How does it work?
All calculations are done locally, there is no exchange with servers. Calculations are fast on newer
computers: each image is divided into as many processors as your computer has (often 8 nowadays).
Each
part of the image is therefore calculated in parallel. Calculation times are in the order of 50 to
400
milliseconds per frame with an Intel i7 process. Only the variant of the Mandelbrot set in cos (z) -
that is,
in complex cosine - of the point requires heavier calculations that are felt on the calculation time
(from the
order of the second to several seconds for deep zooms).
In practice we iterate (we repeat the calculation on the previous result) a function for each point
and see
if it diverges (it exceeds a certain value). At small scales, you may have to iterate more than 1000
times for
each point and this on all the 640x640 = 409,600 points!
The program is written in Javascript. This language models floating numbers (decimal points) with a
maximum of
fifteen significant digits. This is why the image definition becomes bad beyond a magnification of
the order
of 1015. Which limits the magnification to just 1 million billion times...
When you
get into
the game of exploration it may be frustrating to reach the limit but going beyond that limit with
Javascript
would make the navigation much, much, much slower (several tens seconds for each image).
Browser Performance
The microprocessor and graphics card of the computer or tablet play a great deal on speed. But the
browser
used also. Chrome and Microsoft Edge give the best results (both based on the same Chromium
platform). Firefox
does not handle well beyond two processes (“worker”) in parallel, calculation times on the same
machine are
roughly 50% slower.
Where to start?
The easiest way is to let yourself be guided by your instinct. Some areas can be repetitive or even
disappointing. To start your journey in the middle of fractals, you are offered for each set a few
POI
Points of Interest. You can access the POI list by clickingin the command bar.
Orienting
Points of Interests
For each set, you are offered a number of “places” for their interest and help you get started in
your
exploration. By clicking on the corresponding thumbnail, the area will be displayed directly and you
will be
able to continue exploring normally.
Center on a point
To find a view from the name of an image saved from our site - see the filename below - or because a
specialized article mentioned a particularly interesting point, you can use the “center on a point”
function.
Julia Set
For Julia set do not forget to indicate before starting the search the number “c”. Unlike other sets
to
explore, Julia's set, or more precisely “Julia's” sets depend on a parameter. Usage calls
this
parameter (a complex value) “c”.
You can enter the values of “c” in the fields below the image:
rc
indicates the actual part of “c” and ic its imaginary part.
How to zoom in Fractal Explorer?
Zoom with the mouse wheel
By pointing the mouse at a particular point, if you press the wheel you enlarge the scale, and vice
versa.
Each action on the wheel allows you to zoom in or out by a factor of 2.
Zoom finely
If you click on a point in the image a frame appears and follows your mouse. You can cancel by
pressing
Esc.
On the second click, you define the zoom frame.
Zoom on a touch screen
To zoom by a factor of 2, just press twice quickly. The image will be centered on the affected point.
To zoom
out, tap the “Show previous view” icon
Logarithmic scale
A logarithmic
scale
allows you to know the zoom level. The scale turns red around 10-13, area from which
accuracy may
begin to become insufficient. This results in a coarser pixelation of the image.
Filters
Mandelbrot & Co offers some filters to apply on the main view. Filters apply on videos and image
backups
except on Safari. For videos, this significantly lengthens the calculation time.
You can also play on hue, saturation and brightness. In some cases the filters may slightly tarnish
the image,
this effect can be compensated for by increasing the brightness.
The filters used are convolution filters (see links to wikipedia below). For example the Contour
Lines filter
is the combination of the Sharpen filter and the Laplace filter, used for the Edge Detection filter.
Applying the Contour Lines, Edge Detection or 3D shading gives effects very interesting as you can
see below.
No Filter
Edge Detection
Contour Lines
3D - Phong ShadingFavorites
By clicking on theYou add a thumbnail to your favorites in one of the 12 slots on the left side of the screen.
By
clicking on the trash canin the top left, you can select the favorites you want to delete. The removal is effective by
recliking
on the garbage can.
Favorites are very useful for memorizing a “place” during your explorations. Your favorites are
stored
locally in your browser. You find them when you return to the site.
Save Images
You have the option to save the image in png format on your hard drive. The image will be saved in
your
Downloads folder of the browser you are using.
The file name allows you to find the image by its coordinates using the centering system.
The first letter indicates the set (M: Mandelbrot, B: Burning Ship, J: Julia and C: CoSZ). The
following
numbers are used to locate the image according to the nomenclature: (rz) +i (iz) ± (Δ) with the real
part
(rz), the imaginary part (iz) and the Δ
Example: M-0.632173+I0.451271±5.11E-4.png
indicates that this is an image of the
Mandelbrot (M)
set, centered on rz=-0.632173 and iz=0.451271 and for a Δ=5.11E-4
For the whole of Julia, we also need to know the reference point “c”. The file name is completed
according to
the nomenclature: (rz) +i (iz) ± (Δ) rc (rc) ic (ic) .png with rc real part and ic imaginary part.
To infinity and beyond.
With the diving optionyou can create a small movie that will zoom in on the currently displayed area. For Chrome and
Micosoft
Edge browsers a free WebM movie
will be
available for download. If you use Firefox or Safari browsers the encoding process is not supported.
The site
offers for these two browsers only a view of a succession of images (technically it is simply a
slide show)
with an equally striking effect but unfortunately not downloadable.
In any case the process of creating images and, if necessary, encoding in WebM format, may be quite
long
depending on the desired area and zoom level. For the moment, films of reduced size (480 pixels x
270 pixels)
are proposed to ensure an acceptable calculation time, which can nevertheless be in the order of ten
minutes.
As an option you can opt for a full-HD movie (1920 x 1080) but you will have to be patient!
On small screens, such as 9" or 10" tablets, it is not possible to make movies.