Mandel Machine

[Dinkydau]

I think a large bailout is not good. A bailout of 2 is perfect and natural to the mandelbrot set, because it is the minimum bailout required to catch all the points that belong to the set. A larger bailout value doesn't affect the set itself, but it does affect the structures of the iteration bands. There, valuable information is lost. It is especially apparent in the needle/antenna. Even at a magnification of something like E10000, the structure of the iterations bands contains enough information to find your way around the line and find a minibrot.

[hapf]

I think a large bailout is not good. A bailout of 2 is perfect and natural to the mandelbrot set, because it is the minimum bailout required to catch all the points that belong to the set. A larger bailout value doesn't affect the set itself, but it does affect the structures of the iteration bands. There, valuable information is lost. It is especially apparent in the needle/antenna. Even at a magnification of something like E10000, the structure of the iterations bands contains enough information to find your way around the line and find a minibrot.

A large bailout is required for distance estimate when you zoom in deeper or you get artifacts. If you don't need distance estimate 2 is fine.

[Botond Kósa]

I think a large bailout is not good. A bailout of 2 is perfect and natural to the mandelbrot set, because it is the minimum bailout required to catch all the points that belong to the set. A larger bailout value doesn't affect the set itself, but it does affect the structures of the iteration bands. There, valuable information is lost. It is especially apparent in the needle/antenna. Even at a magnification of something like E10000, the structure of the iterations bands contains enough information to find your way around the line and find a minibrot.

Yes, there are some places around the set where a low bailout can help in finding a minibrot, or it can even have an aesthetic value. I might make the bailout adjustable in the future, but it will need some major rewrites since many optimizations depend on the current high value of 32.

[Botond Kósa]

Version 1.1.1 is out, with better handling of very large images (http://web.t-online.hu/kbotond/mandelmachine/, a refresh might be required). I noticed some confusion among users regarding the image sizes created by the program, so here is a short explanation of how it actually works:

The width and height specified in the Image box are the size of the image that can be copied into the clipboard or saved as a PNG file. If there is no supersampling applied to the image, the computed backing image has the exact same size. If a supersampling of N*N is used, the size of the backing image is (width*N)*(height*N). This is now displayed under "Actual size" in the Image box. The backing image must fit into the memory with 8 bytes per pixel, and cannot be larger than 536 megapixels. This means that the maximum image size currently available is about 30720x17280 with no supersampling, or 15360x8640 with 2x2 supersampling, or 7680x4320 with 4x4 supersampling etc.

List of changes:

• NEW: Image size in megapixels shown in the Image box
• FIXED: Application crashes when trying to create an image that doesn't fit into the memory or is larger than 536 MP

[Botond Kósa]

And here is version 1.1.2, with the ability to extract the palette from bmp images as well.
http://web.t-online.hu/kbotond/mandelmachine/

If you still have stability issues, please let me know.

[SeryZone]

Hello! How to correctly load palette? All my palettes is .bmp files, 4096x10. When I load and press 'OK' program do nothing! How to do that?

[simon.snake]

Hello! How to correctly load palette? All my palettes is .bmp files, 4096x10. When I load and press 'OK' program do nothing! How to do that?

After you load the bitmap (.bmp) file, you have to drag a line from one place to another on the bitmap that is displayed on the screen to set the palette, so for your image, you would start at the left hand side, drag a line to the extreme right, and the palette would display in the area above.

This means you do not have to use the entire image for palette, you can load a picture of anything and create a palette from that.

[SeryZone]

After you load the bitmap (.bmp) file, you have to drag a line from one place to another on the bitmap that is displayed on the screen to set the palette, so for your image, you would start at the left hand side, drag a line to the extreme right, and the palette would display in the area above.

This means you do not have to use the entire image for palette, you can load a picture of anything and create a palette from that.

Thanks a lot! I understand! Wait for new colorful images!!!

[Botond Kósa]

This means you do not have to use the entire image for palette, you can load a picture of anything and create a palette from that.

And you should make your palette bmp higher, at least 400 pixels if the width is 4096 pixels, because it will be scaled to fit the width of the palette window. If you load an image with 4096x10 size, it will be scaled to 512x1, so you would have to drag inside a one pixel stripe. I will correct this in the next version.

[Botond Kósa]

Version 1.1.3 is available, with better handling of SeryZone's 4096x10 pixel palettes.

[panzerboy]

How many palette colours can you set from an image?
I've been initially loading a Kalles Fraktaler .kfr file to set the palette then resetting the location and begin exploring.
That means I can't change the palette, loading another .kfr would change the location.
I should be able to write a program to convert the my many .kfp palettes to a bmp.
A 512x50 pixel aspect ratio would work best?
So for the 1024 colour .kfp palette would be a 1024x100 bmp.

[Botond Kósa]

The internal palette representation stores 1 million colors in an array. When loading a .kfr file, the colors (up to 1024) are mapped uniformly into the array, and the remaining slots are filled using interpolation.

When extracting colors from an image, the same method is used. If you drag a line with a length of X pixels, X colors will be mapped uniformly, and the rest will be interpolated. For slowly changing gradients it is best to rely on Mandel Machine's interpolation which uses cubic functions to achieve very smooth transitions.

512x50 or 1024x100 pixels will be OK. It is now even possible to load images with 1 pixel height, they will be stretched vertically to 50 pixels.

[SeryZone]

Hello, Botond! Can you make primitive zoom-maker. At beginning I set location, zooms, width and height and Supersampling. On rendering program calculates from the end to the start coordinates and make .png images. Please, make it! it is simply!

[panzerboy]

I'm finding a frustrating 'opportunity' with Mandel Machine (v1.1.3).
On large images 1680x1050 with high supersampling 4x4 the color offset slider seems set the previous slider position not the current.
If the slider is positioned in the middle and I make a big change sliding it all the way to the left it only shifts the palette on the rendered image slightly.
If I then slightly nudge the slider to the right, a big change to the images palette happens.
It seems the rendered image renders the previous position not the current.
This happens when nuding the offset slider, say I nudge the slider several times to the left.
I've overshot so now I nudge to the right, the image changes as if I had continued nudging to the left.
After another nudge right and the images changes as expected.
Always being one step behind makes it very hard to get the exact background colour, especially with my stripey palettes.
The workaround is to get the palette offset 'just so' with no supersampling before committing to a 8x8 supersample render.


Machined Hooks by panzerboyNZ, on Flickr

[Botond Kósa]

Some info about the RAM usage of Mandel Machine:

The program stores the following structures in RAM:

• computed iterations (4 bytes / pixel)
• rendered image (4 bytes / pixel)
• reference orbit coordinates (16 bytes / iteration)
• auxiliary arrays for series approximation (96 bytes / iteration)

The size first two structures are defined by the actual size of the image, so if you are rendering a huge image, it will need a lot of RAM (for example an 1920x1080 image with 16x16 supersampling needs 4GB just for storing the image). The size of the reference orbit is determined by the iteration limit set under Location. Using a high limit of e.g. 100 million needs an additional 1.6 GB of RAM. The auxiliary arrays usually contain only 2 elements, but can contain limit/2 elements in worst case.

The program is currently allowed to allocate 4.4 GB RAM if needed (max image size = 536 megapixels, multiplied by 8 bytes, plus some headroom for the other structures). The RAM usage is shown in the lower right corner. I will have to raise this limit to allow max sized images to be calculated with maximum iteration limit (536 MP * 8 bytes + 200M iterations * 16 bytes = about 7.5 GB).