Kalles Fraktaler 2.13

  • 90 Replies
  • 4955 Views

0 Members and 1 Guest are viewing this topic.

Offline gerrit

  • *
  • 3f
  • ******
  • Posts: 2265
« Reply #60 on: August 17, 2018, 02:52:47 AM »
Here's a good example of default SA setup  failing, you need to set SA terms manually to 10 to get rid of the glitch.
It fails at resolution 16384^2. Fine with default settings in SMB (about same speed, ~20mins on my computer).

Offline gerrit

  • *
  • 3f
  • ******
  • Posts: 2265
« Reply #61 on: August 19, 2018, 02:08:02 AM »
Small bug with open kfb feature: Save kfr, save kfb, exit. Load kfr (ESC) load kfb: Divide iteration gets reset to 1.

Offline gerrit

  • *
  • 3f
  • ******
  • Posts: 2265
« Reply #62 on: August 19, 2018, 02:53:10 AM »
NR zoom to mini does not work on this power 6 kfr. NR zoom in matlab no problem.

Online claude

  • *
  • 3f
  • ******
  • Posts: 1684
    • mathr.co.uk
« Reply #63 on: August 19, 2018, 03:37:04 PM »
Small bug with open kfb feature: Save kfr, save kfb, exit. Load kfr (ESC) load kfb: Divide iteration gets reset to 1.
divide iteration is saved as an int in the kfb file, possibly a relic from before it was upgraded to floating point in more recent kf versions.
https://code.mathr.co.uk/kalles-fraktaler-2/blob/75401b8192bd1c7e53091add63513642d110fea6:/fraktal_sft/fraktal_sft.cpp#l2598
workaround might be to load kfr to set location, ESC to cancel rendering, load kfb, load palette in colors dialog (*.kfr renamed to *.kfp).

Offline gerrit

  • *
  • 3f
  • ******
  • Posts: 2265
« Reply #64 on: August 25, 2018, 05:16:06 AM »
What's the purpose of the Actions->Arbitrary size menu item? All it seems to do is prevent you from changing the aspect ratio if unchecked.
Why you you ever want this turned off (as it is by default)?

Online claude

  • *
  • 3f
  • ******
  • Posts: 1684
    • mathr.co.uk
« Reply #65 on: August 28, 2018, 11:46:17 AM »
kf-2.13.8 released! download at https://mathr.co.uk/kf/kf.html#kf-2.13.8

 -   new feature: auto skew (via Newton-Raphson zooming dialog)
 -   new feature: show/hide crosshair window (suggested by gerrit and others)
 -   new feature: quality presets ("fast" but inaccurate, "best" but slow)
 -   new feature: page up / page down keyboard shortcuts to zoom in / out
 -   new feature: drag-and-drop parameter files from the file manager to the main window to open them
 -   new feature: drag-and-drop palette files from the file manager to the color dialog window to open them
 -   new feature: CLI prints total remaining pixels (suggested by gerrit)
 -   bugfix: suppress error dialogs when loading metadata from TIFF
 -   bugfix: remove obsolete .ini stuff that was overriding default settings
 -   bugfix: enabled "no reuse center" by default (without it zoom out sequence sometimes glitches)
 -   major documentation improvements
 -   new dependency: GLM 0.9.9.0

Offline gerrit

  • *
  • 3f
  • ******
  • Posts: 2265
« Reply #66 on: August 29, 2018, 06:33:40 AM »
Thanks. Auto-skew seems to work well (tried on mandelbar). Also nice to know what all those formulas are.

Offline Kalles Fraktaler

  • *
  • Fractal Friend
  • **
  • Posts: 13
« Reply #67 on: August 30, 2018, 09:24:51 AM »
Excellent.
The auto skew is really cool! I haven't followed the development for along time so I don't know anything about it, looks like magic to me  :awe:
Quote
-   new feature: quality presets ("fast" but inaccurate, "best" but slow)
How can I test the fast and inaccurate?

Online claude

  • *
  • 3f
  • ******
  • Posts: 1684
    • mathr.co.uk
« Reply #68 on: August 30, 2018, 04:13:09 PM »
How can I test the fast and inaccurate?

Menu -> Special -> Presets

It doesn't immediately re-render the image (only takes effect after zooming or pressing F5).

Excerpt from the user manual https://code.mathr.co.uk/kalles-fraktaler-2/blob/kf-2.13.8:/README.md#l1065
Quote
   - **Presets**
 
     Set groups of settings to suggested preset values.
 
     - **Fast** accuracy may be compromised but it's fast for browsing.  Sets
       ignore isolated glitch neighbourhood to 4, enables guessing, disables
       low tolerance for glitches and approximation, and disables derivatives
       computation unless analytic DE colouring is currently in use.
 
     - **Best** highest quality settings for important images, but slow.  Sets
       ignore isolated glitch neighbourhood to 0 (disabled), disables guessing,
       enables low tolerance for glitches and approximation, and enables jitter
       with a default seed of 1.

Offline gerrit

  • *
  • 3f
  • ******
  • Posts: 2265
« Reply #69 on: August 31, 2018, 01:44:59 AM »
Something is wrong with ADE in this one, NDE is fine.

Offline gerrit

  • *
  • 3f
  • ******
  • Posts: 2265
« Reply #70 on: September 01, 2018, 11:33:26 PM »
Here's another mandelbar location, with KF ADE and with UF using the ADE formula
\( d = ||w||^2log(||w||)/||w \cdot \frac{\partial w}{\partial c} || \)
with \( w=(x,y) \).

Offline gerrit

  • *
  • 3f
  • ******
  • Posts: 2265
« Reply #71 on: September 03, 2018, 09:09:14 AM »
When you load kfr with drag 'n drop, the filename does not get updated (as displayed on top bar and for Save).

Online claude

  • *
  • 3f
  • ******
  • Posts: 1684
    • mathr.co.uk
« Reply #72 on: September 04, 2018, 03:58:03 AM »
thanks for the bug reports re bad mandelbar ade (EDIT: the derivatives in formula.xml are bad, need to redo) and dnd naming, i've noted them and hope to test them soon

not sure what the arbitrary size thing is about, though i've managed to crash KF by dragging the window edge to resize it (in WINE).
« Last Edit: September 05, 2018, 04:18:19 PM by claude, Reason: found cause of Mandelbar ADE bug »

Online claude

  • *
  • 3f
  • ******
  • Posts: 1684
    • mathr.co.uk
« Reply #73 on: September 06, 2018, 09:38:24 AM »
in the auto skew thread I posted a failure for:
- 4th Celtic False Quasi Heart :  z := (|x^4 + y^4 - 6 x^2 y2| + i 4 x y |x^2 - y^2|) + c

I noticed the auto skew works with this variant, which makes more sense as a 4th power formula (all terms are roughly x^m y^n with m + n = 4):
- 4th Celtic False Quasi Heart :  z := (|x^4 + y^4 - 6 x^2 y^2| + i 4 x y |x^2 - y^2|) + c

Siimilar typos (y2 instead of y^2) affect all 4 of the 4th False Quasi formulas' 'et'-generated code (which is only Newton-zooming related thus far, so it's not such a big deal). I've fixed in 'et' and I'll regenerate the formula sources for the next KF release... has been broken for the whole kf-2.13 release series afaik

Online claude

  • *
  • 3f
  • ******
  • Posts: 1684
    • mathr.co.uk
« Reply #74 on: September 06, 2018, 04:45:54 PM »
Something is wrong with ADE in this one, NDE is fine.
I think(?) I fixed Mandelbar ADE, this is what it looks like here in next KF version (hopefully to be released today..) with 10x10 supersampling with jitter, ADE looks a bit grainy.


xx
Kalles Fraktaler 2.15

Started by claude on Kalles Fraktaler

45 Replies
1012 Views
Last post Yesterday at 07:54:34 PM
by claude
xx
Kalles Fraktaler 2 + GMP

Started by claude on Kalles Fraktaler

232 Replies
12664 Views
Last post September 12, 2019, 10:27:53 PM
by claude
xx
Kalles Fraktaler 2.14

Started by claude on Kalles Fraktaler

287 Replies
10484 Views
Last post Yesterday at 07:52:15 PM
by claude
xx
Another version of Kalles Fraktaler?

Started by greentexas on Kalles Fraktaler

9 Replies
1200 Views
Last post September 22, 2017, 02:59:16 PM
by greentexas
xx
Feature ideas for Kalles Fraktaler

Started by FractalAlex on Kalles Fraktaler

13 Replies
543 Views
Last post November 23, 2020, 08:38:20 PM
by the_chief