I've been a cryptic since I joined the group a few weeks ago but I'm finally at the point that I am announcing that I have converted the MATHMAP Droste to G'MIC using the mathematical parser. I guess it has been a concern that there will be a version of GIMP that can't run MATHMAP an then we will be without the spiraling Droste effect.

I used the so called "Droste 10" source hat includes all the features that are typically shown in the MATHMAP Droste tutorials and videos. The source was originally develop by people such as "breic" and Josh Sommers many years ago. My small part was converting the MATHMAP coding and coordinate mapping to G'MIC's mathematical parser (and a whole load of expanding out complex math into regular math).

As I said it's the full Droste 10 source so it has support for multiple strands, "hyperdroste" fractal effects and the all important grid. I won't claim that it is pixel perfect but it certainly should be close. The only thing that isn't implemented is the antialiasing/supersampling that comes "free" on all MATHMAP filters. I'm up for discussion on the best way to anti-alias.

In terms of speed it works as fast if not faster than the MATHMAP version and it should use all the engine cores on your machine instead of just one for MATHMAP (so I hear).

The only question I have is how do I move it out into the community? I have it in my user.gmic file but how do I distribute it to everyone else. It's about 400+ lines with some comments and spacing. I have some real quick images I did with it, below:









This last one shows the filter screen showing a preview which updates quickly and seems to be accurate and the grid option superimposed on it.



Souphead.

Now that would be cool Souphead. Looking forward to it. Hopefully you can get it added quickly under test and David, I'm sure, will add it to Deformations once you are happy with it.

excellent work, Souphead!
Me too I hope David can release the filter under official G'mic

Hello Souphead,

That's awesome news ! This particular script was on my TODO list for several years now, but I hadn't take the time to convert it properly.
I'm so glad someone else did it
Now, about sharing it to the community :

1. If you think you will have the envy to develop other filters for G'MIC in the future (that would be nice ), the best way would be
that I add you as a contributor to the 'gmic-community' repository, available here : https://github.com/dtschump/gmic-community
and you could add your user.gmic file there, naming it as 'souphead.gmic' (or with your real name of course).
This would define a new filter source that I'll add to the filter update system, to make it available for other people just as easily as with other filters
developed by contributors of this repository. The cool thing here is that you would have the possibility to modify this file when you want, to
make updates, add new filters, or fix bugs in filters (if any).
(That's the preferred solution of course )

2. If you think this is the only G'MIC filter you will work on, then I'll manage the inclusion of your filter in my own filter source 'ronounours.gmic' in 'gmic-community'. You won't have easy ways to modify and update this filter, except sending me a mail with the modifications.

That should be as simple as this. In any way, please tell me also how I can add you to the list of contributors, eventually your real name, a link to your web page, you G+ profile if you have one.
This filter has always been awesome in Mathmap, and having it running in G'MIC is such a good news !
I'd want to make a proper announcement with all the credits you deserve.

Thanks again!

Also, about the 'anti-aliasing' option, I assume this is only working because Mathmap performs some interpolation on the resulting pixels when the pixel coordinates are float-valued ? If so, this is also something G'MIC can do. Depending on how your script has been done, it can be enabled in several ways.
If you are using expressions as : 'i(x,y,z,c) = i(X,Y,Z,C)' to set the pixel value in your long expression (with X,Y,Z,C float-valued), then you can specify the use of linear interpolation in the right-hand size of the equation, as : 'i(x,y,z,c) = i(X,Y,Z,C,1)'.

OK, take me through the gmic-community process. Hopefully, it's not too hard. There's probably more stuff in MATHMAP to convert. Don't know if I am image enough to create anything really new though.

As to the anti-aliasing, my approach is to read from the original image and write to a second as it would trash the original otherwise. So if I were to drop in the pixel with interpolation, wouldn't I only have some other pixels around it for the interpolation calculation. In other words, I would have to build the entire image and then parse it a second time to add anti aliasing? And at that point I might as well use something like a -CONVOLVE or similar. Or am I mis-understanding how the fifth option of I() works.

Souphead.

To add you in the repo, I just need your github account name.
Also, I can have a look at your script to see how interpolation can be done (as well as setting the boundary conditions,
to avoid "black" area in the generated images).
This is something I'll be happy to try.
Maybe it's better if you send me a direct e-mail for these technical details, my e-mail is available on my web page :
https://tschumperle.users.greyc.fr/

Could make the black areas transparent. That what I like to do when using Mathmap.

And here it is! Thanks to Souphead, this popular droste filter is now available for everyone.
Thanks again Martin ! We will enjoy this cool effect !

Thanks folks! Nicely done. I will enjoy this one also.

Now that's and awesome mouth you have there, David. lolol

Rod gave me the head's up. Thanks a heap, Souphead for getting this one ported to G'MIC and of course to David too.

Impresssive; pretty fast as well.

Lyle,
I'm unable to get acceptable results: could you pls post some hints on how to get images like yours? (source image, parameters)
thanks

OK; had to toy with the settings a bit Diego, but here is my original (for the example, above, I blew the image up, but did not do so for the log file) and log file.

And, why not click the show both poles option (only lowrez this time).

And, using the same settings (but unclicked the show two poles) is my extracted Tithonia from a few years back. Just amazed at how fast G'MIC processes these.

Thanks a lot Lyle ! ! ! Saved your preset !
(I still don't understand why filters are released with presets which do nothing acceptable...IMO each released filter should create a nice outcome with defaults, from where to start to personalize, but this is just my opinion...)

I saved it as a favorite too; didn't want to have to re-invent this one again. lol

I think everybody must save your preset for Droste, Lyle! Fantastic!

As an hommage I took that outcome and played with some crystallizing ...

adding a step to restore (some) original color...