Welcome to Fractal Forums

Hi all, i am testing out some heightmap rendeing methods for mutatorkammer, look how far i came this weekend ;;

detailed gallery here:

http://fractalmovies.com/index.php?option=com_gallery2&Itemid=3&g2_itemId=841&g2_highlightId=859

a valley
(http://www.fractalmovies.com/gallery2/main.php?g2_view=core.DownloadItem&g2_itemId=866)


.. the sharp edges seem to make problems ;(
(http://www.fractalmovies.com/gallery2/main.php?g2_view=core.DownloadItem&g2_itemId=859)

neato, how about some fractals on a sphere next? ;) it would be so neat to spin around and zoom in!

Impressive, even with the jagged edges.  I like the "3d_3d" one the best: 
http://fractalmovies.com/index.php?option=com_gallery2&Itemid=3&g2_itemId=843&g2_GALLERYSID=7dbd5cd6db1addf673dc994391e56747

Go Kleinhuis!

Christian Kleinhuis (Trifox) wrote:
>
>    Hi all, i am testing out some heightmap rendeing methods for Mutatorkammer,
>    look how far i came this weekend ....  ... the sharp edges seem to make problems  ;(

Very impressive!!    8) 

Several of these images have aspects about them which I like, but the "map_1_3d" and the "3d_3d" images are the ones I like the best.

Can't wait to try this out myself!!    :)

i finally opened the box of pandorra  :)


this is going to be the layer mode of the program, each layer is a entity in 3d space, with a particular mapping, the idea is to have standard photoshop/ultrafractal layer mode when looking right from above on to a stack plane ( or sphere( damn it, map makers dilemma, how to map a plane on to a sphere :( ) or displacement, or torus 8) )transformed formulas
so you can define water sky planets and stuff ;) but this will take a while :(  any help is welcome ;)
but..

as you can see the jagged edges fluff me totally up, you(lycium) said the paper describes correct rendering of fractals structures?
at this time it is a sampling problem, the regular grid does not hit the edges well enough, i will try out hexagons next time , or some sort
of 3d simplyfing algorithm to detect planes and edges

lol, at this point ( mutatorkammer v0.648 ) is somehow *frozen*, the 3d parts will be together in the final name of the program: PHATFRACTAL2  ;D
i will see that i get some of the functions running in user interface and save&edit mode, but the point is leaving good old 2d grounds is sooooooooooo much work,
llight pos, attenuation, camera, target camera uargh- but this would be all, and some nice general mapping formula to map the formula to other object rather than a plane, it would be nice to use existing parts of the engine for doing that as you know the engine features vectors of complex numbers - so each vector formula should do the job ;)


btw. i have added some more renderings to the gallery
http://fractalmovies.com/index.php?option=com_gallery2&Itemid=3&g2_itemId=841

noisy

(http://www.fractalmovies.com/gallery2/main.php?g2_view=core.DownloadItem&g2_itemId=880) (http://fractalmovies.com/index.php?option=com_gallery2&Itemid=3&g2_itemId=878&g2_imageViewsIndex=1)

superformula
(http://www.fractalmovies.com/gallery2/main.php?g2_view=core.DownloadItem&g2_itemId=883) (http://fractalmovies.com/index.php?option=com_gallery2&Itemid=3&g2_itemId=881&g2_imageViewsIndex=1)

pardon, which paper? i previously linked hart's thesis somewhere on this forum (eh losing track with all my recent spamming), are you referring to that?

argh 6:50am, probably time to stop coding...

nice work...your program is looking good...by the way I like the original name better (sounds more esoteric  ;D) it's not really important I just had to get that out
You're really moving along on the 3-d work looks great!

Marc

and some more:

http://fractalmovies.com/index.php?option=com_gallery2&Itemid=3&g2_itemId=884&g2_imageViewsIndex=1
(http://www.fractalmovies.com/gallery2/main.php?g2_view=core.DownloadItem&g2_itemId=886)


http://fractalmovies.com/index.php?option=com_gallery2&Itemid=3&g2_itemId=887&g2_imageViewsIndex=1
 (http://www.fractalmovies.com/gallery2/main.php?g2_view=core.DownloadItem&g2_itemId=889)

Ok, i see, it all lies in smoothing it.


These are regular smoothings on a 1024 grid, right now i am working on a gaussian smoothing ;)

http://fractalmovies.com/index.php?option=com_gallery2&Itemid=3&g2_itemId=893&g2_imageViewsIndex=1
(http://www.fractalmovies.com/gallery2/main.php?g2_view=core.DownloadItem&g2_itemId=895)

http://fractalmovies.com/index.php?option=com_gallery2&Itemid=3&g2_itemId=896&g2_imageViewsIndex=1
(http://www.fractalmovies.com/gallery2/main.php?g2_view=core.DownloadItem&g2_itemId=898)

i'd be interested to know what fps you get for these 1m triangles :)

as for gaussian smoothing, i can pass on a tip: just apply a box filter several times, the process converges to a gaussian result quite quickly (even photoshop uses this technique). what's more, if you do it right* a box filter can be implemented in O(pixels), that is, the speed is completely independent of the size of the filter width. and one more thing: probably you've noticed the boundary-handling if-statements are annoying, but in this case we can do it well. for each successive box-filter the image grows by filterwidth/2 on all sides, so for N applications it's trivial to predict the final image extents... so if you work with a slightly enlarged buffer you can skip all those special cases.

hmm, i'm not so sure how useful the above is to you though, since you're hardly in a position to call the resulting heightfield "fractal" if you smooth it a lot ;)

edit: fine, fine, i will drop the piece of info you need most: what you really want to achieve is not postfiltering, but prefiltering. this is true of all your mutatorthingy fractals, be they 2d or 3d...

* this is where it's convenient to have a demoscene background :D

lets not talk about it, it is rendering a 1024x1024 image, and then render 1024x1024x2 triangles ;)

if its just the triangles, you could move, but no refinment of the formula ...


i am right now using a filter to which i can pass a matrix(which sums to 1 ) , wich is then applied to each pixel, the size of the matrix/2 is the border i need to calculate if i want to render more than one region

the filter used for the above images is 20x20 pixels wide ( a 3x3 filter applied 10 times).

just think of high resolutions, if i render it 10.000x10.000 the filter just gets me rid of the nasty spikes a fractal can cause, filtering them away just makes a sharp 3d image, even for zooming ;)


yeah, thats the thing i stated above :D
btw. supersampling is already implemented ;)

i am enjoying demo secene since 1990 ;) even if i have not released anything meaningful, i started releasing c64 stuff lastyear ;)
check this out:
http://noname.c64.org/csdb/release/?id=38303
 

i think that explicitly storing and filtering all that is misdirection of computational resources (rather supersample/prefilter), but well, if it works for you :)


yes exactly; if you render 10k by 10k you should be able to use each and every one of those texels, otherwise it's a waste, no?


that's too cool mate, respect for oldskoolness! :D i checked your juggling demo (another hobby of mine is juggling), neat stuff! i only got interested in the scene during mid-90s unfortunately, but have since caught up much of the history... of course that's nothing like experiencing it directly (as you lucky europeans get to do), but oh well.

edit: for the benefit of those who don't know what we're talking about, visit http://pouet.net and be amazed!

here is another one:
http://www.deviantart.com/deviation/50901318/

(http://tn3-2.deviantart.com/fs13/300W/f/2007/074/1/d/Barnsley_3d_by_trifox77.jpg)