That’ one of the cases where Automatched tone curve in RawTherapee fails because there’s quite a strong vignetting in the embedded jpeg.
I tried to match the JPG but was not really successful. Also, its quite boring. So I did some freestyle 
.
@Entropy512 I may try to get a better pair or better explain but I never seem to quite get the color, tone and for sure the sharpness to reflect that obtained by the JPG. Generally when the light is good etc I am happy with the JPG however I just thought I would be able to do a bit better given all the info in the raw also the phone does a digital crop on zoom so sometime the raw allows me to re think the crop. Likely I really didn’t explain myself. but I have tried many combinations of modules and I never can seem to hit it so that I am at least comparable…Thanks for your comments
Thanks Thomas…I could probably find a better pair of images…I chose it as I liked the tone and sharpness and detail of the jpeg and I was trying to apply a set of modules to produce a result that was at least close…
I love this. I’ve tried recreating a JPG and it’s not as hard as you think. I’ll make a video about it!
1 Like
Thanks for your comments Sebastien. Its just something about the tone and sharpness that I cannot seem to craft back to what the phone makes as a jpg…Likely my lack of ability…
There seems to be a lot of processing going on in your phone with regard to the colors. One might be able to come close but it is not straight forward.
Thanks Thomas…the sharpness is also hard to nail…
Hmm… Yeah, if the phone is doing heavier postprocessing than just a simple tone curve it may be challenging to match it exactly. There’s a possibility it may be doing a more advanced local tonemapping technique (such as RT’s dynamic range compression)
This one has more fine details and less sharpening artifacts when used fill size. I didn’t care much about the noise though…
WP_20190728_17_02_38_Raw__highres.jpg.out.pp3 (12.1 KB)
You need the RT capture sharpening branch for the details
2 Likes
2 Likes
Thanks Ingo. I will look at this I have seen references to the capture sharpening but not sure what exactly it is …I am set up to compile darktable so I will have to see how hard it is to have a go at RT branches…Thanks for your comments
Dave_playraw.NEF.pp3 (11.7 KB)
I don’t think it’s possible to match the colours without an appropriate colour profile
One question: Is the JPEG from a RAW+JPEG combo, or the embedded JPEG within a DNG, or did you happen to change camera modes in between? (Sorry if the former two are the case, I’m kind of in a time crunch at the moment…)
I’m not sure if any mobile cameras which do burst stacking + tonemapping will save a DNG that is the result of the burst stack but prior to demosaicing/tonemapping. I don’t know if the Lumia does burst stacking at all, I know Google does it, and I’m assuming Apple is also doing it “under the hood” in some manner.
This is unfortunately one of the limitations of DNGs out of phones - I don’t know if any phone allows you to save a burst of DNGs or to save an intermediary “stacked but not further processed” DNG (sort-of like HDRMerge but instead of merging a bracket, merging a stack to reduce noise). As a result, often all of the processing in the world won’t be able to match what the phone can output as a JPEG because a single DNG doesn’t actually contain all of the information that might go into a JPEG.
1 Like
@Entropy512 @heckflosse @Thomas_Do Thanks Rico. I had subsequently found some complaints about the colors that come from the lumia 640 being dull and brownish…so I think a color profile would help. My question really stemmed from the fact that many of my photos are family shots and vacation pics and really to process them I hit the JPG’s with dehaze and or the equalizer in DT. From there I tweak it with exposure and WB and they are pleasing to the eye to me. I can save this for a series taken under the same conditions and process quite quickly. However when I get messing around with the raw files from these I can spend a lot of time and throw a lot of modules at the image and still I rarely get one that I say ya that is so much better…now maybe if I was printing them or something I would notice that this was the better route but I think the JPG files are usually closer to the actual conditions most often…alas I appreciate everyone’s comments…
They are shot as JPG and Raw…not extracted JPG’s…
Thanks for your insight…
Without inspecting the raw file, the op jpg looks like it has been processed by a contrast enhancement algorithm. Its signature is a distinct darkening and desaturation seen in the image. Even without the compression and the other baked in processing, contrast enhancement tends to enhance noise and artifacts in the preexisting image; therefore, some sort of denoising must have occurred prior to its application.
The advantage of in-camera processing is that the manufacturer knows the properties of the signal, noise and distortions of the optics and the rest of the hardware, on which they can make informed trade-off decisions. This is the reason that the jpg looks nice overall. Still, raw files give users the power to do as they see fit given the skill and determination, esp. if they don’t like the manufacturer’s signature look.
I started trying to match the jpeg but then I gave up and let it go.
WP_20190728_17_02_38_Raw__highres.dng.xmp (19.1 KB) (DT 2.7)
So, here is my version in darktable 2.7. As stated above, to match the color is difficult. However, I tried to get good sharpness without to many artefacts. Also, in the original JPG there is not much detail in dark areas. I tried to preserve these, too.
WP_20190728_17_02_38_Raw__highres.dng.xmp (4.9 KB)