I meant to post this a few days ago but got sidetracked.
This is an extremely powerful technique. In fact I think it's almost necessary to get good terrains in many circumstances. I know people have utilized things like this because I've seen it but this is the way I'd do it. TG2 has some weaknesses but this is not one of them. It allows you to do it. You can altitude blend terrains so that you can get extreme variations at altitude without it effecting everywhere or you can use it subtly for beaches for example.
If you hook a fractal through Tex coord from xyz then you can use a Distribution shader for altitude.
You obviously have to think about the altitude relationship of your two terrains. Same seed for example or what displacement amplitudes they have. Same seed for two terrains but other slight differences is an obvious starting place.
Here is a way that you can use this technique to good effect. I've negatively displaced a voronoi 3D A and applied strata and outcrops for the higher terrain. You can get more complex and interesting effects by subtracting a voronoi 3D A from another terrain before doing the blend or use other functions you want to modulate it about a bit since in this case, voronoi noise is a bit simple.
You will also be able to altitude blend distort as described in the other thread.
Blending a terrain "by itself" is a nice way to get more variation. I've desribed another way to blend by "altitude" there: http://forums.planetside.co.uk/index.php?topic=4484.msg105229#msg105229 (http://forums.planetside.co.uk/index.php?topic=4484.msg105229#msg105229)
Now, the second one has great potential I think. Have you tried to also control the size of the voronoi function by altutude? If you create smaller sizes on top and let them grow to the bottom, you could almost get the basic look of a fluvial erosion on rocky terrain. Also you could try to blend out the strata to only ~20% where the voronoi cuts in, to make it look smoth-eroded.
Best regards
Sönke
I have to say that this looks very interesting.
Goms,
Yes, good ideas. I've thought of some that. I have a stepped canyon type terrain which better demonstrates the direction this can go but I'm only scratching the surface. Probably today I'll post some WIPs of what I've been doing and you'll see that. The altitude thing is a bit tricky. I tried driving sizes by altitude and it had an unexpected result. It actually seemed to squeeze the shapes vertically - but maybe that's just due to my graph. This is in fact really cool but doesn't work when you move from origin if you have displaced the shapes. That's all a bit of a mess with dodgy graphs and I'll have to go back to study it.
The great thing about blending terrains by altitude is that you can apply quite extreme effects to one part of the terrain but this won't be all over the terrain. This is one of the key successes of using this technique. You can have very jagged type rocks higher up.
One thing I found was that the very regular unnatural shapes created by the voronoi is not a problem (I thought it would be) because when you add other functions to mess that up a little, it suddenly looks quite natural.
I saw your terrain on the terrain sharing thread and you did exactly the type of thing we usually need to do to get good terrain. It's an excellent terrain.
I have further ideas that I haven't explored yet. Severely blended terrains. Stretching them out on opposing axis and slashing them up etc then using this with altitude blend. It goes on forever as usual.
Nice work Efflux. Those shapes do look like erosion. From what I've found, noise doesn't like being scaled through altitude, as you will always get a lateral shift towards the origin as the scale decreases.
Yep, noticed that too, when trying yesterday.
But I think you also could use more then one voronoi and blend them by altitude. On top you have three, in the middle two and in the low areas only one. I will try this when I have some time to spare this weekend :)
Here's one where I simply ignored whether one terrain relates to another in the "right way". The sharper terrain also supplies a little warping.
Don't mean to highjack this thread with images, however i found a couple of images of a world's surface and sky from the upcoming film Prometheus and thought that the terrains on that world would compliment what you are doing here.
Yes, that looks similar.
I'm pretty much done with the altitude blend thing on here. That's fairly straightforward but I'm definitely interested to explore the voronoi thing way more. You can also subtract the voronoi rather than negatively displace but that's much of a muchness. Just slightly different results. I've added an image of another terrain in my WIP thread where I've furthered this with more voronoi layers.
Goms.
You can create smaller voronoi and stack them up by clamping them but then you have to bring the clamped value back to 0. I just haven't created a specific graph for that. It might work in this scenario.
Quote from: efflux on March 12, 2012, 10:24:37 PM
Yes, that looks similar.
I'm pretty much done with the altitude blend thing on here. That's fairly straightforward but I'm definitely interested to explore the voronoi thing way more. You can also subtract the voronoi rather than negatively displace but that's much of a muchness. Just slightly different results. I've added an image of another terrain in my WIP thread where I've furthered this with more voronoi layers.
The only element i see missing in your results as opposed to the film terrain is i noticed some deep vertical cuts scattered about the inclines. Could be deep, long erosion but there should be a procedural way of figuring this out i would hope. There may be a case were the strata is warped as well in certain zones and not all straight.
You can cut terrain up by blending one with another, especially if it's stretched out on an axis. The next phase would be to experiment with the voronoi technique but on more complex terrain. I've done this with a heavily stepped terrain which is obviously ideal for it.
I think you can almost do anything with TG2 except make erosions which are quite flat like rivers etc. We have the Alpine Fractal which goes a bit in this direction but I've noticed there is a nasty problem with that shader. You get horrible vertical breakages.
Forget that I mentioned voronoi "subtraction" which I've talked about a few times. That creates some problems. The negative displacement for the voronoi in the file provided earlier is the way to go.
I've got a new terrain every 5 minutes here. I'm still only using voronoi. Here is a screenshot that will hint at how I did the terrain in the picture and why it works with this technique used in this thread plus the voronoi on the slopes. It's based on the second file posted here. Terrain displacement has a low spike limit setting - I usually keep that low to avoid extreme spikes. The chain of voronoi are obviously different sizes and different displacement values. This is a dead easy way to create interesting terrains. Took me one minute to create this terrain. Where the bias is, try other maths functions. Try a smooth step etc etc.
@efflux. I dont if it would be faster to delay the conversion from vector to displacement by replacing upper displacement functions by "add scalar".
Mogn.
That's probably true but the displacement nodes are a way of giving each voronoi a different displacement value.
To be honest, this all renders reasonably fast anyway. My other stepped terrain on the other hand isn't too fast.
OK. I've rebuilt that graph to do essentially the same thing but without the extra displacement nodes. It obviously requires a lot more nodes and is not exactly the same results but is certainly more or less the same effect. I'll do a test to see the render differences. It could be interesting.
I've run the test. I built blue nodes to do as close as possible to the version with the red node displacements. Blue nodes rendered in 8.57. Red nodes in 9.17. So it's not earth shattering but does make a small difference.
Here's a file of the chained voronoi. Quite simple but I'll put a file here anyway. It's the exact setup in the last render from the stepped terrain thread. Nothing tweaked for artistic effect.
The voronoi are just added unlike the last screenshot which shows displacement nodes. You could multiply before the add.
This really needs some serious experimentation. Different noise functions and different maths effects applied to them.
Thanks for sharing this efflux!
Quote from: efflux on March 12, 2012, 11:19:10 PM
I think you can almost do anything with TG2 except make erosions which are quite flat like rivers etc. We have the Alpine Fractal which goes a bit in this direction but I've noticed there is a nasty problem with that shader. You get horrible vertical breakages.
There is a fixed version of the Alpine Fractal Shader in the next release (2.4). You'll need to use Alpine Fractal Shader v2.
Matt
Nice and clean programming. The use of a "Get position" for each of the of the Voronois is free of charge.
A curve tool (a 0..1 soft profile with 3 settable points) is soon coming.
Thanks Matt. I never used the Alpine Shader much until recently but I kept seeing those breaks.
Here's that last file slightly tweaked. Colour added and terrain slightly redirected. Global ocean added. More pronounced beach areas could be created with this technique. There is still no surface texture other than one flat colour. Image is exported as exr then edited in Blender and Lightzone.
Thank you efflux, this is a marvelous piece of work.
Thanks Choronr.
I thought I would add that last picture for better example of the idea here. Most of the files I've worked on lately have been kept very simple and grayscale to concentrate on getting simple basic forms to expand on. Especially before any fancy surfaces go on.
For anyone wondering how to create global ocean (I'm assuming it's elsewhere on the forum but since I did it here), it's just a sphere the same size and place as the original or very near. You'll have to slightly tweak it's size to fit. It's just got a water shader applied.
EDIT: That global water method is wrong. A second planet casts shadows. I've changed it to sphere which can have cast shadows disabled.
The last image is unique and well done :)
You ain't seen anything yet ;)
Hi efflux,
Have worked with your Tab7 file and came up with this. Added some additional nodes, surfaces and vegetation. The additional nodes included files by FrankB (wavey strata and terraces) and dandelO (1 and 10m cracks).
Bob
Hi Choronr.
That's great to see because it's obviously planetery scale. You colours are nice and cool to see the trees. I've never got into trees so have no idea what problems that poses. Probably the set up in my file is a bit overall the same charactor in several ways but that's where all the time spent tweaking things around comes in. I'm trying to make these files demo just the ideas so other people can add the details.
I meant to do more TG2 in the last few weeks but I'm trying to do too many things. I need to work on concept stuff in 2D because I have a ton of this type of thing and 2D drawing is fast. I have hugely overambitous plans for animations (including soundtracks) but that requires me to concentrate purely on the concepts or I'll never get anywhere. I actually have a first class degree in Fine Art so I can paint and draw. I have a Cintiq 21UX here that largely gathers dust. To be honest, much as I love the idea of building 3D landscapes, it's not the best use of my skills and I do find it a bit frustrating because the results I want and what TG2 can do aren't always the same thing.
So odd stuff I add on this forum will be sporadic.
Thanks efflux for your reply. I am not privy for the use of the blue nodes; and, certainly appreciate what you have and others have shared here. My use of Terragen started almost ten years ago with 'Classic' where creations appeared as somewhat ethereal but beautiful ...had loads of satisfactory images - but, were lacking in having the fine details of what TG2 eventually offered.
I'm looking forward to continued learning; and, will be trying to incorporate you setup in future projects ...thank you again.
Bob
.
Just trying to kick start this again. This is an incredibly easy way to create interest in terrains and this is just a starting point to mash the terrains up in other ways. I've shown a picture of the nodes. One fractal has displacement roughness of 1, the other 0. Those values are just reversed to get the different results in each picture. The stones are the voronoi.
Here's a further variant on the previous terrain. Smooth low terrain and rough high terrain using the Tex coords from XYZ to altitude blend. The new lowest terrain is another smaller version of the bigger terrain but blended with a merge shader set to Highest raise. Since this terrain is less displaced it is below the altitude blend of the other two. Now we have a much bigger sense of scale. More like the actual scale because it looks like large hills with smooth deposition of rocks or stones or whatever rolling down then hitting the smaller terrain. The lower POV shot shows how this creates interest at these low level views. Ideally you'd put rocks on these gentle slopes to enhance the effect. This effect works simply because we know what real hills look like and these features give us the sense of scale.
The lower POV render also has some clouds as higher atmo. Haze is low and not so glowy. That's why the higher haze effect varies. I'm still experimenting with this and it needs further tweaking.
I'm not supplying the finished terrain in this case. Try tweaking the settings from previous files. My advice is to get a nice shape with one fractal before doing anything. Change the shape while the roughness is 0. That way you will see the general shape better before adding roughness. Start with quite simple shapes to see the forms and scales then add detail with higher roughness. Various results from tweaking the fractal settings is another thread though. I'm concentrating on why altitude blended terrains are good. The last few renders are just developments of how it goes when you get nicer terrain shapes and much better than this is possible.