My bet is the output options that this cool java jar does can be done in G'MIC as well. Got the link from [Günter Bachelier](https://www.researchgate.net/profile/Guenter_Bachelier). This cool applet actually renders output faster and better than mathmap does. I'm quite impressed.





https://karczmarczuk.users.greyc.fr/TEA ... pStill.jar

That is awesome! LOL! I still do the renders the old fashioned hard way, in Linux.

Since it's a java jar, it probably can run in Linux as well, Wall.. um, er, Mahvin. lol

Why not some more fun.

https://www.flickr.com/photos/34520999@ ... 6/sizes/o/

And used this equation: ((1.1+(i*z)/6)/(1.04-(i*z)/6))^6.2

and chose a 1600X1600 output for this result.

Man that's fast!


Thanks for the link Lyle.

Fast and very clean output. They must use some form of gpu acceleration is all I have to say about this, Rod. Also, unlike mathmap and the seam problems, this applet produces very clean output. Now to see if Dave can do this in G'MIC (posted request here).

Lyle, let me know what you think about this

So sweet David! Thanks.

Used your values. This took about 3 minutes to render.

Hi Rod,

3 minutes ? That is quite a lot, what was the resolution of your image ?
All the tests I've done have been generated in less than 10 secondes, even for a 3200x2048 image.

Yes; this one took around 3 minutes too, David. Used the default custom formula. If you use just z it still would take 3 minutes; of course no change happens afterwards with just z. Still nice to see this ability; maybe can be speed up some.

Some of them almost look like continual droste. I like that Conformal maps Lyle.

I'm assuming you need to use the pre release gmic to get that filter?

Yes, Sallyanne, you do need the Gmic 2 but you can install both. They live quite happily together on my machine.

Great turing, Lyle, but very hard on the eyes.

Yup; lol

I got this complaint about this particular pattern before. I guess I'm a sadist. lolol

I'm a bit worried about the timings you get.
Could someone try the filter on Linux. Even on my laptop, it takes like 8 seconds for a 3000x3000 color image, so I can't understand why it should take 3x60 = 180 seconds on another machine, even on Windows ! There must be something wrong that should be fixed.

Gimp 2.8.20 - Kubuntu 16.04 - laptop 8 GB mem 2.8 cpu - settings as shown - original 800x600 to 3000x3000 new image

about 7 to 8 seconds according to my stopwatch.

Thanks James, how many available cores do you have for the computation ?

EDIT : I've always found that gmic on Windows was a bit slower than on Linux, without any explanations why it is like this. Maybe there's only a simple hack to perform to get the same performance as in Linux, I'd be happy to know

Wow; maybe I need to install GIMP on my 64-bit Ubuntu VM and give it a whirl. lol

Yes, Lyle, that would help, really
If you find that it is way faster, even on a VM, then we can worry a bit much about the Windows version !