DJI Mavic, Air and Mini Drones
Friendly, Helpful & Knowledgeable Community
Join Us Now

Mavic 2 Pro Photography – HDR

You must be looking at JPEG images, which are always 8 bit. RAW is going to be 16 bit but, as I mentioned, that's not what defines dynamic range - you just need enough bits to record the sensor resolution of its dynamic range.
There's no escaping the fact that the larger the pixels, the better the dynamic range. Large sensors, large pixels, small sensors, small pixels. Large sensors, good dynamic range, small sensors, Ughhh!

The Hasselblad Hasselblad H6D-400c Multi-Shot has a very large sensor with pixels the size of you little finger nail (maybe not quite) but It's 400MP sensor shoots 2.4 GB files and boasts an incredible dynamic range allowing for unprecedented detail capture – from the deepest shadows to the brightest highlights.
 
Last edited:
RAW images from the Mavic 2 Pro are 16 bit, not 10 bit (that's video), but it's the dynamic range that is the limiting factor, not the bit depth. The Sony sensor has a dynamic range of about 12 EV at ISO 100. The JPEG mapping to 8 bits will potentially miss detail in the highlights and shadows, so you can independently tone map a RAW image to pull those out. Every stop change in an HDR RAW sequence adds 1 EV to the dynamic range over and above the sensor's 12 EV.

There is no way Mavic 2 Pro RAWs are true 16bit - that Sony sensor maxes out at 12bit readout if it's the sensor model I think it is. At an absolute maximum it would be capable of 14bit and I am quite sure it isn't doing that. Also DR does get a measurable improvement with increased bit depth, all else equal.

There's no escaping the fact that the larger the pixels, the better the dynamic range. Large sensors, large pixels, small sensors, small pixels. Large sensors, good dynamic range, small sensors, Ughhh!

The Hasselblad Hasselblad H6D-400c Multi-Shot has a very large sensor with pixels the size of you little finger nail (maybe not quite) but It's 400MP sensor shoots 2.4 GB files and boasts an incredible dynamic range allowing for unprecedented detail capture – from the deepest shadows to the brightest highlights.

Pixel size has no direct correlation to DR. Lots of sensors with smaller pixels have way better DR than sensors with larger pixels. That 1" Sony sensor (equivalent to approximately a 54MP DSLR) has better DR than some full frame DSLRs, such as much of what Canon was producing until recently. There are also lots of medium format cameras that do not have better DR than a full frame DSLR with a much smaller sensor.

The sensor technology, sensor tuning, and overall light collection ability, as well as how the ADC is done has the biggest effect. Also, the more room there is on the sensor wafer to get the electronics out of the way, the better - that is a problem with the small sensors, which is why it's the really small sensors that see much higher performance benefits from BSI and Stacked designs.

For example a 46MP Nikon D850 has some of the highest DR of any camera, and a 20MP Nikon D5 with much larger pixels has way worse base ISO DR because it is tuned for high ISO performance.

Also that Hasselblad H6D-400C has pixels orders of magnitude smaller than anyone's fingernails :) The pixel pitch is about 2.4um (a micrometer is 1000 times smaller than a millimeter).


The only way mirror vibration is affecting sharpness is if you are using a very long lens.

Not really an issue for most drones, but It depends on the camera. Mirror vibration almost always plays a factor. It's also not limited to long lenses - macro photography is even more sensitive to it, and really any time you need absolute critical sharpness - even with landscape shots the difference is visible depending on the camera. Electronic shutters are one way around this, and some DSLRs now have electronic first curtain shutters which offer huge improvements for critical work. The other thing many high end camera manufacturers are doing now is using very specialized actuators that gradually accelerate and decelerate the mirror mechanism to reduce vibrations.
 
Last edited:
Also that Hasselblad H6D-400C has pixels orders of magnitude smaller than anyone's fingernails :) The pixel pitch is about 2.4um (a micrometer is 1000 times smaller than a millimeter).

I was being a tad cheeky ;) and yes I agree with what you say but there's no denying that ('as a general rule') larger sensors have larger pixels and offer better DR.
 
There is no way Mavic 2 Pro RAWs are true 16bit - that Sony sensor maxes out at 12bit readout if it's the sensor model I think it is. At an absolute maximum it would be capable of 14bit and I am quite sure it isn't doing that. Also DR does get a measurable improvement with increased bit depth, all else equal.

They are 16 bit files. That has nothing to do with the actual dynamic range. As I pointed out, and you repeated for some reason, that's around 12 EV for that sensor.
 
Do you see much difference with the extra shots? Are you stepping them in values of 1 +/- ?
 
Yes - but the misalignment is fixed in the stack processing - it's the very first step. What kind of software are you using? This is a complete non-issue for most HDR work. If your concern is that the vibration from the shutter is moving the camera between shots (but somehow not motion blurring the individual images?) then that concern is misplaced. Far greater effects arise from movement in the view - wind motion of leaves and branches, objects in motion etc., and the software will use various methods to try to identify and remove the resulting ghosting. I've no clue how you think that you have identified this as an issue, but it isn't.

sar here are three photos taken of the same scene, same focal length/point, same ISO=800 and aperture of F8 though: 1 of them was a single shot, another was bracketed in post-production, and another bracketed as normal HDR. I used 2 steps for bracketing [-2, 0, +2]. without reading the file can you tell me which one looks better?

Thanks,

Grons
 

Attachments

  • 1.jpg
    1.jpg
    1.5 MB · Views: 52
  • 2.jpg
    2.jpg
    1.5 MB · Views: 50
  • 3.jpg
    3.jpg
    1.4 MB · Views: 51
sar here are three photos taken of the same scene, same focal length/point, same ISO=800 and aperture of F8 though: 1 of them was a single shot, another was bracketed in post-production, and another bracketed as normal HDR. I used 2 steps for bracketing [-2, 0, +2]. without reading the file can you tell me which one looks better?

Thanks,

Grons
IMO (and I realise you didn't ask for it) the second image is superior and without looking at EXIF and pixel peeping I'd bet it was bracketed in post production.
 
The only way mirror vibration is affecting sharpness is if you are using a very long lens. Anyway - either the individual images are sharp or they are not. If they are not sharp then the problem has nothing to do with HDR. If they are sharp then post-processing isn't going to make them sharper, but it is going to align them if there was camera movement between the exposures. That movement doesn't matter unless it is large - you can take perfectly good bracketed shots handheld and let software alignment take care of it - no tripod required.
I thought I posted my 3 shots on this thread; pardon me if I did but here are 3 shots; one is a single RAW processed file, another is a HDR file and shot as bracketed at [-2,0,+2] and another was bracketed in post-production. Which is the best and can you guess which ones are which? All settings are the same, with the exception of the shutter settings between the HDR as shot and the HDR bracketed in post-processing.
 

Attachments

  • 1.jpg
    1.jpg
    1.5 MB · Views: 28
  • 2.jpg
    2.jpg
    1.5 MB · Views: 25
  • 3.jpg
    3.jpg
    1.4 MB · Views: 26
IMO (and I realise you didn't ask for it) the second image is superior and without looking at EXIF and pixel peeping I'd bet it was bracketed in post production.

Noosaguy I believe number 2 was bracketed in post-production. But as always I screwed up and forgot to write the order down. The 3rd shot was the single photo processed in the HDR converter, and I believe the 1st shot was truly bracketed as shot.

For my purposes there is very little difference among them however, the drone shots are much noisier and the method of processing my indeed have a greater impact on the pics!

Thanks for the comments.
 
  • Like
Reactions: noosaguy
IMO (and I realise you didn't ask for it) the second image is superior and without looking at EXIF and pixel peeping I'd bet it was bracketed in post production.
Bingo you got it! I took the photo with an exposure value of 0 and created 2 copies one at +2 EV and the other at -2 EV then merged them. Good eye!
 
  • Like
Reactions: noosaguy
Noosaguy I believe number 2 was bracketed in post-production. But as always I screwed up and forgot to write the order down. The 3rd shot was the single photo processed in the HDR converter, and I believe the 1st shot was truly bracketed as shot.

For my purposes there is very little difference among them however, the drone shots are much noisier and the method of processing my indeed have a greater impact on the pics!

Thanks for the comments.

Looking at the 3 images I would have picked the first as the single image, the second RAW post processed and the third bracketed in camera. That conclusion was based on shadow detail alone. I'm not sure there's not much between them though because DR is better in No 3 and better still in No 2. If you tried to drag out shadow detail in No 1 I reckon you'd get noise.

Where are you Gronz?
 
Bingo you got it! I took the photo with an exposure value of 0 and created 2 copies one at +2 EV and the other at -2 EV then merged them. Good eye!

Why on earth would you do that - it doesn't generate any extra information? Why not just map the original raw image down to 8-bit JPEG to get the result that you want? This is not HDR.
 
Looking at the 3 images I would have picked the first as the single image, the second RAW post processed and the third bracketed in camera. That conclusion was based on shadow detail alone. I'm not sure there's not much between them though because DR is better in No 3 and better still in No 2. If you tried to drag out shadow detail in No 1 I reckon you'd get noise.

Where are you Gronz?

The first was bracketed in the camera, the second was bracketed in post-processing and the third was the single shot. I'm in sunny AZ where your cell phone shuts off after flying your drone for 5 minutes during the summer. I have to bring frozen gel packs and hold them against the phone to keep it on during the summer...
 
  • Like
Reactions: noosaguy
sar here are three photos taken of the same scene, same focal length/point, same ISO=800 and aperture of F8 though: 1 of them was a single shot, another was bracketed in post-production, and another bracketed as normal HDR. I used 2 steps for bracketing [-2, 0, +2]. without reading the file can you tell me which one looks better?

Thanks,

Grons

I'm really confused - what are you trying to demonstrate? I thought the issue you were worrying about was loss of sharpness due to movement between multiple bracketed exposures. Apart from anything else, the resolution of these images is clearly limited by the downsampling - they are far too low resolution to conclude anything. And only one of them is actually constructed as an HDR from multiple exposures.
 
Why on earth would you do that - it doesn't generate any extra information? Why not just map the original raw image down to 8-bit JPEG to get the result that you want? This is not HDR.

Why? Because the clarity is much better from a drone with regards the movement between the three shots is enough to make the detail fuzzy because ALL post processors are not 100% full proof when stacking images and each image will vary a tad. Secondly you probably notices no significant differences in the photos, so why take the chances of a blur caused by movement in the lens as it takes 3 shots? Oh well, to each his own. Generally exterior photos seldom need to go through an HDR processor and can be done in the HDR software. If camera movement was not significant, why does some guys spend $hundreds on their tripods? I've been shooting HDR for 4 years now and often have to go back to a single shot, if the HDR shot loses it's sharpness due to a leg sliding on the tripod, due to tile floors, or the fixings not tightened all the way down.
 
I'm really confused - what are you trying to demonstrate? I thought the issue you were worrying about was loss of sharpness due to movement between multiple bracketed exposures. Apart from anything else, the resolution of these images is clearly limited by the downsampling - they are far too low resolution to conclude anything. And only one of them is actually constructed as an HDR from multiple exposures.

Yes and you can't tell the difference among the 3, which was one of my points in that HDR is often not necessary for outdoor shots. Again, I don't care how great the HDR processor is, when you take 3 shots from a drone there is going to be movement from shot to shot, and this movement is not in sync to allow the software to stack one on top of the other.

Anyways, I enjoyed the chat,

Peace brother,

JG
 
Why? Because the clarity is much better from a drone with regards the movement between the three shots is enough to make the detail fuzzy because ALL post processors are not 100% full proof when stacking images and each image will vary a tad. Secondly you probably notices no significant differences in the photos, so why take the chances of a blur caused by movement in the lens as it takes 3 shots? Oh well, to each his own. Generally exterior photos seldom need to go through an HDR processor and can be done in the HDR software. If camera movement was not significant, why does some guys spend $hundreds on their tripods? I've been shooting HDR for 4 years now and often have to go back to a single shot, if the HDR shot loses it's sharpness due to a leg sliding on the tripod, due to tile floors, or the fixings not tightened all the way down.

Real HDR (i.e. higher dynamic range than one exposure provides) is, by definition, only achievable by multiple exposures. Taking one 12 EV dynamic range raw image, processing it to get three differently exposed 8-bit JPEGs, and then merging those JPEGs to get an HDR is completely pointless. It's not HDR. You gain nothing at all in terms of dynamic range over the original raw image, which you could have mapped any way that you wanted to the final 8-bit image.

And in terms of your concern about loss of sharpness due to merging multiple exposures, you didn't address that at all with your examples. As I said before, ghosting from movement of things in the view are far more difficult to fix in the HDR processing than pixel or sub-pixel offsets. The deliberate use of sub-pixel offsets is actually a well-established technique to improve resolution. All HDR stacking software that I've used aligns the images well enough that it is not the resolution-limiting factor.
 
Real HDR (i.e. higher dynamic range than one exposure provides) is, by definition, only achievable by multiple exposures. Taking one 12 EV dynamic range raw image, processing it to get three differently exposed 8-bit JPEGs, and then merging those JPEGs to get an HDR is completely pointless. It's not HDR. You gain nothing at all in terms of dynamic range over the original raw image, which you could have mapped any way that you wanted to the final 8-bit image.

And in terms of your concern about loss of sharpness due to merging multiple exposures, you didn't address that at all with your examples. As I said before, ghosting from movement of things in the view are far more difficult to fix in the HDR processing than pixel or sub-pixel offsets. The deliberate use of sub-pixel offsets is actually a well-established technique to improve resolution. All HDR stacking software that I've used aligns the images well enough that it is not the resolution-limiting factor.

I merged 3 RAW files, and converted to Jpg in order to post. Again, you have to shoot the way you wish to. Most photographers believe that bracketing 7 exposure values is uncalled for. To each your own.

I don’t respond to threads to argue or debate but rather to share information based on my experiences to edify others. Good day, signing off this thread.

Good evening gentlemen
 
Lycus Tech Mavic Air 3 Case

DJI Drone Deals

New Threads

Forum statistics

Threads
131,123
Messages
1,560,064
Members
160,099
Latest member
tflys78