vkdt, did not spend time on the rest of the image, just to show the highlights. i think rawtherapee looks really good here and the filmulator results is excellent due to a lot of diffusion i assume 
time spent in full res highlight reconstruction on a gtx1650 max-q laptop: 9.893 ms.
Looks very naturalā¦
I love DT and for most images find the editing capabilities far better than LR. However, I have thrown this image into Dt and then LR and concede for this image the Highlights were handled better by LR just sliding the highlights recovery to the extreme. I do note that a DNG file is supplied and wonder if the RAW file instead of Adobeās DNG file format might work better with DT. Probably wouldnāt.
I would suggest with any image use the best tool for the job. With this image it would be LR. However, donāt judge DT on this single image. DT has far more artistic control than LR. I still use LR for panorama stitching and exposure merging of bracketed shots. Bracketed exposure would work well with image as the bright section of the sky needed far less exposure than the foreground.
Itās a Pentax, they can write DNG natively.
I doubt that shooting in PEF format would make a difference. However, I own a couple of Pentax cameras and for myself have chosen to shoot in PEF format so when I look back over image files in my catalogues I know which were taken with the Pentax gear. Personal taste, but with a DNG file I am unsure which camera it is or if it was created by another program such as LR or Adobe DNG converter. Adobe DNG converter helps with camera files not supported by LR6, DT or other programs.
That DNG file was written by the camera itself. No conversion (Itās possible that LR added some XMP data because here I prefer the single file approach without additional sidecar files - but the image data is untouched).
In my early days I used PEF but switched to DNG very soon because of the better interoperability. When I switched, I did a couple of comparisons and could not see any difference in image quality (nor noticable in overall file size). But to be honest, I did no comparison with my newer cameras that are able to capture way more dynamic range than the old K10D.
From what Iāve seen so far from the dt developments, the guided laplacian seems to be ideal for such images. I couldnāt get it to run by myself so far. Either I used the wrong branch or did something wrong. But Iām looking forward to test it.
And again: Many thanks to everyone who plays around with my image. Itās incredible to see how different the approaches and results are.
When shooting with high iso, you loose rapidly a lot of quality (tones, dynamic, noise). I never shoot over 320 iso (specific to Canon) & 200 iso with Phaseone. LR is easy to use, Dt require some knowledge . Lr will be better for people without knowledge.
ā¦if you want to go old school with dt, this is probably a case for ulrichās colour reconstruction module:
doesnāt blend in quite so smoothly as the better examples weāve seen above in this thread, and requires a bit of fiddling with the parameters. (and man, dt takes time to process)
@heckflosse are you sure, you uploaded the corresponding pp3? With opening your pp3, i find an almost neutral profileā¦
Damn, youāre right. I uploaded a wrong .pp3 
I will post the (hopefully) correct .pp3 in my original post nowā¦
Does the LR highlights recovery have tone-mapping in it ? Because if so, thatās not only highlights recovery anymore, itās a local contrast enhancement. Recovery is the process of trying to fill-in the clipped pixels with a value that makes sense. Thatās how open-source does it. If LR is doing more than one thing at a time, then you compare apples to oranges.
And automagically doing 2-in-1 does sound like Adobe, but itās not the FLOSS approach.
@heckflosse thanks, now its correctā¦ great edit
Avec Art + un petit passage sous Gimp pour lisser les bords du nuage un peu plus clair Ć gauche.
āWith Art + a small passage under Gimp to smooth the edges of the cloud a little lighter on the left.ā
Iām an amateur photographer, great fan of DT since a few years.
I use DT both as a digital asset mgt tool and as a software development tool for my RAW pictures (EOS 1300D). My library contains about 70K of pictures. I shoot mainly portraits and landscapes.
Like 0x4243, Iām sometimes also facing issues to recover shadows and highlights of landscape photos.
In this thread, containing many DT xmp files, I could see interesting trials. Unfortunately, I couldnāt really find any satisfactory end results. To my opinion, most results are too dull. I donāt want to hurt anyone.
Does this mean there is no way for DT to process this picture ?
Could someone help me out of this ?
PS : Iām nor a data scientist, nor a programmer, Iām just an amateur trying to get the most out of DT.
It looks like @anon41087856 has something promising cooked up for future versions.
I think that most of us concentrate mostly in highlights recovering than getting a perfect picture. 
Did you calibrate your screen? The image was taken 1st week in september. The sun shall rise one hour later. Itās still dark, & colours are dull.
No, but, 0x4243ās LR result looks not ādullā on my same screen. So I donāt think itās a calibration issue. Anyhow, Iām facing this ādullā issue on a regular basis with my landscape pictures as well.
