The result of my third full JWilfire-Povray run. I'm considering implementing an algorithm that ditches outlier plots in order to clean the output. However, the output still looks... scatterbrained for larger fractals. The fractal in this image is small and easier to control the output for. What I would like to do is make the surfaces smooth using fewer points, but that's no walk in the park.
:origin()/pre00/cf30/th/pre/i/2015/091/7/4/experment_004___artificial_gamma_effect_by_ablipintime-d8o1t70.png)
:origin()/pre00/b6ea/th/pre/i/2015/089/9/a/isolated_in_the_canyons_by_ablipintime-d8nt47b.png)
:origin()/pre00/153d/th/pre/i/2015/089/9/6/snow_blue_recluse_by_ablipintime-d8nt3j6.png)
:origin()/pre00/1002/th/pre/f/2015/089/9/d/blue_recluse_by_ablipintime-d8nszqc.png)
:origin()/pre00/3953/th/pre/f/2015/089/6/f/light_of_another_world_by_ablipintime-d8nsyin.png)
:origin()/pre00/c64e/th/pre/f/2015/089/d/0/fractor3_experiment002_by_ablipintime-d8nsv17.png)
:origin()/pre00/cee1/th/pre/i/2015/089/6/f/experiment001_by_ablipintime-d8nsrns.png)
:origin()/pre00/fc17/th/pre/f/2015/081/1/1/scripted_elliptic_11_by_ablipintime-d8msd0q.png)
:origin()/pre00/2639/th/pre/f/2015/017/7/c/knot_underestimated_by_ablipintime-d8ecg7u.png)
:origin()/pre00/0e68/th/pre/f/2013/204/5/0/cute_flowers_and_bees_png_by_hanabell1-d6erd1y.png)
:origin()/pre00/4c91/th/pre/f/2008/328/e/6/kyoto_flowers_by_mogcaiz.jpg)
:origin()/pre00/c980/th/pre/i/2013/222/8/1/low_poly_trees_by_error_23-d6hii0k.png)
:origin()/pre00/d3e8/th/pre/f/2014/188/0/a/realistic_dead_fir_trees__fh__by_dirty_couch-d7pno72.png)
:origin()/pre00/5fb2/th/pre/i/2015/106/5/6/free_to_download_tree_by_wulfthewolf-d7q1f2z.png)
:origin()/pre00/496f/th/pre/f/2013/204/9/5/cute_stylish_abstract_flowers_png_by_hanabell1-d6erhi4.png)
:origin()/pre00/b4a4/th/pre/f/2014/039/a/5/3rdfullpov_by_ablipintime-d75mk3r.png)
That's really cool. I've been working with JWildfire & 3D meshes although I took the easy route and basically converted the 2D image to 3D mesh using image stacking. Have you considered utilising MeshLab to clean the mesh before passing to PovRay if at all possible?
I did consider outputting a mesh (e.g. wavefront obj) by simply building it with points, but that results in color loss (for a correct surface) or a messed up surface (for no color loss). The problem there is that what looks like one color for a region of the fractal is actually composed of several colors and maybe one or two dominant ones.
I elected to go with Povray because primarily because of the ray tracing, but I also get the benefit of not having the problems as I just mentioned. However, it would be cool - and much more useful - to create meshes. I may look into that next, assuming I finish this project. One thing at a time.
Image stacking works - they do it with MD3D for voxels and the results can be pretty good.
Are you currently working on a 3D model output for JWF?
The routine I use kind of happened by mistake and was never really my main aim - was while playing with MD3D voxel output I decided to try it in Apo. I wanted to merge my previous Apo scripts anyway so built in the option to create a stack of images at different heights just to try it and got some suprising results. Trying it in JWildfire was the next logical step. I'd like to make my routine more 3D based as opposed to simple height adjusting but that wouldn't be in the near future.
I've noticed the color region problem too whilst messing around, for example what looks like black / transparent is a near-black shade if that makes sense. Fiji then plots this as a point in 3D space instead of ignoring it - I could edit the stacks (if I could see the offending pixels), use photoshop to clean the 'noise', use a 2-color pallete or use MeshLab and/or Mesh editing software such as Blender to remove the offending plot points.
jwildfire.org/forum/viewtopic.…
And then Maschke (thargor6) told me he's making changes that may conflict with what I've done. The good news is that JWF 2.0 is coming soon.
But in the mean time, I wouldn't mind tossing around ideas for how we can tackle the 3D challenge. What do you think?
Forgot to mention - have you seen PoseRay? Apparently allows conversion to/from PovRay to/from obj mesh.
Has other features including ability to modify material & lighting.
Now if only I had some compatible .pov files - says that it can't find vertices etc on the files I've tried.
May try it from obj to pov see if I get any joy that way......
I'm guessing it only imports and exports a certain shape of povray (I'm thinking "blob", but I'm not sure), just to keep the mesh conversion simple. That being the case, trying to open a file with other primitives may cause it to crash. If that is the case, it's poor exception handling on the part of the programmer, but who am I to complain? - it's free software.
Well had a play with PoseRay and came up with this - fav.me/d76u8j7
Its not to bad as a port between a 3D mesh and PovRay although I did initially get stuck with using PovRay's materials - I imported them but they didn't appear. Checked the FAQ which didn't help much. Thought it was where I had stored the program as Win7 does some strange things with permissions, went to re-install it but checked the folder first. The materials were there so I restarted PoserRay and they suddenly appeared. Ah well, live & learn I guess - a simple refresh was obviously hard to code
I've downloaded your exporter code (I </3 MediaFire as I have downloaded iLivid by mistake so many times).
Hope to get to try it but don't know when .....
I'm currently playing around with LuxRender which has some nice features when it comes to lighting and material settings.
I have only touched the surface so far, and its shiny
Would love to work in colab but not sure what help I would be tbh.
Help-wise, it all depends. I don't know how much you know about programming, but if we come up with a good idea, I might take a shot at implementing it. I can also easily modify my Povray exporter so we can output a different file type. That gives me an idea already. I could output a wavefront .obj and build the faces based on point proximity (the result would be a little jagged, but that's fine). Then we could use MeshLab to simplify it and Blender to output. HMM! This sounds very intriguing! What do you think?