Inspired by a thread from earlier this month, I decided to do some camera testing.

A user has claimed that the D-Log is baked into your RAW/DNG photos. I finaly found the time and weather to test the drone/camera with our current firmware (V01.03.0600).

The good news is that D-Log is NOT baked into your DNG files (at least not in my case). The bad news is, there's still something going wrong when shooting photos with cameraprofile set to D-Log...

I did a JPG+DNG shot with every profile selected, and generally the DNGs looked the same, completely independent of profile selected. But there was one strange exception. The version when D-Log was selected...

Apparently the camera with D-Log selected is underexposing approx 1.5 stop compared to shooting photos with another profile selected!

I have made an overview with DNG+JPG pairs of the first 6 cameraprofiles:

When D-Log is selected, the camera underexposes your photos

During my testing I discovered another thing. I have heard people talking about a "watercolor painting effect", but never noticed anything like that on my own photos. Apparently that is because I have been shooting in RAW/DNG mode. Because at a closer look, the JPGs here are all REALLY  bad. It is especially visible in the green trees and bushes behind me, but generally details are few in the jpgs:

JPG vs DNG. Really bad "watercolor effect" in JPG.

I have decreased highlights in the DNG file a tiny bit, but besides that, it is default settings in Adobe Camera Raw.

Camera has sharpening set to -1. I have read that setting sharpening to +1 should decrease noise-reduction and remove the watercolor effect, but that is not a solution I would be satisfied with if I was shooting JPGs. Default sharpening is already rather high in my opinion.

CONCLUSION: Shot RAW/DNG and avoid using D-Log when taking photos.

BTW, cameraprofiles for video and photography/jpgs are not independent. Neither is for example whitebalance which I prefer fixed in video (to avoid it changing in the middle of a recording), but prefer in Auto for photography.
Please DJI, (also) try to fix that...

Interesting. Thanks for doing and reporting on your testing.
Regarding watercolor effect, 0 sharpening usually is sufficient in good light in video, in lower light you might have to try +1.

Mine is set to +1 0 0 works well for me

Interesting, I was indeed confused with my stills (they looked washed) I'll try to disable the color profile and set it back to none before taking pictures! Thanks for the test!

I honestly think however that these color profiles for video should be independent if taking stills... it doesn't make sense to be changing it all the time!

I can add that the exposure-time (auto-exposure) for all the cameraprofile test-shots are in the range 1/850-1/950sec., except for the D-Log where it suddenly jumped to 1/2400sec.  (All shots at 100 ISO and with the fixed aperture of f/2.2).
A coincidence and just the auto-exposure that accidentally got confused exactly when trying the D-Log cameraprofile? Unlikely I would say, but I will verify it the next time I find the time and weather, unless others can verify the behavior before me...

I suppose if we establish this is happening it's not too difficult to increase exposure when taking stills.
Also using RAW you can change exposure at edit stage significantly more than jpg with no bad effect.

4wd Posted at 2017-5-4 10:11
I suppose if we establish this is happening it's not too difficult to increase exposure when taking stills.
Also using RAW you can change exposure at edit stage significantly more than jpg with no bad effect.

Regarding doing post-corrections:
Even if RAW have more "room" for adjustments, you need correctly exposured photos for optimal quality. When pushing the exposure in post-processing you increase the noise and you won't see the same level of details in the shadows as on a correctly exposured photo. This is also true for RAW.

Regarding increasing exposure when taking a photo:
This is almost as easy/difficult/annoying as changing the profile before you take a photo.... And remember to switch back if you do D-log video afterwards...

Btw, personally I'm not using D-log (at least not yet, I need more experience in other areas of video before experimenting with that). I'm just reporting the problem I see.
But I really think that settings for video and photography should be independent as long as possible. Including cameraprofile and whitebalance settings.

Glad i found this thread....
Ive had the drone about a week, recently switched to D-Log for video.  Reviewing my RAW photos (i only ever shoot RAW, not JPG.  I wouldnt have bought the drone without RAW) they were all underexposed by about 1.3 stops.  I initially put this down to human error and me screwing up a manual setting but im now 100% convinced its not.
DNG raw is simple not RAW is D-log is selected.  It is performing post processing to the image (and is underexposing resulting it noise when you bring it back up).
I can see it being really annoying filming video with D-log and the next minute having to remember to change a camera setting to shoot a raw photo of the same subject - its clearly a bug.  A far better solution would be completely independent settings for camera mode vs video mode.
I may have to stick with Cinelike for video purely so i dont forget when i want to take a photo which isn't ideal.

Was this addressed in DJI Go 4.1.3?

Ex Machina Posted at 2017-6-29 17:31
Was this addressed in DJI Go 4.1.3?

I tried to test this last night but found with 4.1.3 I could no longer open DNG. sigh.

I did notice different apparent exposures with D-Log and other color profiles, but wonder if the exposure variance actually appear in RAW files rather than just the embedded jpg preview in the DNG. Occurred to me because obviously a B&W RAW file is only going to be a valid file coming from a dedicated B&W camera -- the Mavic's embedded preview might be B&W but the RAW data should be full-spectrum.

Of course, until I can once again open DNG I won't be able to test this.

I tried to test this last night but found with 4.1.3 I could no longer open DNG. sigh.
I did notice different apparent exposures with D-Log and other color profiles, but wonder if the exposure variance actually appear in RAW files rather than just the embedded jpg preview in the DNG. Occurred to me because obviously a B&W RAW file is only going to be a valid file coming from a dedicated B&W camera -- the Mavic's embedded preview might be B&W but the RAW data should be full-spectrum.

Got my Mavic able to save RAW files again. Confirming that RAW files are underexposed when using the D-Log profile, not just the embedded preview jpg. The first example is using TrueColor, the second D-Log.  
  

Ive done some more tests and can confirm that sharpening and noise reduction has NO effect on the DNG (which is good) on any profile setting.
The D-log DNGs are under exposed by 1.5 stops for sure though.  BUT its the same exposure settings used for the image as cinelike so its actual software processing causing the under exposure not the actual camera.  In other words, take 2 photos, 1 in cinelike, the other in d-log with identical camera settings.  The Cinelike will be exposed correctly, the D-log will be under exposed by maybe 1.3 stops from that  
This means a raw file under d-log isn't raw at all.
Havent tried 4.1.3 at all yet but i cant see that making any difference at all - this is firmware on the bird itself.  The RC sends take photo command and the mavic processes and saves it on the SD card.

For now my only work around is to not use D-log (i really dont want to remember to change profiles every time i go for a still shot vs a video).

Ex Machina Posted at 2017-7-1 15:31
Got my Mavic able to save RAW files again. Confirming that RAW files are underexposed when using the D-Log profile, not just the embedded preview jpg. The first example is using TrueColor, the second D-Log.  
  [view_image] [view_image]

Now that I re-read StigNygaard's original post, I see he could not have been confusing DNG preview with the RAW data. ;)

I recently watched a YouTube video about shooting D-Log video. The guy said that D-Log can handle overexposing by 2 stops! After reading all here I started to think like if you set it to D-Log and leave it for both video and photo shooting you should at least overexpose any of them by 1.5stops. Only then you can get the most dynamic range and the less noise I guess. I will test this and let you know asap.