Preview of my upcoming denoising filter

I’ve been working on updating my denoising filter and decided to do something a little different and make a video about it.

I have uploaded the filter to my testing folder so it should propagate through the system and be available. It will only work in one of the latest 2.7 pre-releases because it relies on a recent update from @David_Tschumperle


Awesome @Iain!
This shoud really be moved to the Repair/ section, with a warning about the version requirement.
That’s a really nice video you’ve made, if you agree, I’ll make some “noise” ( :wink: ) about it on Twitter and Framasphere.

I might make some changes to it before I move it to the repair section.
I don’t mind you sharing the video if you call it a work in progress.

1 Like

What an excellent filter with an excellent explanation video! :astonished:
When filter comes out, I’m gone for months to clean photos!

I’ll be back when such nice explanation videos for all filters in GMI’C are ready. :grin:

1 Like

I wish you a good retirement in advance! :smile:

1 Like

All, like the universe, is ever expanding. Long live the @s7habo. :rofl:

1 Like

I quite enjoyed making that video so I might make some more videos.


I’d like to see an video comparing the different inpainting filters.

That seems like a really interesting filter!
Could you give details on the algorithms please? The “lookup” parameter makes me think about non local means or bm3d, or something similar?
I also guess you have a scale decomposition, is it with wavelets, or downscaling or something else?
Also what method do you use for detail recovery?
And one last question, what kind of transform is resulting from the shadows and light sliders?
Thanks a lot :slight_smile:

Hi @rawfiner, thanks for your question. I figured I would make a video to explain it so you can see what is going on.


Thanks a lot for taking time to explain all these steps with this video! :slight_smile:
That answered my questions very well!

No problem.
Something I was going to mention but forgot was that the detail recovery method fails if the image has a repeating pattern over a lot of the image. It’s good for removing banding though.

Ok, thank you for this information :slight_smile: