Today I was working on an improvement for the pixel shift motion mask in RawTherapee.
My work led to this experiment: How many flying birds can you detect in this screenshot?
What is the fifth dot representing?
Which fifth dot?
In (almost) every group there is four dots (obviously one bird in four places) but there is also a single dot nearby (the fifth).
Ok, now I see it. Busy scen. Nice too see that you are still improving the pixel shift functions.
What are you doing with it now?
Currently I try to fix some bugs and improve the monochrome motion mask to better visualize the motion. In current code the mask is not showing all the motion because it’s only black and white, no shades of grey. The new code will have a better mask.
For reference
Nice, I guess that “bug” explains what I saw a week ago when I was playing with a PS image with RT5.4. I was pixel peeping very closely at an old TV antenna on a roof top. And the motion mask had only some pixels on the right side masked, yet the whole antenna were effected somehow when motion correction were enabled compared to when it was disabled.
Yes, that should explain your experience
@Gimbal @nosle Currently only the monochrome motion mask will get the improvements. Do you think, the green motion mask also should be more exact (with shades of green)?
In my opinion the shades of green would be sometimes hardly visible because the background is not black but the image, but I may be wrong.
What’s your opinion as pixelshift users?
If it’s not too much work I would like to have a more exact mask. It doesn’t have to be many shades though. One green for complete substitute and another shade for mixed. (Or is it like that already? I believe I noticed two greens the last time I worked with it and I didn’t understand what it meant at the time.) I’m not at home right now so I can’t double check that.
No, the current green mask is only one bright green, but if you enable sharpening you get more than one green (though the other greens don’t mean anything, as they are only caused by sharpening the preview)
It’s not much work to have a more exact green mask. I’ll have a look how it comes out.
Thank you very much for the feedback 
I improved the green motion mask.
Here’s a screenshot of the old one:
And here’s a screenshot of the new one:
It’s still under development, but I will push soon to pixelshift_fixes branch
I pushed some changes to pixelshift_fixes branch. Motion masks are improved now…
Sweet, will try it out as soon as I get home. (Tomorrow that is.)
I added another feature to my branch. Currently it’s only for tests (to get feedback from pixel shift users):
I extended the Dynamic Profile Rules by an Image Type field:
Currently it’s only an input field, where you can enter for example PS.
Rules marked with Image Type PS are for Pixel Shift files.
For RT 5.5 we will make a better solution (with a combo box to select the image type instead of writing PS for Pixel Shift files).
I would be glad to get feedback from you about this feature.
To be honest I didn’t even know “Dynamic profile rules” existed.
I always start with the neutral profile to get an idea of what I have captured and then I work from there.
Well, you can for example use a profile neutral + Pixel Shift now to automatically use Pixel Shift for PS files instead doing that manually for every PS file you open.
The new Image Type setting for Dynamic profile rules works really well! Nice to see PS spreading it’s tentacles! I use dynamic profiles for my normal files and now the PS ones can join the club.
Thanks!
