mario837
Iterator
Posts: 158
mario837
|
|
« on: February 14, 2011, 05:42:54 PM » |
|
Is it possible to use textures in addition to colors and lights? I think some of the works may improve substantially should we be able to use a determined skin.
|
|
« Last Edit: February 14, 2011, 05:46:26 PM by mario837 »
|
Logged
|
|
|
|
Jesse
Download Section
Fractal Schemer
Posts: 1013
|
|
« Reply #1 on: February 15, 2011, 10:05:51 PM » |
|
Common texturing can't be done because there is no meshlike structure the texture can be connected to (at least i have no clue how to do it). What can be done is using a light worldmap in combination with the normals, but this is not really the same, of course.
|
|
|
Logged
|
|
|
|
|
mario837
Iterator
Posts: 158
mario837
|
|
« Reply #3 on: February 16, 2011, 03:16:25 PM » |
|
there is no meshlike structure Thank you for the response Jessse. I understand the lack of a mesh to control 100% the texture. However, a feature like the TX in Fractal Explorer can help to fill some black holes or even replace the use of an image as background. Bring Back Bitmap Lighting!
I will try as soon as I find what you are doing
|
|
|
Logged
|
|
|
|
lenord
Fractal Bachius
Posts: 611
No Matter where you go there you are
|
|
« Reply #4 on: February 16, 2011, 05:16:29 PM » |
|
Bring Back Bitmap Lighting!
I was just learning how to make it do things like this when it went away:
Guess I hadn't jumped on the MB3d Bandwagon yet when that was an option, was that like Incendia's Bitmap for surface material? I have to admit it was one of the few things I did like about Incendia, along with the save Render Buffer.
|
|
|
Logged
|
|
|
|
fractalrebel
|
|
« Reply #5 on: February 16, 2011, 10:50:54 PM » |
|
My 3D Fractal Raytrace progam in Ultrafractal can add texture to the Mandelbulb. Attached is an example of a Mandelbulb zoom. The upper portion is the plain Mandelbulb, while the lower portion has texture added.
|
|
|
Logged
|
|
|
|
David Makin
|
|
« Reply #6 on: February 17, 2011, 12:51:51 AM » |
|
You can add texture maps (or even bump maps) based on any colouring you like that can produce two different values - for instance allow the user to choose 2 of the 3 (or more) values from orbit trap distances as the look-up coords in a map (to a given modulus).
|
|
|
Logged
|
|
|
|
DarkBeam
Global Moderator
Fractal Senior
Posts: 2512
Fragments of the fractal -like the tip of it
|
|
« Reply #7 on: February 17, 2011, 12:02:23 PM » |
|
My 3D Fractal Raytrace progam in Ultrafractal can add texture to the Mandelbulb. Attached is an example of a Mandelbulb zoom. The upper portion is the plain Mandelbulb, while the lower portion has texture added.
I think that this involves in more fuzziness in most cases...
|
|
|
Logged
|
No sweat, guardian of wisdom!
|
|
|
Jesse
Download Section
Fractal Schemer
Posts: 1013
|
|
« Reply #8 on: February 18, 2011, 10:30:38 AM » |
|
My 3D Fractal Raytrace progam in Ultrafractal can add texture to the Mandelbulb. Attached is an example of a Mandelbulb zoom. The upper portion is the plain Mandelbulb, while the lower portion has texture added.
Quite good, my guess is that there is no general solution that works in all cases. For example could you do a julia seed(0,0,0) sphere and bind the texture to it? Would be interesting to see the whole sphere. You can add texture maps (or even bump maps) based on any colouring you like that can produce two different values - for instance allow the user to choose 2 of the 3 (or more) values from orbit trap distances as the look-up coords in a map (to a given modulus).
Yes, the question is to find two values that gives the desired effect. Meaning that they are more or less orthogonal and have the same amount of gradient so the texture will not be stretched to much. My first guess is that such parameters does not exists for a general solution because you can choose always certain directions of a plane (the texture should be attached) so that one direction has no gradient? (but not sure about it) Animation would be the next issue, you might have to do two overlayed textures in different sizes, to allow zooming without nasty effects.
|
|
|
Logged
|
|
|
|
David Makin
|
|
« Reply #9 on: February 19, 2011, 01:34:00 AM » |
|
Using orbit traps of many types produces useful results in 2D even with the stretching/skewing inv olved, I don't really see any reason why some options wouldn't get useful results in 3D too. Specifically with the appropriate paramers on an orbit trap it shoud be possible to reduce the affect of "whipped cream" areas without making areas with true fractal detail too busy e.g. such that the mapping is stretched/larger scale in the fractally areas and the mapping is better/smaller scale in whipped cream areas.
When I get chance to use UF again I'll experiment since it wil be pretty easy to write a formula/colouring to allow the mapping of a source image onto the surface of a 3D+ fractal using any available colouring formula/s from the UF database to get mapping coordinates.
Actually, maybe Ron's formulas let you do that already, I'm not sure how sophisticated he's made the mapping as I haven't tried the latest version.
|
|
|
Logged
|
|
|
|
fractalrebel
|
|
« Reply #10 on: February 19, 2011, 02:16:24 AM » |
|
My changes to 3D Fractal Raytrace are still WIP, but the following features have been added:
1. A 2D or 3D transform which can be applied to any number of iterations of the fractal, ranging from 1 to all iterations and starting with iter = 0 to any chosen iteration point. 2. The transforms are plugins so there are several hundred available in the UF library database. Mark Townsend has a special plugin which allows any Trap Shape to be used as a transform.
These changes aren't public yet, as I am still working on them.
There are several options that have been in the program for a while which are on the coloring side (3D Fractal Coloring Direct). There is a texture option which takes Trap Shape plugins. There are 5 different flavors for mixing the texture with the z value passed to the coloring formula. With the correct choice you can selectively add texture to different regions. some of the trap shape textures have convolution options so bump map style effects can be obtained. There is also a coloring option in the coloring formula which allows the mapping of images and colored textures. All of this takes advantage of plugins already available in the UF library database.
|
|
|
Logged
|
|
|
|
Jesse
Download Section
Fractal Schemer
Posts: 1013
|
|
« Reply #11 on: February 19, 2011, 09:34:26 PM » |
|
Hi David! Using orbit traps of many types produces useful results in 2D even with the stretching/skewing inv olved, I don't really see any reason why some options wouldn't get useful results in 3D too.
There are surely many combinations of parameters that give good results in many situations, i don't want to neglect this. But my choice of implementing this is also dependent on other things: - more than the currently recorded params (only a veclength orbit trap and the smoothit what is surely not enough) would needs a new calculation and extensions in all iteration functions, what is more a time effort problem. - not working in animations, ..maybe more user questions popping up. I think texturing is more a feature for blocky meshes to make them more interesting (or maybe i am getting a bit lazy). Isn't it the positive thing with fractals that they have mostly their own unique textures? If not, you can often produce some by choosing after n iterations another formula.
|
|
|
Logged
|
|
|
|
David Makin
|
|
« Reply #12 on: February 20, 2011, 01:53:33 AM » |
|
I confess it's been so long since I programmed "raw" code for fractals rather than using UF that I'm probably underestimating the implimentation time. Having said that of course, increasing the possible calculated values with respect to mapping also alternatively gives two or more new values for plain colouring the surface so adding such code is a big plus Come to think of it - adding an extra iteration or two to the surface pixels orbits after bailout using a different formula could be used for colouring purposes too (or, if you know the bailout iteration count for the pixel concerned prior to performing the iteration for colouring purposes then in the iteration to get colouring/mapping simply replace the formula used at some point in the orbit as in your suggrestion with respect to mapping).
|
|
« Last Edit: February 20, 2011, 02:00:57 AM by David Makin »
|
Logged
|
|
|
|
Jesse
Download Section
Fractal Schemer
Posts: 1013
|
|
« Reply #13 on: February 20, 2011, 09:56:27 PM » |
|
By now i have 8 iteration loop functions and 5 of them coded also in assembler with sse2, a changing here is always a bit work. I did no shader programming yet, maybe you can answer me this: Is it possible in shader programming to call functions only by pointers, like variables? Like: function1 = mbulb8; function2 = mbox; call function1; call function2; Then it might be possible to transfer my code to gpu sometimes, if double precision is common maybe. ps: i already did, more by accident, a kind of bumbmapping: if you choose the raymarching stop decision on DE's and calculate the normals on smoothits, you might get more structures or less structures then surface details... (it is just another possibility out of many)
|
|
|
Logged
|
|
|
|
xenodreambuie
|
|
« Reply #14 on: February 21, 2011, 12:41:39 AM » |
|
I think texturing is more a feature for blocky meshes to make them more interesting (or maybe i am getting a bit lazy). Isn't it the positive thing with fractals that they have mostly their own unique textures? If not, you can often produce some by choosing after n iterations another formula.
I agree. Fractal detail only needs coloring and/or lighting for best effect. Textures are most useful on smooth areas, such as orbit traps, or base shapes (Incendia), constructors (XenoDream), raymarching with low res, etc. When these areas get iterated down to fine detail, the textures should be filtered to minimize aliasing.
|
|
|
Logged
|
|
|
|
|