Is it possible to apply a g'mic filter to a batch of images or to video? I've tried Help -> Plugin-Browser, I've tried via File -> Batch Image Manipulation and I've tried figuring it out using the g'mic language tutorial.

This comes up from time-to-time for images, do not know if any video editor implements a gmic plugin. You will probably need a frame-server of some sort.

However, using BIMP for a folder full of images.

Determine the G'mic command line using layer name or logfile logging for a typical image. In linux the log file is in ~/.config/gmic/gimp_log, it will be there in Windows somewhere.


In BIMP search for gmic
put 1 into 'input layers mode
put the command line into the G'MIC command string << much easier to copy paste from the logfile.
then the rest of BIMP is the usual: add images, set output folder etc.



example of the command string


edit: Flowblade Editor uses gmic: https://jliljebl.github.io/flowblade/ linux only
AFAIK limited implementation, no dream_smoothing filter but worth keeping an eye on.

I'm no GMIC expert, but GMIC is first and foremost a command line tool. So if you can do something with GMIC as a Gimp plugin, you can do the same thing on the image using the command line version. And "batching" a command line use is what .BAT files are meant for in Windows.

I think you would have to figure out what functions are called in gmic. Nothing directly applicable that I can see.

Originally a naggobot 'test' filter.

A bit about it down the page: https://www.flickr.com/groups/1316653@N ... 746212046/ lots of anisotropic smoothing.

A real world update.

Dream smoothing does state that it is slow...very, very, slow.

I did not think that the gimp_gmic command would work with standalone gmic but it does. Syntax is straight forward



Unlikely but for a small sample, 10 frames BIMP was marginally faster (48 seconds) than standalone gmic in a bash file (1 min 5 secs). Speed obviously depends on computer power, this using 2.3 GHz laptop and writing to a SSD.

On that basis for anything other than a small clip will take a long, long time. 1 hour plus a bit, for 840 frames. (30 seconds)

Gives this: https://youtu.be/z4r8cP9Phmc