Page 26 of 48
Re: Phantom Scatter | Procedural instancing technology [1.6]
Posted: Wed Mar 11, 2015 1:10 pm
by smicha
Phantom107 wrote:As stated, I don't have a solution for this... the Octane Render cloud material is full 3D and so the bottom and top layers essentially cut it off.
Maybe someone has an idea on how to fix this? Maybe with a special turbulence generator?
Can you mirror/duplicate a cloud horizontally?
Re: Phantom Scatter | Procedural instancing technology [1.6]
Posted: Wed Mar 11, 2015 1:19 pm
by Phantom107
Download the *.orbx scene file linked in my last post and see for yourself. The problem has to do with the Octane Render material.

Re: Phantom Scatter | Procedural instancing technology [1.6]
Posted: Wed Mar 11, 2015 1:23 pm
by glimpse
Phantom107 wrote:Download the *.orbx scene file linked in my last post and see for yourself. The problem has to do with the Octane Render material.

changing "Final Blend" seems to make a change - 1.5 - 1.7 seems to give the most realistic result (cooking a piece =) I'll post after..a day or two.. =DDD..
Re: Phantom Scatter | Procedural instancing technology [1.6]
Posted: Wed Mar 11, 2015 2:18 pm
by linvanchene
Update / Edit:
Parts removed by user because a solution has been found.
Re: Phantom Scatter | Procedural instancing technology [1.6]
Posted: Wed Mar 11, 2015 2:33 pm
by Phantom107
Maybe you can elaborate more on the changes to geometry to solve this? I don't entirely get what you mean, so it will help if there is something I can look at.
I have been thinking about doing other forms of geometry but I keep running into the same problem: the cloud turbulence is fully procedural and will not match up with the geometry I generate... I made a little image in GIMP to illustrate what will happen:
Looks like it will be cut off regardless.
Re: Phantom Scatter | Procedural instancing technology [1.6]
Posted: Wed Mar 11, 2015 3:15 pm
by linvanchene
Update / Edit:
Removed parts since an alternative solution has been found.
Phantom107 wrote:
Looks like it will be cut off regardless.
There will be cut off with volumetric materials but the idea was to use so many different stacked volumes that the cut off will not anymore be that noticeable from a distance.
To illustrate some images how it is done in e-on vue:

- stacked spherical clouds

- stacked spherical clouds - rendered

- Metaclouds front - cloud layers background
- - -
In the example above:
For each e-on vue spherical based metacloud we would need about 25-100 volumetric based stacked spheres in OctaneRender...
And as profbetis allready pointed out in his original thread more than 25 layers is going to slow down render speed massively.
The problem in practice may be that most systems will not be able to handle it.
- - -
So for now I can live with the flat cloud bottoms.
It still looks amazing and I am looking forward to the next Phantom Scatter update!
Thank you for your work!!

Re: Phantom Scatter | Procedural instancing technology [1.6]
Posted: Wed Mar 11, 2015 3:53 pm
by Phantom107
I think I found a solution to the problem. I was just working on a GLSL shader for my videogame and it gave me an idea... I'm letting it cook as we speak.
I can already give away that this makes it a lot more demanding... stay tuned.
Re: Phantom Scatter | Procedural instancing technology [1.6]
Posted: Wed Mar 11, 2015 4:10 pm
by Phantom107
Alright so here's what I came up with:
I think it's a lot better! Instead of the 'final blend' node, I am now multiplying the turbulence texture by another turbulence texture, but the latter is at a much smaller scale. So what happens now is that even the big cloud layers have another turbulence messing with them. This prevents them from getting very solid, which was causing the 'flat cloud bottom' problem.
It is still not 100% solved, but I think it does make a hell of a difference. It could theoretically be solved entirely if I add some more turbulence textures to randomize further... but it is already very demanding so a tradeoff has to be made.
Re: Phantom Scatter | Procedural instancing technology [1.6]
Posted: Wed Mar 11, 2015 4:14 pm
by smicha
Yes, much better. If we could have some small variation in terms of cloud height level above ground - this would reduce 'flatness' feeling.
Re: Phantom Scatter | Procedural instancing technology [1.6]
Posted: Wed Mar 11, 2015 4:18 pm
by p3taoctane
Very cool
A huge improvement.
It does look like a Kansas Prairie cloud layer... but not an ocean off shore cloud layer
I am wondering of you stack different sets of clouds using placement it would give different layers.
A lot of GPU overhead but def a nice effect
Keep up the research man... really cool
Peter