I’ve processed the raw file below and am happy with the result. I’ve used DS starting with the watercolour preset then tweaking to taste. I output to jpeg but it’s as if the DS has only partially worked. On my system, the jpeg has less diffusion and in particular, the highlit leaves on the upper right are very differently rendered. Can anyone reproduce please?! Any ideas what is happening…
I’m using Ubuntu 20.04, DT 4.2 Flatpak, and no GPU.
If you zoom in to 100 what I see on my screen matches the detail level of the export… if viewing the Darkroom preview zoomed out to fit it looks more blurred…
I agree. So I’m wondering why the zoomed-out view if so adrift.
If I switch off DS and compare DT’s view with the jpeg by importing the jpeg into DT, they look the same, as per this screen grab of lighttable (jpeg on the right) -
‘high quality resampling’ turned on during export , just to be sure ?
The ds algorithm will produce quite different results when fed with fewer pixels . Maybe it only now is noticeable ? Or maybe there were changes to previewing and ‘region of interest’ code that causes this change …
Remember that, while great care has been taken to make the algorithm’s visual result fairly scale-invariant, the preview will be exact only when zoomed 1:1.
I have indeed with another post…this time is was noise. So the OP looked at the noise reduction show on the screen and was then surprised about all the noise in the jpg output…It was a similar thing…zoom in to 100 and to me they matched but zooming out… it was clean of noise… On my monitor zoomed out to fit is at around 25% so lots of pixel scaling/averaging going on… I can dig up that post if you have any interest in what came of it… I think it was part of a post about noise on a image of sourdough bread and then the image of two men was introduced into the post followed by some discussion of the previews and the different ones used in DT…
Certainly gives different results, if I understand the change correctly… I output my image without the D&S to Tiff, one at full res. (approx 5200x3600) and one at 1300x911. Then applied the D&S to each and output, producing these jpegs -
I think he was referring to the change made by AP in ANSEL… he has exposed the scaling module so it can be moved in the pipe to alter when scaling happens and what happens after it… in DT I think it comes after the output profile which I think has been determined to be not the best place…
