but the obj format works real well straight from ST to TG. The serpentine shaped sand drifts are cool, but from my viewpoint such formations need a more 'open' foreground composition to look appropriate, think of the open wind paths necessary to make such things. I really like the image so far, it's quite a vista!
Thanks for the scene Badger, I'll take a look at it when I get home. That link you posted does have some nice powdery looking sand but what was lacking was the ripples themselves. I've tried to do this kind of stuff before with limited success. I guess I'll start with that file and go from there.
Zaxxon, I'll try putting higher res images on the trunks of the trees and see what happens. I'm far more used to doing asset development in programs like Maya and Vray so getting the hang of how Terragen handles these types of object imports will just require some time. Not being able to see and adjust the UV sets is a little frustrating too but I know that's not what Terragen was made for. I do wish Terragen had normal map support since Speedtree will shoot them out as part of the export to obj process, and having access to that channel information would speed things up a lot when setting up initial shaders.
Thanks for the info though, I'll hit it later tonight and see what I can get accomplished.
-greg
dandelo? made a sand pack which would be perfect for your needs but I'm not sure where you'd find it!
Here's v2 of this project. I was trying to reproduce the sand ripples for the foreground like the earlier pic I posted and somehow wound up here, which is typical Terragen for me which is I often am trying one thing and by happenstance find myself somewhere else. In this case I'm not sure if it's an improvement or a distraction. I like to cragginess of the foreground but it seems to get all jumbled together. It does on the other hand make it look like a real place, but does it make it look like a good composition? Just because something looks interesting doesn't always mean it looks good. I feel I need to edit it a bit more.
As I write this it's raining pitchforks here in Los Angeles. It's actually coming down sideways. God knows we need it.
-Greg
Greg nothing much to add.You know what you do.
I think a scene that is kinda in the middle with the first and this version(a little more to the first even maybe) might be nice.
That lighting is very believable and the cuts in the rocks are interesting as well. Overall, this is a very convincing varied landscape. I do not know what other input to add other then that.
The image is proceeding quite nicely in my estimation. The greater detail in the foreground helps set the panoramic scale to my eye. As to the 'jumbled'; that's fine, but a little more bounced light to fill the detail in the shadows might reveal some cool stuff. It's a pleasure to watch a Pro 'take' Terragen on, and I'm looking forward to the next step, thanks for sharing the WIP.
I like it, particularly the lighting, but at the same time the foreground is not all that interesting. There is a large area that is in the shadow and the rocky ground is, well, just not that striking. How about a nice tree that is in the sun perhaps?
That is one NP that I would definitely like to visit. Must be a cool place.
Thanks for the inputs. I agree with all the comments. It's a lot different doing the work vrs critiquing the work; it's too easy to lose one's objectivity, especially after a lot of this was done late at night after several pourings of burbon. (Thanks for the suggestion yossam) 8)
I think I'm going to try for the sand ripples again. With regards to the shadow areas being too dark, it seems that perhaps my monitor at home is likely running a bit too bright. I noticed here at work that what I posted seemed a little blocked up and dim as well. Guess I'll need to recalibrate my stuff at home.
thanks again
-greg
QuoteI often am trying one thing and by happenstance find myself somewhere else.
Yeah, me too. I think it is because Terragen is so open there is no way to know how exactly you can get from point A to point B in any given situation. It can feel like nothing is constant and there are no rules to guide us. After you spend a long time trying to make one specific thing, you easily loose sight of what you were after in the first place. This is why having a large and well maintained collection of clip files is so important I think.
All 3D work is so repetitive that doing things twice can become a torment.
Your image is still looking good though, man. Just have to sit there and tweak the hell out of it to get what you want.
I think there is at least one clip file around here for getting good ripples in sand, I haven't been able to find it yet though. It my be buried in a seemingly unrelated thread. I'm sure I remember that there was one shared though. Actually I was wishing that there had been one in the TG3 pack that came with the download before you brought it up. As far as I understand it they require either functions or an image map. My preference would be for procedural. But I saw an Image by Kadri that used an image map, and it looked pretty good.
If I bump into that thread I am trying to remember, Ill post the link to the ripples... If I'm even remembering right in the first place.
Thanks Badger, if you find something that you think'll work feel free to shoot me a PM. I think I'll give it another go tonight and see what I can come up with. I don't mind using an image but I'd prefer procedural too since my end goal is to build out something that'll allow me to virtually wander around the place and and "photograph" it as would someone traveling in a real location. I'm always amazed at all the new vistas that present themselves when I wander around a world once I've created it that weren't part of my original composition.
Here you go with 'sandelO' clip file ...thanks to Martin (dandelO).
Awesome! Thank you!
-Greg
I've been working with this sand shader now and it works great for flat surfaces but I can't figure out how to apply it to the foreground part of this image in the manner I've envisioned. The idea is to basically "fill in" all the little valleys in a stepped manner; the bottom gullies would fill to a point and then stop once they've ramped up against the wall. The the layer above would fill, and so on until I cut the whole process off at whatever elevation I want. It's a real brain bender for sure. It's kinda like how snow looks on a mountain. You only see it where it lands on the flatter parts which get filled in. The steeper parts remain exposed.
Yes, I had this problem many times. The only way that I was able to find in the forum is to limit the shader in a way similar to a population (as with the bounding box), and then use additional sand shaders controlled for different slopes and elevations... If I understand what you meat?
But even so, others may know a better way.
I think Blue nodes, redirects, and color hsv nodes are the answer. The problem is I don't know the question. Your balls hurt, my brain hurts, and the bourbon isn't helping me anymore.
;D
Did you use displacement intersection for your sandy area? In that case the wavy displacements added to that 'layer' would only be in the sand.
And for ripples you need to use a get position in texture, an X or Z to scalar, a divide scalar to get the size right, a sinus and a warping setup, perhaps a transform shader to rotate the ripples, and finally a displacement method, and some patchyness by a fractal.
Really enjoying this one - great scene with interesting development. Look forward to the finished result
Hey Dune,
Choronr was kind enough to post a tgc file that handles the ripples in the manner you described which I've hooked into the flow. The issue was I just couldn't get them sit where I wanted. They looked great on flat areas but fell apart on anything else. I think I need to build a very simplified scene first and just get it working there before trying to meld into something as complex as what I've bulit up so far. I also need to learn to label my stuff better so I know what the hell I'm doing. My Nuke scripts are only marginally better organized, and that's just so I don't get laughed at work if I need to pass one off to someone who actually knows what they're doing.
With regards to the sand ripples, the file Choronr posted had ripples with sharp cusps. I adjusted this by changing the ridge smoothing so they weren't so sharp. They feel more natural now, but what I'd really like to do is shift the bias of the cusp across the ripple so it favors one side or the other, so instead of the highest point of the cusp of the ripple always being center ripple, it would maybe closer to the edge of the ripple. Is this possible?
I'll post something later today for everyone to get a chuckle at.
Thanks again everyone for all the help.
Another question Dune,
I'm using a Surface Layer to add the sand ripples shader, which I assume is the correct approach.
Next I've selected Displacement Intersection under the Effects tab as the method of displacement merge. I've left off Intersect underlying as it didn't seem to do anything. I've set the appropriate altitude and slope constraints to limit the sand ripples to just the parts that I want. Now here's the problem: When I connect the sand shader to the Child Layers input triangle on the now configured Surface Layer, about half of the main displacements from the main upstream flow from the Input Node vanish. Only the biggest displacements reamain. All the lower level stuff vanishes, stuff that's still above the sand level. It shows properly when I don't have the sand ripples input and am just using Enable Test Color to verify everything's working, but the moment I add the ripples as the child have the terrain displacements vanish.
Any insight?
Thanks again
Greg
I very like the first one !!!
The tgc with the ripples was made before the new fractal settings were on board, hence the sharpness. And I know what you mean about the shifted top, having experimented with that as well. I don't think I can find those experiments easily, but remember it was done with an additional cosinus of the same X to scalar, and bias add or subtract functions. I'm sure some of the Math guys have excellent solutions for this. If I find something I'll post, but don't hold your breath.`You could also try something with a tilt and shear shader to shift the tops a little to one side.
If you have the x to scalar + sinus set up, you have straight soft lines, if you could add a number to this base and thus shift the line a little to one side, you could add or multiply these values for a bias to one side, then continue with the rest of the line of shaders.
The problem with the vanishing of the rocks has to do with the compute terrain patch size; if you decrease that they will (probably) re-appear. And/or changing the values to smaller values in the displacement intersection.
Hope this gets you on the way....
EDIT: I quickly put some nodes together, but I'm sure there's a better way.
Well Dune you've certainly given my a lot to think about. I've set up render that'll take about 2 hours to compute. I'll put it up when it's done using these new sand ripple ideas.
Thanks again
-Greg
Well here's what I've managed thus far with the sand ripples:
[attach=1]
Overall I think they're working. I still would like to make the texture of them have a little more variation in color, tone, and size of individual sand grains, but that's just fiddling at this point. It feels a little too powdery still. Also once I've got this to where I like I need to start adding all the dead little twigs and other debris one sees on the ground in places like this. Still a little too sterile at this point. I'm still looking for some tall grasses if anyone knows where to find these things here on the forum. I've looked and haven't found much. I saw some Walli posts that looked promising but the links were dead.
thanks again for all the comments and help everyone.
-Greg
With some no-displacement colored fractals and/or surface layers of color you have your color variation in no time. And you might need to add some warp to the ridges to make them more diverse and 'windblown'.
I was playing with a fractal warp on the ridges at one point but removed it when I was trouble shooting something else. I'll have to put it back on. Thanks for the reminder.
Greg, if you're still using Speedtree in this project then perhaps the grass primitive in the ST geometry types might serve your purpose. The sand color seems to have a 'temperature' value much 'cooler' then the mid ground values to my eye, but the flow of the composition with the ripples looks great.
Thanks zaxxon, I'll take a look at Speedtree grass primitives. I still intend to adjust the color of the sand, because you're right, it feels temperature wise a little out of place and dull. I also think some of the cuts in the sand feel a little "cuspy" and hard edged, something I think adding a fractal warper might help adjust. I'm also thinking about bringing back some of the old ground level rock stuff because it's all sand now.
Thanks for the input.
-greg
Quote"(...) what I'd really like to do is shift the bias of the cusp across the ripple so it favors one side or the other, so instead of the highest point of the cusp of the ripple always being center ripple, it would maybe closer to the edge of the ripple. Is this possible?"
Like this ? http://www.planetside.co.uk/forums/index.php/topic,16174.msg158959.html#msg158959 (http://www.planetside.co.uk/forums/index.php/topic,16174.msg158959.html#msg158959)
It could be done with a combination of sin functions, to obtain this kind of profile :
[attachimg=1]
On the scheme, wind is coming from the left, so F1 is the windward face, and F2 is the slip (collapsing) face. F2a is for a sand dome profile, F2b is for a dune profile.
Xm is the collapsing point position, 0,72 in that case. It can be changed in the functions.
You will also need a modulo function to repeat the [0->1] values from the coordinates.
Don't know if I'm really clear... perhaps I should write a tutorial... :P
It would be best if you put this together in blue nodes in a tgc. That would really be terrific...
This is exactly the profile I was thinking if. Most sand ripples I've seen in photographs tend to show a bias as the wind forms them. Always having the cusp center mass limits the realism in my opinion. I wish I knew more about the math nodes too Dune, but then again I did manage to figure out how to make the rock formations using them from someone else's assembly of them.
Math isn't my strongest point either, but I've been fooling around with the blue nodes, so I know a little. By the way, a tilt and shear shader can tilt the sinus ripples somehow as well, but it tilts the whole ground with it.
Speaking of that operation, I haven't had much success using that operator yet. Are there any good examples on the forum of its use?
This image is getting really cool. The color, shadow, lights are awesome!
Cant wait to see the end product :D
It's quite simple; it tilts the X or Z displacements (or y if needed) by a factor, with the base height as pivot point. So if at 100m high, set at 100m high, or it will tilt from 100m below, moving the whole terrain sideways.
I'm taking a huge step back on this until I can figure out how to do this correctly. No color corrections this time. The issue is that I can't for the life of me figure out how to get the sand ripples to lay over the rocks without also adding their displacements to them. I'd also like to vary the pockets that the sand falls in but it seems that all I have available to me is based on slope or elevation. I suppose I could use a simple shape shader to limit where things fall but I was hoping to find a more procedural approach so when I'm done with this particular vista I can wander around the landscape and "take pictures" of other parts as well without having to retool the new camera location.
I have to say that masking different types of topology, be it the rocks themselves, the sand that flows over them, and even the fake stones is proving challenging. I'm trying to think of this like a compositor and lay things down in a logical fashion. Is it better to work from the "ground up", where your lowest levels of the terrain are at the top of the flow and subsequent rises are added as you work down closer to the planet or would one work the other way, establishing the large structures first and then sort of carve out the ground level detail as you move along? I've done it both ways here and it seems that both approaches have their pluses and minuses.
Anyway, here's where I'm presently stuck. I can't get the sand ripples to ride over the base rocks without also adding the same ripples to the base rocks. Any insight would be mucho appreciated.
thanks
-greg
Dune,
That almost looks like the profile I want for the sand ripples...
Interesting.
-greg
wiwine, that looks exactly like what I want. My kingdom for something I can study that does that.
-greg
Greg did you try to use the merge shader? It does have many possibilities.
And one more basic thing is if you don't want that one feature influences the other one at all
you can use two planet objects (for limited places in the landscape you could use displaceable plane object etc).
One for the rocks and one for the sand etc.
Not sure about render times i haven't used this method.
You could then put some lower scale displacement for rocks,sand
in which planet you want by copying the same nodes and lowering values etc.
Looking really good Greg, better with each one.
Thanks for the compliments.
I have used the merge shader with some success but the trouble start when I get to the second tab over with regards to how the shaders should be handled. There's Mix, Add, and then a bunch that don't seem to be what I'm looking for. What I think I want is an Over operator. Mix I think of as a blend, add is additive, which I don't think I want, and the others don't seem to make a lot of sense for what I want to do. What I think I want is for the sand ripples to sit "over" the rock floor, not become additive to the floor. I don't want the ripple displacements to also add ripple displacements to the rocks. This is why I'm wondering if I'm going about this wrong from a build up perspective; should I grow the main rock structures out of a sandy floor base, or should I start with the rock structures and then layer the sand around them?
I'll need more time to think about it.
-greg
These are the kinds of issues a lot of us (me too) are pondering about and experimenting. I don't really recall if I got that solved (ripples also in stones). One idea is that you might need to reduce the size settings in the displacement intersection tab. It's basically at 2/-2/2 or something, but if you reduce the first to a millimeter or so, the second to minus 1 and the third to small + you can see changes occurring. That might help.
The thing Kadri mentioned should work also, I did that too several times, and doesn't add much in render time (not in my case at least), but you most liley get a sharp distinction between the two planets layers. Needs playing around a bit, I guess.
Hi Greg,
To have sand "over your stones" you need to change the order of the shaders probably.
There's probably 2 ways of fixing this and always takes a bit of play to figure out what works best. It really depends on how much displacement you have going on and how many times you are compute the terrain already.
First way is to have your fake stone layer and connect those as a child to a surface layer.
Set the surface layer's smoothing to 1.
This will use the render state of the first compute terrain node up in the network from that surface layer.
It allows you to cancel out displacements for those stones and displace them separately.
Then add your ripples.
Second way is to have fake stones first too, then compute terrain with small patch size, then add your ripple layer.
This will probably be pretty slow to render!
If you look up some of my canyon renders you can see I managed to get it to work.
If you like I can dig it up and have a more exact look.
First let me know how it goes with these 2 suggestions.
Cheers,
Martin
Yo!
Not sure of what exactly is your problem here, but if you want to do something like that :
(http://f1evv.free.fr/tg2/roche-et-sable-0007.jpg)
... check this file (http://f1evv.free.fr/tg2/roche-et-sable-sur-deux-plans-0007-bis.tgd)
It's on TG2 (because TG3 don't run on my computer), but, I think, It could work well on TG3.
Really nothing complicated here...
Just 4 things :
1 - The merge of the 2 textures is done with a merge shader (option Choose by altitude : Hightest). This shader can break if you have lateral displacements (on TG2, I don't know for sure on TG3)... You can solve this problem with a simple trick... Delete the merge shader and apply your second texture on an other plane (or planete, or whatever)... We do it with lakes, so why not...?
2 - If you want your sand to follow the relief you must use 2 instances of the same relief... On this file the nodes "TERRAIN HARD" and "TERRAIN SMOOTH same seed and size than TERRAIN HARD" are the same, except the second one is smoother. ;)
You could also do that by using a surface shader with smooth option ON (0,5 or 0,75 value).
3 - Because of point 1 issue, warp on your sand isn't necessary a good idea. So I prefer to use a vector deformation ("VECTOR TO POSITION" group). Also you can use that to "elongate" your dunes.
4 - After that you add fake stones and plants, and your good. I hope this could help. And nice work! ;)
Thanks Martin and Chinaski, I'll take both your suggestions and give them a try.
-greg
Here's Chinaski's nodes just plugged in as ingested into my script. Overall they're working, I just need to get the scaling working better and adjust the shapes and overall surface properties. I am wondering why they're also displacing the underlying rocks in the foreground, I'll have to see where that's coming from.
Since I don't know a lot about the blue nodes and this is almost an entire blue node setup it's going to take some time for me to figure this one out!
thanks again
-greg
Sorry, was longer than expected (I hate my PIII)
So, this time no more merge shader (by the way, on your last render, and if I must guess, I'd say it's a merge shader issue, something broke it)... I work on two differents planes, try dunes tweaks, and add some fake stones and vegetation. I also made bigger displacements ; This way you can see that the sand follow the relief and "jump" on the rocks.
I think the last one is the simpliest, and the best one for reverse engineering. ;)
(http://f1evv.free.fr/tg2/roche-et-sable-0007-ter.jpg)
(http://f1evv.free.fr/tg2/roche-et-sable-0007-quater.jpg)
(http://f1evv.free.fr/tg2/roche-et-sable-0007-quinto.jpg)
Tired. Sleeping.
Forgot to say one thing : There is not a good and a bad way to do that. The two method are very good. With Martin's technic you work on an axis (and repeat & deform), with mine you work on cells (and deform with vector). The first one give you a regular aspect, the second one a cheesy aspect. Maybe you could try to mix them together... ;)
NB: About the cell method you can read this thread (http://www.planetside.co.uk/forums/index.php/topic,15345.0.html). Efflux explanations are, I think, the best way to understand it.
Thank you for your generous work and great shared examples. All the trig talk with blue nodes is a bit heavy, but these types of shared files really help the process of understanding TG's more 'esoteric' abilities. Again, thanks!
Good to see you around and posting Chinaski! Nice files as always, thanks from me too :)
I agree. I'm really glad that this thread has developed into something that anyone can take something from and not just about my images. I can't wait to get home tonight and plug these in.
Thanks again for the help!
-greg
These are the best kinds of threads from a learning perspective. When people start sharing everything they know on a single subject, man, you can learn a ton fast! Your image and project creates the common denominator. OF course, now we are all going to want to see what you do, greg! ;D
Ha.... no pressure... :o
It's a good thing I have an understanding wife.
Hi Chinaski,
So I managed to plug in your 3rd version into my script and overall it's working, but I can't figure out how to get the sand to not crawl up the side of the rocks in places. You can see on the foreground parts here that there's sand on places that I don't want it, mostly over on the lower right side of frame. I really only want it on flatter areas. It shouldn't be on vertical parts around the base of the rocks themselves.
Is there a way to control this? I tried using a surface layer and the slope controls but it's just not working.
For the most part the sand is looking great, I just am going crazy trying to control how it rides up the sides of the rocks where it shouldn't.
-greg
Yep, that's not the result we expected. :D
Can I please see your file (not sure I can do something, or even open it in TG2, but...)?
I'll send it to you.
Thanks for helping me out.
-greg
Ok, since you're in 2.5 I've stripped this down to the minimum of nodes to (hopefully) avoid possible version issues using mostly what you've already posted but switching out some of my displacements for yours. It still illustrates the same problem even if it's not the exact same file I rendered in my last post so if you can get it working here I should be able to get it working on my end.
Thanks again for taking a look.
-greg
Hi greg,
QuoteI just am going crazy trying to control how it rides up the sides of the rocks where it shouldn't.
So this was one of the original problems right? And its because you can only limit by hight and slope one time, which effects everything?
Oshyan and ulco told me how to deal with that, but I don't remember what they said. I tried to paraphrase earlier. Basically I think you need to have more then one layer of the sand limited by area, so that you can use more than one distribution shader, so that you can have different levels of control for altitude over your not flat terrain.
Is that right Greg, and guys? If not, the issue I described is still a problem for me at least.
???
I think I pretty well understand the concept behind Chinaski's approach. He's using two planets (or planes in his posted case) to generate his displacements and then combining them. The problem I'm having is the "sand planet" has sand on areas that I don't want it and it's seeping through to the main terrain planet, manifesting itself in ways I don't want.
I have managed to control the slope of the sand on the sand planet so it really only shows up where I want it but the problem is that it leaves behind areas of black that carry on through to the main rock planet. I tried making a mask layer from this slope controlled sand planet and merging the two together but again, it created black holes in the main render.
I know this can be done, he's already got it working on his file he posted but for some reason when I swap out my displacements for his part of the terrain parts of mine break.
Hmmm guys, we are in trouble!
(http://f1evv.free.fr/tg2/TG2errors.gif)
I'll try to figure out, but it will be hard (working sort of blind here).
About the second plane (or planete or whatever object) trick... If I have two planes I don't need to merge my textures. It's just a little solution to a specific problem with the merge shader. I don't even know if this problem still exist on TG3.
On TG2, WHEN (and only when) I have artefacts with the merge shader, I thinking like this: sand = water, problem solved! So I make a "lake" of sand, and it's solve my problem. It's not elegant, it's a cheat, but it work (typical example: sand dune on bottom of a martian crater).
Now, if you want to distribute some sand in your scene, TheBadger is right, you need a distribution shader (or a surface shader) to place it in the scene.
I gotta say this setup you made Chinaski for the dunes is very cool. It reminds me a lot of these old fractal patterns I used to mess with way back in the mid nineties during my early Lightwave days hanging out with a guy named Steve Worley. He used to come over to my house when he was in town and show me all his new fractal based plugins for Lightwave and the patterns they made look exactly the same as what you have here. It's like a lightbulb just went went off in my head and I now realize what this things do.
I think I'm getting a grip on it now.
Thanks again!
-greg
Greg you mean the famous " http://www.worley.com/ " ?
Hi Greg,
You should be able to do it with Intersect Underlying alone, see some of my examples below:
http://www.planetside.co.uk/forums/index.php/topic,6272.msg66508.html#msg66508
http://www.planetside.co.uk/forums/index.php/topic,11967.0.html
http://www.planetside.co.uk/forums/index.php/topic,6366.0.html
If you like you can send your file to me, my e-mail is in my forum profile (just click on my nick for the info).
We should be able to sort it out.
Cheers,
Martin
Yes Kadri, that's him. He used to write custom Lightwave plugins for me a long long time ago. He'd charge me $20 for each one and write them on the 40 minute plane ride from Los Angeles back the San Francisco. Some of them later became part of his Polk collection. He wrote one once that he called "Disgusting", which was this huge panel that had probably 100 sliders on it and a small viewer window. That little window displayed what all the fractal patterns were doing, and they looked pretty much the same as what TG's math nodes do now. He never did sell that plugin. I don't know why, it was very powerful. He also wrote a lighting control plugin for us for Lightwave that later he sold called Gaffer. He made a bundle on that one. After than came FPrime. That was a huge hit.
There was another guy back then named Ken Musgrave that worked at DD at the time. He and Worley were friends. Ken later went on to write Mojoworld, but when he was at DD he wrote the code for the pyroclastic clouds rendered for Dante's Peak. I still talk to Ken off and on to this day. I haven't talked to Worley in years.
Those were the days.
Hi Martin,
I'll send you the whole file. I tried intersect underlying and it didn't seem to make any difference, but odds are at this point like a lot of things I've discovered in life, I'm probably doing it wrong.
I'll send it out soon.
Thanks for your help.
-greg
Thanks Greg!Don hesitate to write about such memories.
Don't know the others but i certainly would not be bored to hear more :)
Hey, I still have all of the procedural texture .dlls that Worley wrote for Imagine 3D way back in the day (mid-90s). :) His name popped up in another thread recently on adding new noise types to Terragen ... at least I assumed it was him - some of the noises had Worley in the name. Glad to hear he's still around and doing stuff.
I have used FPrime with Lightwave............that plugin used to amaze me for some reason. :)
Are you sure you need all those nodes for the rocks and sand you want? It can probably be done simpler, IMHO. Anyway, it's computing tediously slow on my machine, so I can't do much at this moment. But adding a surface layer right at the end and using the displacement intersection to add the rocks, playing with the settings (even coverage influences things, try 0.1 or 2) and using the compute terrain patch size (and perhaps smooth terrain) might get you somewhere.
I'm having that problem too right now with the render times. It's agonizing. I don't know why it's taking so long at this point other than I've probably got too much going on at the moment. I'm going to try and strip it down again and see what happens.
Martin is also giving me a hand with it so I'll see what he can figure out.
Learning, learning, learning...
Thanks again
-greg
Here's Chinaski's sand layer with a couple modifications to the fractals. This took about 8 minutes to render on my machine, which is a 3.4ghz quad core with 12gb ram. Somehow when I start plugging in the rocks the render times go berserk. Looks like it's going to be a late night.
Your rocks may have too many small displacements, and may be way to complicated in terms of nr. of nodes to build them.
I did this yesterday (on my other machine, an i7 2600k 16gig memory) and it took 17 minutes with detail 0.7 AA4. Just testing some stuff, nothing serious.
Holy cow, that sand looks fantastic, Greg!
Hey Dune,
Your scene looks great and the render times are well within what I would consider acceptable. Let me ask you, how would you keep the sand ripples from rolling up in the stones? Whole most of your rock structure seems to sit nicely on top of the sand, I'm seeing the same problem I'm having on some of them with the rocks looking like they're "growing" out of the sand ripple. It's like the sand is morphing into rock. How do you keep a very discreet boundary between the rock structure and the sand base without having two planets? That's the problem I can't seem to crack at the moment.
Thanks!
-greg
Thanks Hannes, but Chinaski really deserves the praise for that last one.
Now I'm getting somewhere render time wise. This image took about an hour and a half. For a 2k+ render that's this clean, I consider that a pretty good render time.
I'm still seeing the sand rolling up the sides of the rocks in an unnatural way over on the far right background formation, but overall it seems to be behaving. I'm wondering if there's a way to smooth out the sand base planet layer to it doesn't so closely ride the displacements of the rock formations. I think at this point I also need to start thinking about how I want the sand to look. It's a little too extreme right now. I like what Dune did a few posts up so maybe I'll try for something like that. Overall I really like Chinaski's sand nodes he posted. They took a lot of the mystery out of making these structures.
Onwards...
-greg
@Dune: Very very very nice!
@gregtee: I tried to understand your build with what I can see on TG2 (half of the nodes are missing, and a lot of connections are broken)...
The principal issues I can think of:
First, the scale of your scene. Very small (when I work on small details I personnaly use a x10 or x20 scale). And if you reduce too much the scale of my build you're going to have a lot of problems with rocks, micro-displacements, sand... and your render time going crazy.
Second, you have negative altitudes. That's never good in my point of view (TG2 user), can mess up with a lot of things (merging in particular). So I generaly try to compensate and have only positive altitudes.
If you try to solve this two issues, you're going to destroy your scene (lose your POV, which is very nice, change your terrain, also very nice, etc.). So I think you should stop here with this technic on this specific scene... ;)
Now, if you want to know (maybe for futur works) how to integrate my nodes in a scene... I took a similar relief, but with positive altitude and a bigger scale:
(http://f1evv.free.fr/tg2/septimo0.jpg)
(http://f1evv.free.fr/tg2/septimo1.jpg)
The file is here (http://f1evv.free.fr/tg2/septimo.tgd) (contain some notes). See? Classic. Just define a zone to texturise, add a surface layer, plug in, and you are done (in a perfect world, in reality you have a lot a little adjustements to do). The merge shader do its work perfectly. No need of a second plane to simulate it. No interaction with outside the zone. The cut between rocks and sand is clean. So far so good. You can define a second zone, with other textures, and work for hours! \o/
Now, with this first image there is two issues :
- In the bottom right there is an odd displacement. It's because a "rocky" part of the texture is on the border of the zone. There is several ways to solve that properly (modify the blending/breackup of the zone, create a transition zone, smaller, to blend the rock branch before merging, paint a little, etc.)... And ways to ignore it (change the rocks seed, adjust your POV, etc.). I was lazy, I changed the rocks (all in the file, as an example). ;)
(http://f1evv.free.fr/tg2/septimo2.jpg)
(http://f1evv.free.fr/tg2/septimo3.jpg)
(http://f1evv.free.fr/tg2/septimo4.jpg)
- Second issue (and it's about render time)... There is a small zone (background right) where we apply a very complex sand/rocks texture for nothing (nobody can see a sand dot 2km away). This kind of complex texture is for closeup only! Because the render time is 34 min at 0,8 on a 2001 bi-PIII 1ghz with 2 go ram (3 min on a modern computer?), I did nothing about it (it's bad, very bad, I know). You should.
Wow!
I feel like putting down the mouse and going and watching Lawrence of Arabia after looking at your images. I can tell I'm going to have a lot of fun going through all this stuff. Thanks for the notations on the nodes too, it helps me to understand what you're doing.
I feel like I want to take a up a collection here on the board and buy you a faster computer and a TG upgrade. You're a gentleman among men Chinaski!
Thanks and I'll let you know how it turns out.
-greg
This would be a good topic for a tut series. Geekatplay?
Actually, given the pretty big difference in appearance and methods people are showing here, I doubt I would be the only person happy to see a sticky of ready to use sand ripples.
And to tell you the truth, I wouldn't mind a "ripple shader" built into TG in the first place! Seems like a fundamental aspect of most any desert scene (to greater or lessor degrees). And clearly its pretty labor intensive for the user to go from scratch... Is it a possibility Jo?
I think a good answer to a sand ripple shader is something that Nuke has, which is a feature called Gizmos. Gizmos are simply arrangements of nodes that are "wrapped" into a single node with specific controls exposed to the end user to make adjustments. For example, all of Chinaski's sand nodes could be compartmentalized into a single node, and when that node is clicked on, only the sliders and input values he decides are presented to the end user. It looks just like any other shader panel to us, but it's made by him and not the Planetside team.
It's a great feature because it encourages users to write their own "shaders" that can easily be shared and limits the confusion end users have when trying to decipher what the originator of the script had in mind as only the parts that the originator wants them to have access to are exposed.
-greg
Well, the answer to the Gizmo is a TGC, I guess. Save it in a Gizmo folder and pull out when needed. All variables are inside.
I've made my file into a GIZMO aided, very simple tgd. This is my way of doing rocks on sand. It might benefit you.
But I also believe it's not more than natural if sand rides up to rock structures, that's what happens in the real world as well, windblown up againts slopes.
QuoteGizmos are simply arrangements of nodes that are "wrapped" into a single node with specific controls exposed to the end user to make adjustments.
Yeah, in Maya or anything with MEL and python too I guess, you can make any node or set of nodes, as well as UI. In TG you can collapse an entire node tree down into one or two nodes and use that as a TGC as Dune pointed out. But the problem with that is you can't hide parameters like the way you described in nuke. So once the end user gets down into the guts of it, he will be faced essentially with the same problems you are discussing.
In TG it does help a lot to keep work space neat and orderly. But from the perspective of sharing, it requires every node to have a note explaining whats going on, or for the end user to have the same knowledge level of the person who made the shader... which defeats the purpose from the perspective of what you said.
So for TG, really the only way to get what you are describing is for TG to have the ripple shader, and then people could simply share the settings recipes, and connection formulas. Still, that would cut the work load for making ripples down to 1/4 or less of what it is now, I would guess.
@Ulco, is that file you shared of the sand in the image you posted on the last page? I thought your ripples looked nice. I downloaded your share too. Thank you!
Yes, it is. But if you build a Gizmo/TGC and add some notes about which variables can be adjusted and which should be left alone, it will not be too hard. But I don't like hidden stuff, and would prefer all to be seen and changeable. Like the inside working of a water shader; you can't change the fractal flavor...
I downloaded your file Dune and saw your gizmos. That's close to what I'm talking about but the main missing items are the access to the internal nodes input variables inside the first surface layer. When I click on the surface layer all I get are standard surface layer options. I don't have access to any of your internal nodes. Sure if I click the little + to get access to the internal nodes I can edit them but what if you could design your own interface based on your internal nodes? So there's a node called Dune's Sand Ripples that has as its interface when you double click it a bunch of sliders and input fields that just reference back to the deeper internal nodes you've made that produce your sand ripples. This way someone like myself or Badger wouldn't have to dig down into the nested nodes and instead just access thier input variables from the top node. We don't even have to know how your internal nodes work. All we're presented with is what you decide you want us to fiddle with. Now if we wanted to dig down and see what really is going on we still could, and make changes, additions, etc, but it makes for a much cleaner workflow if we don't have to or want or need to.
Are we not just talking about Macros here? Something I wanted to see in Terragen for a very long time. Vue has them and World Machine has them.
I think it would be possible if you put the adjustable nodes outside (in a neat row, in a group) and link them to the inside stuff. But it would be up to Planetside to make a real one-node gizmo with different variables linking to all the stuff inside, like the water shader for instance (which I don't like, although not liking is relative, because it's so secret inside).
Any of the Gizmo/Macro/TGC interface ideas suggested would be welcome. Making the power of TG more accessible with fewer learning curves works for me, I have to believe that Planetside is moving in that direction. Using Vue as an example is a mixed bag, however ease of use to focus mostly on image making with friendly tools certainly would help to expand the user base. BTW, despite all the cool learning examples and high level discussion I still find the original image and concept to be the most pleasing to my eye.
My goal ultimately is go back to that original image but with some of these sand features mixed into the landscape. This was a huge detour for sure but it taught me a lot. And I haven't even put the actual Joshua trees in yet!
Hi Greg,
I didn't forget to look at your file, but haven't found proper time yet....just to let you know I'm still keen on helping you out here.
Cheers,
Martin
Thanks Martin. I'm looking forward to it.
I rendered a 300 frame animation of this thing over the past 4 days that I'll put up in the animation section tomorrow sometime. It's just a simple forward dolly track over about 300 meters of terrain. No plants yet but it's still fun to watch.
-greg
Here's some keyframes from an animation I rendered over the past weekend. Still no veg, and not exactly what I'm shooting for, but I thought a little interesting none the less. These are part of a 300 frame run rendered at HD with pretty high settings so there's no artifacts or flickering of any kind. Times per frame were in the high end because I'm using the two planet gag to add the sand ripples but for a proof of concept anim I wasn't trying to optimize yet. Camera is nothing special, just a straight push in.
-greg
Very impressive you seem to have got the sand to work really well. Great stuff!
Thanks. I'm going to render it again this weekend with vegetation.
The sandy formations look really good in the frame shots, what kind of rendering horsepower in is use here? The addition of some foliage should really enhance what already is a fine piece of work, looking forward to your next examples.
Horsepower is basically Digital Domain's render farm. Well, what little of it I can use that is. I don't get to use the whole thing.
Hi greg
Out of an abundance of curiosity about everything I hear and see in these forums, and since you seem ok to talk about your work. Did you ever say why you are picking up TG? I mean are you doing it just to increase the number of tools you can use, or because your job told you to learn TG?
Again, Im just curious is all. I figure there are a lot of people in these forums who have cool jobs and use TG, but only once in a while do people talk about their pro work (really unfortunately). So it can be real informative when people do talk about it.
And I'm just a interested person anyway.
I like this very much especially the colors and contrast, which give these renders an exceptional sense of realism.