I've been working for maybe... oh... several years. Trying to make natural looking canyon walls. I've got some approaches that turn out looking alright, but very "fake" or cartoony almost.
(http://i.imgur.com/a2yMLMu.png)
Maybe someone can explain how to setup a start, or look at my start and advise?
Have you tried working from the preset? Most of the people who know canyons and other complex procedures on this forum are pretty dodgy when giving out details on techniques.
Quote from: Upon Infinity on November 28, 2015, 04:50:29 PM
Most of the people who know canyons and other complex procedures on this forum are pretty dodgy when giving out details on techniques.
Not sure that's fair. What WASasquatch is looking for is something we've all been trying to achieve. I think the discussion and development would go further if we all understood the problem better. Lets have a real life image we can all look at and work out how to work out further.
I guess the provided scene gives the base shape in the first three nodes. I want to see what detail he's looking for TG to provide in reference pictures.
Here are reference images that can help.
https://duckduckgo.com/?q=canyon+walls&iax=1&ia=images
I met some success with a distant canyon from a bird's eye view but close up is a whole other challenge. It looks like you have the general shape started well. Having the correct colors is the easy part. Maybe we can stretch some function nodes and i was looking for a function to use for that but with no success yet. I may already have some ideas in that area. That layered chiseled and deep cracked mix that is so hard to duplicate. And lastly of course eroding the entire structure correctly.
Quote from: Chris on November 28, 2015, 07:48:29 PM
Here are reference images that can help.
This looks like a good one to start with
(http://www.virtualoceania.net/australia/photos/landscape/grampians/au1592.jpg)
Quote from: Hetzen on November 28, 2015, 05:41:56 PM
I think the discussion and development would go further if we all understood the problem better. Lets have a real life image we can all look at and work out how to work out further.
I think Hetzen is right, we seemed to go quite far, really fast with the eject topics, far more has been learned and discussed then in the past years.
Users that would like to join in and help solve these issues for the world of TG users sounds like a great idea to me.
From what I planned in the blue nodes, you can see I have some pretty "decent" channels that I mixed with a voronoi billows (for buldges). My theory was that it would pan out like cracks and ridges in lateral displacement. In practice though, it seems the translation to displacement isn't exactly as pictured and coming off a lot softer and rounded. It also isn't taking on the form I'd hoped for which means the angles I have based on X/Z are not going to work I don't think.
The reason I have new interest is viewing the beautiful pieces of René like this canyon wall which is just stunning (must be part of Terragen Galleries) https://www.facebook.com/photo.php?fbid=934437656638635&set=gm.915201408605883&type=3&theater
That one photograph looks to be a good one to work off of.
René's canyon wall is very nice. His coloring represents the region fairly well. Erosion is needed otherwise the canyon lacks any type of geomorphology other then the main feature itself. Fluvial and weathering effects are needed on the smaller scales.
I found way to flatten certain smaller scale noises in varying alternative methods and stretch them out so i will post some ideas soon. The sheer rock is caused by weathering called exfoliation. Sheets of the exposed rock slip away and fall due to cracks or joints within the rock. This is what i mean...
By the way Dune's square rock tests as good to browse over also.
Quote from: Chris on November 28, 2015, 10:59:21 PM
Erosion is needed otherwise the canyon lacks any type of geomorphology other then the main feature itself. Fluvial and weathering effects are needed on the smaller scales.
That's actually a good point. I've made a bit more head-way and rebuilt my function. I'll post it up in a minute after I try to do a bit with Daniil's Erosion Shader. Not sure how will it will work AFTER the strata, but I'll try.
The big thing I have discovered is "Lateral Only" is based on the inner most displacement points of the slope. So you displacement you add with appear more smooth as it's blending with the outer displacement of the slope which extends PASSED the displacement you are adding.
To solve this you use "Lateral Normalized" which uses the displacement surface of the slope from what I gathered. This just all being my explanations of what's happening with these settings for the Displacement Function.
Quote from: Chris on November 28, 2015, 10:59:21 PM
I found way to flatten certain smaller scale noises in varying alternative methods and stretch them out so i will post some ideas soon.
I look forward too it. Thanks for all the help Chris.
As a note: Due to Render Times, I think Erosion should be something we look at towards the end. I'm on almost 20 minutes of a test render with erosion masked. Actually looks "decent" for the erosion effect. Could be smaller scale though.
Finally finished at 40 minutes.
(http://i.imgur.com/GwmssIv.png)
There are docs for the displacement methods here...
http://planetside.co.uk/wiki/index.php?title=Default_Shader_-_Displacement_Tab
- Oshyan
Erosion should always be a last resort until you know you are satisfied with the overall terrain. Those adjustments can take enough time as it is. Not a problem helping out in this. This to is quite a challenge so why not be up for it. ;)
Thank you Oshyan for the link. It always helps to refresh the memory every now and then.
(http://i.imgur.com/krhD7Id.png)
Here is what we have going on now. Looking a lot better.
Quote from: Oshyan on November 29, 2015, 12:59:12 AM
There are docs for the displacement methods here...
http://planetside.co.uk/wiki/index.php?title=Default_Shader_-_Displacement_Tab
- Oshyan
Thanks for the link Oshyan. Not sure why, but there was definitely a lot more detail present on Lateral Normal.
So with this it seems there needs to be yet another, larger scaled displacement layer to create even more varied jetted areas and compressed areas with detail. Lateral like cliff hangs. Also needs possible another, even smaller scale crack setup.
Also there are some rough spots that need some polishing. I'm not sure where they are coming from though...
(http://i.imgur.com/uEXrTX0.png)
Certain a lot closer in certain areas. Here in red is what i see and green still needs to be achieved, at least from what can be seen so far. A mix of more flat, some softer corners and wider strata.
I am currently rendering a very crude example of sharp scarped cliffs. There are several ways of doing what i have in mind so as crude as it may be, there are still man ideas left to explore.
Quote from: Chris on November 29, 2015, 02:49:52 AM
Certain a lot closer in certain areas. Here in red is what i see and green still needs to be achieved, at least from what can be seen so far. A mix of more flat, some softer corners and wider strata.
I am currently rendering a very crude example of sharp scarped cliffs. There are several ways of doing what i have in mind so as crude as it may be, there are still man ideas left to explore.
Good eye, what I think can be done is a surface layer, using the base cracks as a input, and child layers as a new function setup for the new desired target look. Masked with a PF at high smallest scale settings and high colour contrast settings to give us are varied setup like seen in the image. Also maybe limited to a altitude as shown above where it seems to be common to lower altitudes. However that I'm not sure how to do besides a simple Displacement Shader to Scaler and Colour Adjust 0 - 1000 to create a height mask from displacement.
I tried something similar but I broke the original setup somehow and can't track down what I did wrong. Too late. I'll re-download this and try again in the morning. Maybe you can give it a shot.
May need to also use a Surface Layer as a Displacement Shader alternative to limit the displacement more to the walls and not the plateau buildup where it seems to go a bit crazy.
You're getting pretty close. I've had a quick play with your initial file and changed some stuff. Nowhere near, but it'll give you some additional lines of thought ;)
My smooth voronoi cracks file - which is free from the NWDA store might help here.
So I have some larger, more X stretched noise for the larger bulbous areas, but I'm having trouble masking them through. Are my scales wrong or something? It just doesn't want to mask in even with Colour Contrast at 10 making a hard mask. Or maybe the design for the larger variated clumps is just wrong. I'm not sure. Also seems to be causing some displacement breakup (sky showing through). Oh the headache. xD
I managed to put some tests together but some parts may work well more then others. Use what ever parts may help. This has been very hard without a clear way of the proper fractal noise/s.
Quote from: Chris on November 29, 2015, 02:49:52 AM
Certain a lot closer in certain areas. Here in red is what i see and green still needs to be achieved, at least from what can be seen so far. A mix of more flat, some softer corners and wider strata.
I am currently rendering a very crude example of sharp scarped cliffs. There are several ways of doing what i have in mind so as crude as it may be, there are still man ideas left to explore.
Just in case this helps you guys with visualisation I've added a bit more geological interpretation to Chris' photo. Think the strata at the bottom is thicker than the upper layers hence the larger amount of vertical joints and horizontal layer boundaries. The red I've added seems to look like an erosional cut down (think a fast flowing river) into the yellow layers as they seem to terminate at that boundary. The blue layers have then filled the channel (see how they lap onto the red). At the top it has filled so you get thinner layers from maybe a more slow moving body of water. If you can put these properties into your models would certainly help with a more heterogeneous look.
N.B. the erosional cut down could also be a fault, but I don't see a damage zone so erosion seems more likely :)
To me it seems like what is needed is an iteration function or a function that splits a fractal noise into frequencies (size variant gradient) using features to drive it by slope and or altitude. Certainly several fracture layers can be blended but having the fractures be a consistent flow might be hard to achieve.
Quote from: Chris on November 30, 2015, 06:07:29 PM
To me it seems like what is needed is an iteration function or a function that splits a fractal noise into frequencies (size variant gradient) using features to drive it by slope and or altitude. Certainly several fracture layers can be blended but having the fractures be a consistent flow might be hard to achieve.
I think for a basic setup we can differentiate the noise and have a basic shader, but yes, a more accurate, slope driven function would be appropriate. This, however is where I would be at the end of what I know of functoots. I still do not understand the functions clearly from docs, and my math dyslexia makes my literally cry sometimes on frustration beyond simple times tables that I also cannot retain.or work...
I'm still exploring various functions. Maybe, just maybe there is a way to create something like a frequency function at least.
Quote from: Chris on December 01, 2015, 12:05:57 AM
I'm still exploring various functions. Maybe, just maybe there is a way to create something like a frequency function at least.
Yes you can Chris. By using Voronoi 3D Cell, you can create 'random' intervals with the greyscale output. Because you can't blur the edges, you will get sharp lines where the shade changes between the cells and then get ugly edges. What you can do is use a Voronoi 3D Difference noise to multiply the Cell noise to always fade the edge to black, as long as they have the same seed and scale as well as position, they will align with each other. To tighten that edge, you can use a Colour Adjust, or, you can use a Smooth Step, where the low is 0 (black) and for the top end say 0.1 to become white. This will plateau the gradient of the V3D Difference noise. Try and play with Get Altitude in Texture to drive the noise functions instead of a Get Position, this will give random banding through altitude, which you can use to break up other noise effects.
Using the above technique, you can get variation in altitude effects.
By the way, thank you for chiming in WebbGeoscience. Good of you to contribute your expertise. :)
Quote from: Chris on November 30, 2015, 06:50:29 AM
I managed to put some tests together but some parts may work well more then others. Use what ever parts may help. This has been very hard without a clear way of the proper fractal noise/s.
Sorry Chris, that doesn't just drop in. Can you send a scene file.
Just make a Simple shape shader using it's displacement inverted/subtracted about -100 meters and smooth step about 50 meters, the clip files will be placed in the Shader group in between the Compute terrain and Planet node. The file is designed to displace only the lateral sections of a terrain.
Quote from: Hetzen on December 01, 2015, 05:28:30 PM
Quote from: Chris on December 01, 2015, 12:05:57 AM
I'm still exploring various functions. Maybe, just maybe there is a way to create something like a frequency function at least.
Yes you can Chris. By using Voronoi 3D Cell, you can create 'random' intervals with the greyscale output. Because you can't blur the edges, you will get sharp lines where the shade changes between the cells and then get ugly edges. What you can do is use a Voronoi 3D Difference noise to multiply the Cell noise to always fade the edge to black, as long as they have the same seed and scale as well as position, they will align with each other. To tighten that edge, you can use a Colour Adjust, or, you can use a Smooth Step, where the low is 0 (black) and for the top end say 0.1 to become white. This will plateau the gradient of the V3D Difference noise. Try and play with Get Altitude in Texture to drive the noise functions instead of a Get Position, this will give random banding through altitude, which you can use to break up other noise effects.
This technique is certainly useful for other ideas in mind but this is what i mean. I forgot to mention in my previous post that the frequencies need to be reduced in size along the slope gradient like this crude illustrated example.
That's a neat little flock of blues. Just playing with it. If you rotate it, and add as child (or whatever), you can get nice blocks. Another technique again, good to have at hand. Thanks, Jon.
Quote from: Dune on December 02, 2015, 03:08:13 AM
That's a neat little flock of blues. Just playing with it. If you rotate it, and add as child (or whatever), you can get nice blocks. Another technique again, good to have at hand. Thanks, Jon.
Can u give some hints how u do that? I am not really good with blues but would like to learn what u said. :D
Very nice Hetzen, as Dune suggested rotating it can produce some interesting results.
Maybe some hints on rescaling the noise like that in this thread called 'Rescale Noise', Chris... ;)
http://www.planetside.co.uk/forums/index.php/topic,12144.msg122162.html#msg122162
@archonforest: rotating has nothing to do with the blues, you can just use an ordinary transform shader, rotate X(Y)Z by 90º and merge that with the original (or add as child for 50% or so, lots of ways). Now if you mask that by some vertical billow you may get interesting shapes, where the squares only appear where the billows are and you have some smoother gullies left over, which give the appearance of erosion. Just play with it.
By the way; the same principle goes for mogn's square blues setup from a while ago, though that's a little more nodes, but also more variation.
Quote from: dandelO on December 02, 2015, 10:08:59 AM
Maybe some hints on rescaling the noise like that in this thread called 'Rescale Noise', Chris... ;)
http://www.planetside.co.uk/forums/index.php/topic,12144.msg122162.html#msg122162
Alright, that looks like it has potential. As long as we can re-scale along a slope then it's one step closer to one part of this dilemma solved.
Okay, simple addition to Jon's Blues, for Archonforest (and others).
Thx a bunch! 8)
Nice work Ulco.
I had a quick go at creating a Canyon lands base shape. It's a bit rough, but there maybe some interesting things in there for people to use.
Also posted a quick render with Ulco's base method and the above scene, but I didn't like the regularity of the displacement bands those blues give, so will work on something to break that up.
Quote from: dandelO on December 02, 2015, 10:08:59 AM
Maybe some hints on rescaling the noise like that in this thread called 'Rescale Noise', Chris... ;)
http://www.planetside.co.uk/forums/index.php/topic,12144.msg122162.html#msg122162
Good to see you around Martin.
I think i may have an idea using blend modes to lateral fractal colors.
Quote from: Chris on December 03, 2015, 06:03:51 PM
I think i may have an idea using blend modes to lateral fractal colors.
Cool, post your tests. I used a colour adjust to temper the blacks to multiply a colour fractal in the second image.
Quote from: Chris on December 03, 2015, 06:03:51 PM
I think i may have an idea using blend modes to lateral fractal colors.
This is getting intense. I have a setup which is just gorgeous looking, I think. But with my detail/aa limitations I am not sure 100% or if the small scale anomalies I see are just quality or displacement.
From what I am looking at it is just really nice looking, even if not 100%. Only thing I wish there was more of on this new setup combining our ideas is smaller details. I think that's easily added with just a PF honestly though.
Soon, i am still testing the martian flake stones. Rendering takes a while but this is next on my to do list.
Quote from: Chris on December 04, 2015, 12:04:57 AM
Soon, i am still testing the martian flake stones. Rendering takes a while but this is next on my to do list.
My test was not as I anticipated on a larger scale formation... little upset Haha. Back to the drawing board.
The laterals you use, Jon, are too rounded, IMHO; I've used the same squares for those, but smaller.
Sort of related but i am weighing in all options to be explored. It still needs work in the spiky small scale displacements.
A simple strata slope reduction test but it would be better served to have a more even noise gap for the octaves.
Quote from: Chris on December 04, 2015, 02:37:21 AM
Sort of related but i am weighing in all options to be explored. It still needs work in the spiky small scale displacements.
This is actually pretty nice. Is there a way to taper the fine displacement only to the canyon floors?
The strata rock lines along the steep slopes?
Quote from: Chris on December 04, 2015, 03:54:50 AM
The strata rock lines along the steep slopes?
I believe so, the layer ontop of of the highest peaks. I feel like they would be pretty smooth and windswept.
They would be. I think the erosion is causing this but reducing that is easy to do.
Here is what I've finally gotten back on my study. Combining the lateral/vertical noise based on the altitude in texture, and my voronoi cracks setup, and a few PFs for variations.
File is a bit of a mess and I'm hoping to reduce functions down a bit, or figure out why it hung so long during rendering.
No where near perfect, and as you can tell from the left-hand side, you can't get too close without the surface noise getting soft.
Not bad. Please continue to work on this. Who knows what other neat thing will happen along the way.
I think i have a sheer rock cliff face ready to go soon. I have to render the result to see what it looks like more clearly.
Really close now...keep on keepin on...
Quote from: bobbystahr on December 10, 2015, 11:17:09 AM
Really close now...keep on keepin on...
I've come to find out this scene if very heavily dependent on my base PF setup. Basically it's just a SSS canyon with inverted displacement, then a Power Fractal along normal to achieve terrain variance, and another copy of that PF on lateral, both masked by the SSS (to give us a canyon floor)
Without the nice settings I have for this, the lateral detail layers can look very silly on other rock forms.