Hi Everyone !
It has been a long time since my latest post concerning the G'MIC project, so I would like here to give some informations on the advances of this project. I don't want to bother you with technical details, I just would like to show an overall view of the improvements.

First, just a word to introduce G'MIC : it is basically an open-source framework to do generic image processing. It includes amongst other things a plug-in for the GIMP. This plug-in offers a growing list of various filters that can be applied on your images (about 150 for now), which are largely customizable. There are filters for artistic purposes, but also for image denoising and sharpening, color correction and so on...

The G'MIC plug-in web-page can be found at :

http://gmic.sourceforge.net/gimp.shtml

There have been a lot of interesting improvement lately, and I released the 1.3.3.3 version of the G'MIC package yesterday. For the plug-in users, here are the main advantages of the new version(compared to the latest one I presented here, i.e. the 1.3.2.9) :

- A high gain in computation speed. There have been important optimizations in the code, which lead to a visible speed-up of G'MIC-based computations. In previous versions, one had often to wait for seconds before the preview is updated when one changes a filter
parameter for instance. Now, this has been really optimized at a point where the preview is updated in less than a second for most of thefilters. This is really really appreciable in terms of user comfort.

- A 'smarter' preview : Each G'MIC filter is now able to set the most adapted zoom factor for the preview image, so that it better corresponds to the final filtered result. For some filters, the default zoom factor would correspond to the entire image (for instance, if you do global deformations on images), for other, the best zoom factor would correspond to a 1:1 pixel correspondance between the preview and the image (all the filters that do local processing, such as the smoothing filters for instance), and finally, others do not require a default zoom factor for the preview (the ones mixing color channels for instance). All this is now handled properly by the G'MIC plug-in. It is particularly interesting for denoising filters, where you can now adjust your parameters once (by looking at your preview window), and obtain the same-looking image when you finally apply the filter. This is particularly true since the preview computation is now faster than before. To me, that is one reason to abandon our previous GREYCstoration denoising filter, since it is not maintained anymore and does not have this essential feature.

- I've opened a "G'MIC" group on the Flickr web site, to discuss about the G'MIC evolution. That's interesting since people can post G'MIC-processed images in the group image pool, and they automatically appear on the G'MIC gallery which shows all these contributions. It gives an idea of all the possible uses of this plug-in. The Flickr gallery page for the G'MIC group is located at :

http://www.flickr.com/groups/gmic/pool/show/

If you have a Flickr account, do not hesitate to join the group to discuss or suggest filters for the G'MIC plug-in (or to post images in the group pool).

- Finally, lot of debugging and code improvements have been done these last months, so the whole thing becomes more and more stable. Also, some new filters have been included.

Well that's it. I really think that all these improvements are important evolutions of the G'MIC plug-in, for a better user experience.

Let me know if you find it useful (or totally useless)

Regards,

David.

I have the previous release installed on a W7 64 with GIMP 2.6.7 and it runs well. Had some head scratching moments trying to figure out where to install everything.

I've used the previous release for denoise and it seems to do a good job while still retaining detail.

Am I imagining that? Or have you actually addressed that issue in your denoising filters?

Well, as I said the improvements have been mainly done for the plug-in interface and speed. The denoising algorithms stay the same, but they are much more easy to use since the preview is really more accurate (so you will probably set the parameters only once), and the parameters are easier to adjust (since the preview is updating faster).