Welcome to Fractal Forums

I have uploaded a new version that I have been playing with for a while.

* I have tried to make the zooming experience similar to Fraktal Extreme by zooming centered to the mouse pointer when using the mouse wheel or +/- keys.
* I implemented arbirary zooming from an old request - hold the Control key and use right mouse botton to make an arbitrary track rectangle.
* When using animated zooming, simple left click behaves as FX and dragging the view.
* Animated zooming can be disabled, which changes to the old behaviour of mouse click etc.
* I also implemented arbitray size ratio of the window. This is disabled as default.

I had some ideas of movies that I wanted to realize.

First, I wanted to combine slopes and rotating side scrolling to emphasize slope shadows:
https://www.youtube.com/watch?v=Sbu767eDDDg

I also got a excellent suggestion from chillheimer to use timestamps instead of "Movie frames per Key frame" for speed. This movie tries to synchronize with slow and fast parts in the music. However, since I realized the acceleration from slow to fast takes "long" time I should have offset that speed change one second earlier:
https://www.youtube.com/watch?v=LlLHaCocjHE

I also wanted to implement Distance Estimation.
This is actually a cheat that is related to slopes, so no derivative is calculated because I don't know how to do derivative with perturbation
https://www.youtube.com/watch?v=NKwkujUkMr0

All functions in the dialog boxes are now explained with tool-tips.
Since the test department is on holiday, there may be bugs ;)

scrollwheel-zoom rocks! feels great!
and finally drawing a box in the size you desire! thats so perfect! been looking forward to this a long time! :D  and dragging! woohoo! all this combined really makes the navigation so much better. mission accomplished!
one small thing:
rectangle drawing: would it be hard to change it so that the point where you first click is the centre of the opening rectangle and not one corner? you click on the detail you want in the center and then you determine the size of the box around it. seems more intuitive for me..

mouse overs in the moviemaker are really helpfull!

time-sync in moviemaker! yeah!  now if we had this for cycling as well...  ;)

your program really has come a loong way.. :)
keep up the great work! (whenever you feel like it ;)

I'll admit that I put KF down when I found navigation too unpleasant without these new features, so am looking forward to picking it back up now   O0

When drawing a rectangle, and I have not tried it yet, but does it allow for moving and resizing the rectangle after it is drawn?  That is the most important part of drawing an arbitrary zoom thingy!

Glad to see the custom aspect ratio too - another reason I stopped playing with KF is no longer an excuse   :police:

Does window/image size save with files now?  When deep zooming I use teeny tiny itsy bitsy windows for faster rendering, but when opening those old files they always open with larger windows.  Another thing to consider..

Thanks for the updates as I am certain some of those requests were mine  (thought others may well have also requested them either before or after I did!).  Guess I'll go download the new version and play a bit...

I don't think so, but as workaround load your pictures and hit "escape" right after, so it won't compute the whole thing. then change the imagesize and press f5..

Thanks a lot :)

I have never considered storing the size of the window or the background image.
I put a lot of efforts preserving the skew so that you can open a hard skewed location in non-standard aspect ratio :)

Track rectangle could indeed begin from center.
Unfortunately as soon as you release the mouse the zoom is performed. Currently... ;)

Press the '+' button on your keyboard and keep it pressed for a while. Realtime zoom animation! (until you get too deep ;) )

hm, results in ultra fast jumps in the set zoom size - do I have to activate 'floating' zooming somewhere?

so if you enter a time like 4:33 now it results in the speed change starts at 4:33 and ends later, depending on what happened before?
how about using the time es endpoint for the speed-change instead of startpoint?

Nice update, and thanks for updating the histogram. I'll try to figure out how to use the moviemaker. The update to the zooming is very nice.

Is menu "Actions->Animate zoom" checked?
Currently the speed change start on/after the time stamp.
So yes, the speed change should happen before.
It might be difficult though...

However, if "Skip render" is checked, you will be able test the render rather quickly also with speed changes.

I've made some test-renders for the histogram and everything seems to be right except the slopes I tried to add those but they dont show up when using the new moviemaker.

Edit: they do show up but not if the final movie is smaller than the rendered frames.

Slopes cannot be combined with distance estimation...
Some parameters are not read until you start a movie render. So you may need to check the "Test" checkbox and press "Start" for these parameters to be read...

Ok thanks. I'll try the test button first. If it doesn't work I can also change the palette to add a little more contrast which will give an appropriate effect.

The distance estimation is very nice to work with. :)

I found a little bug with saving a jpeg. The 16:9 ratio doesn't work properly.

Yes I know, this happens when you resize the background image size a couple of times.
You can easily solve it by resetting the skew (Menu Actions->Skew or Ctrl+K)

But this is something to improve!
I find it hard to combine arbitrary aspect ratio and stretch/skew. If you set the image width to a multiple of 16 it is easy to calculate the height. But what if you set it to 500, 500*9/16 = 281.25, but the sizes of the window and the background image are integers...
It wouldn't be any problem if height would always be calculated by multiplying width with the ratio of the window.
But the program also allows stretching to reveal skewed locations in abs fractals. :)

Ok thanks. I didn't thing it would be a problem but it is good to know how to fix it.

I have made another small update:

KF 2.9.1

• In this thread, http://www.fractalforums.com/mandel-machine/glitches-and-crashes/msg85895/#msg85895, I saw a nice suggestion of enhancing the exploring experience by not filling glitches with one colors, but instead let them remain the stretched pixels from the previous view. This can be disabled by the menu "Actions->Special->Show Glitches
• Enhanced Distance Estimation

KFMM 1.31

• Fixed color cycling when color division is more than 1. This is for old stylish solid coloring
• Enhanced Distance Estimation
• A second option for entering speed, "Time NS" = Time No Smooth. Because the smooth speed transitions is difficult to control

I really like the exploration experience improvements (mouse wheel zoom around cursor location, using the previous view to give an idea of the zoomed view, animating the transition, only updating completed pixels).

I've not got the hang of the distance colouring though (I'm guessing it's calculated using a method similar to slopes rather than true derivative-based distance estimate?).  The palette repeats in a way that I don't like - I'd rather it stops at the end, so dense regions aren't so aliased and grainy while sparse regions are still bold.  I ended up combining 4 slope-colouring images at 90deg angle increments in GIMP using the Darken Only layer mode, then post-processing a bit (contrast stretch, blur to remove some grainyness, downscaling), to give something that looks more like what I want (top image, second image is one of the angles, parameter file attached too).

Some other minor things I ran into:
* auto-solve glitches isn't on by default, which meant that my first zooming ended up pixelated at double precision limit (I don't remember exactly which option fixed it, maybe it was a different option)
* reuse reference can't be selected at the same time as auto glitch solving
* I couldn't get the palette waves (the one below the infinite waves) to work properly, no matter what I chose for offsets it just ended up shades of grey when all the periods were the same
* reducing contrast then increasing it again gives a posterized effect (colour precision is lost when reducing contrast)
* distance colouring didn't do what I expected (see above, maybe an "all-directions/fake-de" toggle for the slope colouring would be an idea?)
* would be cool to be able to set the bright and dark colours for slope colouring, rather than just white/black - or even a second palette that gets applied according to the angle of the slope (which could be used for the previous point too)
* at one point the timer stopped updating in realtime, but when the image completed it showed the time (5:48.923 done ref 34)
* the menus are a bit confusing (reorganising them into submenus for different aspects could help)

I tested the 32bit version with WINE on Debian.

Thanks :)
Indeed, DE is inspired by Mighty Mandel :)
However I have attached what I wanted to achieve, several layes with different colors.
Background is blue. First layer is black, second layers is dark green and third layer is yellow. Each layer comes from a close passage of a Minibrot.
You cannot combine reuse reference and automatic glitch correction, since reuse reference means that one reference only is used and glitch correction involves several references. Reuse reference is meant to be used for the frames closest to the final minibrot in a Movie :)
If you use e.g. 4 periods on 128 colors, you would need to offset it with 16 in order to move it half a period.
If you use 32 periods on 128 colors, you would need to offset it with 2 in order to move it half a period.
Yes, reducing contrast to grayish colors and then back makes color information to be lost

I need to come back on the other comments :)

Very nice!  I'll try the parameter file soon to see how you did it.
I described a method for finding secondary references using the glitch detection test:  http://www.fractalforums.com/announcements-and-news/pertubation-theory-glitches-improvement/msg84154/#msg84154
The primary reference can be re-used (along with its series approximation coefficients), the secondary references don't need to be recomputed from scratch (they are found at their own period) - it turns out that a single iteration of Newton's method is almost always enough, and using the double precision derivative computed for distance estimation is usually enough, so it's quite cheap to find the next reference.  I haven't released my implementation yet, it still has too many annoying bugs (sometimes zooming too quickly in the GUI crashes in an infinite loop waiting for worker threads that already exited, sometimes a bad primary reference is used giving a blank screen or other wrong appearance...).  But no time to fix them for a few months...
Thanks for the explanation :)  I thought period meant the length of the cycle, but it seems it's the number of cycles inside the palette length (so more like a frequency than a period).

Hi, just want to say that this software is amazing! I never thought I'd be able to view deep zooms of the Mandelbrot set so quickly.


I have a request, though. Could you add a lossless image save option(like PNG or BMP)? JPG has too many ugly artifacts, even with high quality. I was looking into adding this for myself, but realized the source you provide is from an old version. I'd be missing out on newer features if I built onto that, so I have to ask you to add it.

How do u save settings!?

Select "save".....???

but window size, pallete, etc

edit: figured it out  :)