Visual oddities in various fractals since 2.09

[tpe]

I've noticed some rather interesting visual glitches in 2.09 (and 2.10-dev), usually in the form of spheres in certain places in IFS formulas.

An example, using the octahedron settings file that comes with the program:


Another example on the octahedron settings, zooming far into one of the triangular 'pits':


This also crops up on the standard Menger Sponge:


An interesting thing to note with these unexpected sphere shapes is that if I encounter one during a flight recording and attempt to 'fly' into it, the camera bounces back and reverses temporarily, before regaining its momentum and moves forward once more. I can sit there and do that repeatedly.

And one more thing, there's something that has happened to hybrid rendering between 2.08 and 2.09.

This render in 2.08:


Turns into this in 2.09 and 2.10-dev (exact same settings file, just with DOF switched off):


I've attached the settings for these zooms to the post.

Sorry for the rather large post!!

[mclarekin]

Hello tpe,

thanks for sending the settings files. I have tested the first one and reproduced your image.  It looks like it might be a DE problem. I tried a different DE calc and the sphere disappeared,  but also  part of the fractal disappeared as well. 

thanks for reporting, I will  post it as an issue at github.

[mclarekin]

compute_fractal.cpp . code at line 1073.  Temp fix:  Either comment out like I have or try changing code to r < 0.000000001, (both seemed to work for me)

But the bulb hybrid I can not get your image yet, hmmm??.  And  it sure takes a long time to render

Code:

//				if ((z - lastZ).Length() / r < 0.1 / fractals.GetBailout(sequence))
//				{
//					out->maxiter = false;
//					break;
//				}
			}
			else if (Mode == calcModeDeltaDE1)
			{
				if (r > fractals.GetBailout(sequence))
				{
					out->maxiter = false;
					break;
				}

//				if ((z - lastZ).Length() / r < 0.1 / fractals.GetBailout(sequence))
//				{
//					out->maxiter = false;
//					break;
//				}

[tpe]

Thanks mclarekin, I'll give that code edit a try and see what I get. And hmm, that render is interesting because that's pretty much exactly what I see in the preview before loading the settings in 2.08 - i.e. it looks like my render just with a lower detail level value.

And yeah, it's quite a slow-rendering set of params but it was one of my favourites.

A shame to see it randomly supernova when I updated to 2.09!

[Buddhi]

Thank you for reporting this problem. You have discovered side effect of special bailout condition implemented for pseudo-cleinian fractals.
Now I see that I need to code auto-disable this additional bailout condition for fractals without pseudo-cleinian components.

Actual workaround (without changing code) is to disable "Rendering engine / Use default bailout for each fractal formula" and increase bailout value.

Corrected settings for AObulbboxhybridHC2-08.fract. I have changed Distance Estimation Function to Delta DE logarithmic and on first fractal tab (for Mandelbulb) I have disabled "Check for bailout condition". These settings give exactly the same image as you had in Mandelbulber 2.08.

Code:

# Mandelbulber settings file
# version 2.09
# only modified parameters
[main_parameters]
ambient_occlusion 4,5;
ambient_occlusion_enabled true;
ambient_occlusion_mode 1;
ambient_occlusion_quality 2;
bailout 3548130000;
camera 0,0500965750401106 -0,1065077873601582 0,03404257193919225;
camera_distance_to_target 0,002442793005611771;
camera_rotation 25,47095489628087 -23,30182834296036 -2;
camera_top -0,2015220815212444 0,3419000851850035 0,9178742737487703;
check_for_bailout_1 false;
contrast 2;
delta_DE_function 2;
detail_level 2,4;
DOF_enabled true;
DOF_focus 0,00246476917527616;
DOF_radius 7;
flight_last_to_render 0;
formula_2 37;
glow_enabled false;
hybrid_fractal_enable true;
image_height 512;
image_width 512;
keyframe_last_to_render 0;
main_light_enable false;
mat1_coloring_palette_size 255;
mat1_coloring_saturation 2,06;
mat1_is_defined true;
mat1_surface_color_palette ff0c00 7fb700 ffff99 ffee00 005baf f77752 d0004a 00ceff f5fff5 938700 56ffff 5acaff ad2784 00006a 000000 c4ffce 00c35a d64a5e 001fff 0000ff ff73ff cbe6ff 2200ff eabeff 51ebbe 42abff ff2d7b 000074 00c5ff 37ff75 000040 530000 fffd00 0000ff ff0000 88ffcc 7f03ee 0000af 004d00 ff00c6 cd7455 92d404 00fdea 61632c 61ca10 8383ff 000419 00ff00 a0ff9c 0c0021 de3eff 000660 00003c 5083ff aa0008 a32da1 36ff00 c2cd19 ffff36 000000 50f167 0067bf ff1200 ff023c 44ff75 00005e ff002b 69000c ff1d61 a1ffa3 5e00ff 00ff79 ff00ff cdff8d d95fa7 2c00ff 4fff00 29ffda 00ffaf ffffff ff3304 00ffc1 a5fcff ffffff 00d053 ffb41c 006660 ffff37 7718e4 ff00ff a9cc60 ff0000 686200 ffff49 ffcd47 ff00c0 ff00c2 188300 86ff8d 1100ff 00e8ff 4bff53 44c800 3ee536 589600 ffb2a3 a5005b ffff1c eaff6e 19062d ffff00 547f4a f5e0ec ff00ff 0aff00 005115 00b64b 98abef ff22c3 00ffd6 0000f4 ff00f0 ffff00 b3ff00 ffd600 00ff57 b3c9ff 00ff65 00ffe0 ca0010 00e1ff 0000c6 e30f00 00014b 6cffcf e600cc f12100 ffffff 000000 ee0000 cf1279 00b98c ffffff ffffff da1bde 519d74 ffff1f 15c642 ff8350 ff0032 c08f9d 006def ffd337 05fffe 050053 96ff00 ffffff ad00d4 00ffff 59a3ff 00ff7f 3d2500 00e193 5aff00 ffffff ffac00 0000a7 00005f 00a2ff 4b00b6 4eff00 d9a8c1 000000 06ff25 a9f792 a2ffa0 fc47c7 8200ff 975726 ff0009 002d29 ffff00 171000 00ffff fffd00 003d00 00ff00 b4534f c300fc 0e00e6 003000 d70013 6fff48 ffad00 888a5f 00ffff 93d5ff 0083e8 000022 ff3cff ffff00 ff9c00 ffbd00 ff5800 009600 25b308 00ff00 2175ce ff0000 ff00ff 00ff72 7300ff 1900ff ff0000 453914 a200cf 0000ff 5affb0 ffff00 ff21ff ffff33 fff600 00ff04 31d067 b3ff00 ff925c 0cff00 83ff00 ff0000 d13c99 0000d2 beff86 00a098 8b0000 3de456 000011 ffbf35 52002b 000000 673300 ff54ff ff4fc4 00ff63 00db2e 938d00 91ffff 2b6333 4e00ff dc00cf d80000 570000 830dff 9200ff 682aff a200ff;
mat1_use_colors_from_palette false;
raytraced_reflections true;
target 0,04913173129785908 -0,1044823085851727 0,03307626455880695;

[tpe]

Confirmed working for bulb/box hybrid and suggested changes to bailout do indeed fix the spheres problem. Thanks!