New Rocks

Started by efflux, December 06, 2012, 07:10:45 PM

Previous topic - Next topic

efflux

OK Chinaski. That's great. I edited my last mail because that was before I looked at exactly what you mean. I've tried the smooth step thing before the profile. I think this solves our blending issues with the ground but I'll need to test further.

efflux

#31
Quote from: mhaze on December 09, 2012, 01:57:29 PM
Nice work Chinaski.  Better distribution in this one, added another perlin noise

mhaze. This is much improved. This is all we have to do - make more complex blending shapes. That of course could be done all sorts of ways. You could distort it as well. As I've mentioned, there is also a way to make an altitude blend for the rocks independent from the surface layer blending. What you can do is distort the altitude blending region with the actual shape of the rocks. What this does is make the altitude curl around the rocks so that they do not fade out. This is a little difficult to explain and the graph starts to get a little complex but I have a thread called Altitude Blend distorter. There are various ways to make the graphs but if you look at the pictures you'll get the idea. Take the rocks here - they would come in and out at altitudes but not fade into some other surface if you see what I mean. It's one step at a time though. Best not to post messy graphs. I have quite a lot of half baked graphs with stuff in them that I need to bring out and attach to other methods and simpler set ups. Chinaski has provided a fix for a previous file I have here. I'll post the new one soon and remove the old.

efflux

#32
Quote from: mogn on December 11, 2012, 04:28:36 AM
"EDIT: About "radial gradient network, built from a veronoi grid" I was thinking something like that:"

Like this:

mogn, the shapes that graph creates are really cool. The only thing is that is creates nasty shadow problems. This always happens when you shift around position too much in TG2.

efflux

#33
Here's the last file I posted with a better arrangement. My last graph wasn't ideal. I've added Chinaski's improvement. My graph was working sloppily after the divide. You will notice the render looks different from before but the rest of the settings are the same. This is just that the settings are now working over a different range. The old render had no faults with the settings in that version and in fact it might look a bit better with the particular settings it had but I didn't want to tweak the new file to have different settings. This version has more headroom for tweaking without getting problems.

I think we now have a totally satisfactory method of blending in the rock shapes but the surface displacements are another matter. They will still create problems and that needs some thought. Also, exactly how we are going to blend the overlying surfaces? This is a question of what look we want as well as the technicalities.

Probably these threads should be arranged differently so that the best files are put on the first post otherwise people might download more problematic files. I didn't really start this thread with that properly in mind though. I don't want to change all the files I've put here because some of the discussion relates to how they can be improved.

Chinaski

Guys, I think I found a big part of the solution. Max and min values are normalised. Now I need to mix that smoothly, and exclude the small cells. Thanks to mogn, your clip helped me a lot.
You don't understand me ? That's normal, I don't speak english.

mhaze

Hi

I've learnt a lot from this thread unfortunately I'm too busy(decorating the bathroom before family arrive for Christmas) to play, I should be able to get busy after Christmas and there are lots of ideas to explore here.

efflux

#36
There are multitudes of ways to do things here. Do you want to rocks to smoothly transition from the ground surface? Do you want them to pick up all the displacement from ground or not? What profile do you want them to have? etc etc.

I've added another variant. In this one the rocks pick up displacement from the ground but the seam is a sudden transition for the rocks colour but not their displacement otherwise the displacement would be broken. I also hooked in a smooth step which adds to the step end. The actual extra constant scalar is not needed but adding values from 0-1 is around what you'd want which is what the slider does. Other things could be experimented with here. This can squash the rocks flat. It just goes on forever and there are still ways to technically tidy up these graphs. This whole angle deals with so many things but that's why I'm interested in it.

I have other files that play in this area but they need tidying. I'm a bit sick of this though. There are other angles I need to explore. Every area cross feeds into another though.

Xynedia

.
#37
.

inkydigit

This thread seriously rocks!!! Awesome stuff to explore... Thanks!
:)

RichTwo

Always pushing the envelope, aren't you?  This is still scratching scratching TG2's surface, and I see what you've pointed out.  Keep on keeping on!
They're all wasted!

mhaze

Modified tgd attached. much more random distribution and merging two graphs to give two rock sizes

efflux

#41
Quote from: Xynedia on December 12, 2012, 05:12:41 PM
Hi Efflux,

Very interesting work, and the files with it.
Unfortunately, I can no longer do in TG2.
TG2 have the software, digital key and password access
for on Planetside completely erased.
TG2 is for me too complex and not user-friendly in the
World creation and render preview practice.
I would have made ​​her file or a World Interesting
can do with these stones.

In German:

Hallo Efflux,

Sehr Interessante Arbeit und die Files mit dazu.
Leider kann ich nichts mehr in TG2 machen.
Habe die TG2 Software, Digitalschlüssel und Passwort Zugang
für auf Planetside komplett gelöscht.
TG2 ist für mich zu komplex und nicht Anwenderfreundlich in der
Weltenerstellung und Rendervorschaupraxis.
Ich hätte aus ihrer File noch eine Interessante Welt
mit diesen Steinen machen können.

Hi Xynedia.

TG2 is complex and I understand your frustration coming from Mojoworld. I have many gripes about TG2 as compared to Mojoworld but Mojoworld does have other problems. The UI is bad and unfortunately it's out of date. No multi core for example so to continue procedural landscaping we now have have TG2. I'm hoping TG2 gets a few changes eventually to open up more power as in Mojoworld.

Chinaski and mogn, you should make more threads on tweaking the noise functions to get different shapes because that goes across the board for creating any surface whether displaced or just colour - colour is more versatile because you don't even have to worry as much if the shapes have sudden changes. You know how to do these things.

As for that last file. The displacement and colour are separated from the fractal output so they get treated differently. I just used a colour mix for colour but bear in mind the merge shader is one way to separate colour from displacement. One can control colour and another displacement. There are probably a whole host of uses for this. I'd rather have more blue nodes controlling stuff in some of these graphs but they work OK.

That last file also strays a bit close to fake stones. It's questionable whether you'd use that rather than fake stones.

mhaze, I'm a computer that doesn't have TG2 so can't look at that tgd but in regards to creating different rock sizes. Here's my idea. Make different sized voronoi noises and blend them together. You could add or subtract constant values from different rocks to get them to poke through the clamping or not.

efflux

Just another thought about the multiple sized rocks and clamping. I'm talking without testing here but it might be a good idea to just set the clamping and have each voronoi rock have it's own control to poke through the clamping or not. This might be more intuitive.

efflux

#43
Different angle here. I've gone for lower type rocks. No overhangs except what the surface displacement creates. These are voronoi piled on other voronoi getting smaller on the top layer. The surfaces can be blended from one to another. I like this one because of the inconsistency depending on the rock slope at the edge which changes at lot. Playing with the rock density and the blend region can get a lot of different results. There has to be a decent space between blend start and end values or bad displacement joins will occur.

This is file 21 but that doesn't mean I've done lots in between. It just came from a different folder with other files.

efflux

Just another point about the shadow problems. I noticed this when I tried to use shadows on mogn's file but it's not just that file, although it seemed worse. The more blue node vectors and scalars added the more the shadows problem occurs. I'm not sure what the technical problem is.