yes.
For those of you who can compile dt master there is a testing PR for xtrans opposed inpainting.
6 Likes
Thanks for your work, I’ll compile and make some tests tomorrow.
inpainting + filmic
inpainting
Here there’s some magenta artifacts in “busy areas”(dunno how to describe) like in the stairway on the water tower.
color
inpainting
color
inpainting (color gave the same result, but it’s an easy image)
5 Likes
Could you share the water tower image? This is exactly a kind of image i don’t have yet.
(very thin dark structure in blown out sections)
Yes - that’s the kind of border issue that was expected. I think we can fix that. Otherwise the color correction seems to be ok,
2 Likes
Yeah, color correction is looking pretty good and very natural looking
1 Like
As the opposed means highlights algorithm now works fine for xtrans I am working on segmentation based recovery for xtrans.
What I need to verify the quality of the algo on xtrans are more images to test.
Come on fuji friends, if you want something really good we need your help!
We need fotos with large parts of the image blown out, like sky and clouds. In bright light and also classic sunset images. Or indoor with arteficial blue or red lights.
Perfect would be images with dark structures in the sky, window frames, power lines, orangeries…
Look out for badly exposed images in your collection and share!
4 Likes
Sunset:
DSCF8597.RAF (21.5 MB)
Blown highlights on textured rock:
DSCF8663.RAF (28.4 MB)
Here are 3 bracketed images that can be useful because the darker ones can give a ‘ground truth’ for the blown highlights:
DSCF7510.RAF (22.7 MB)
DSCF7508.RAF (22.9 MB)
DSCF7509.RAF (24.6 MB)
Sun through lots of trees:
DSCF2362.RAF (27.0 MB)
Above images are licensed under Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International (CC BY-NC-SA 4.0)
2 Likes
The problem I had was finding poorly exposed pictures that were worth saving. 
Here are a few staged ones I just took for you .
LED ‘Edison bulb’. Bulb is ‘4000k’, tiles are white and approximately teal
DSCF0314.RAF (21.3 MB)
Window with latch and screen:
DSCF0322.RAF (27.4 MB)
Complex light fixture. Bulbs are ‘3000k’:
DSCF0323.RAF (23.7 MB)
Above images are licensed under Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International (CC BY-NC-SA 4.0)
Here are a few:
DSCF0075.RAF (26.2 MB)
DSCF0074.RAF (21.2 MB)
DSCF5466.RAF (25.6 MB)
DSCF5377.RAF (26.6 MB)
DSCF5330.RAF (30.1 MB)
I also have a full training set for my LUTs with 80 exposure series. I will make that image set available to you just as soon as it finishes uploading. It is a bit big (4.5 GB).
/Off topic
Ooh, Are you willing share any of your LUTs? I am always looking for some new good ones.
1 Like
Here’s the aforementioned training set: 400 images (4.43 GB). Please let me know when you have downloaded the images, so I can free up the storage. It’s not a very large server.
4 Likes
got it, thanks 
1 Like
Thanks for all the images, especially for the huge set from @bastibe.
All tests confirmed the current code so there is a pr in github 
7 Likes
Can you please share some before/after samples?
Don’t understand this. Before / after what?
The pr has just been updated with some fixes. A am not aware of remaining issues, might also have fixed windows build issues. My tests show equal quality compared to Bayer.
Fuji friends, could you spare some of your precious time and do some hard testing?
Of special interest are
- Transitions from dark to clipped, do pixel peep here!
- Smooth transitions like partly blown out sky.
3 Likes