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

I am not bothering with the 48mp versions... am i missing something?

Only a Year into my Drone life but over 60 as a photographer.
With my Air2 I have used the 48 mp function often with good results, but I occasionally forget and shoot at the 12mp and you really can't see the difference except when "pixel peeping" .
I tend to print my images 13 x 19 to 16 x 20 inches, and will use Topaz Gigapixel to upsize the files (Usually no more than 2X). The prints are quite good, considering the sensor size
 
  • Like
Reactions: PeteGSW
In an initial test with Siemens Star chart I am seeing better results with 48MP than with 12MP. The IQ difference is not that big, but 12MP has strange dot artifacts. Could be a bug.
I would like to see artifacts with 48MP that do not appear with 12MP. I have not noticed them yet.
When comparing noise, remember to resize the 48MP to 12MP before comparing. I have not seen worse noise with 48MP.
My preference is 48MP. That could change if DJI enables real time HDR (a feature of Quad Bayer sensor) for the 12MP mode.
 
It's with JPG. I can see it through the controller - as soon as I switch into 48mp mode the shadows and midtones get darker.

Edit: here you can see the picture get darker in 48mp mode, but is brighter in 4k and 12mp modes. In this video you especially see it in the trees.
That's cause in 48MP quadbayer mode the pixels are effectively 1/4 their size so need more light or longer exposure to compensate. My phone also has a quadbayer 16/64MP mode with the Sony IMX686 sensor. Very gimicky unless you like grainy photos.
 
That's cause in 48MP quadbayer mode the pixels are effectively 1/4 their size so need more light or longer exposure to compensate. My phone also has a quadbayer 16/64MP mode with the Sony IMX686 sensor. Very gimicky unless you like grainy photos.
The same amount of light is hitting the sensor wether it’s it’s one big pixel or a billion small ones so it should not affect the exposure. Same number of photons either way.
 
That's cause in 48MP quadbayer mode the pixels are effectively 1/4 their size so need more light or longer exposure to compensate. My phone also has a quadbayer 16/64MP mode with the Sony IMX686 sensor. Very gimicky unless you like grainy photos.
What brett8833 said. The resolution has little to do with DR/noise when looking at the output level (output of the same resolution). The sensor size matters.
If you look at a 100% view, the noise is more visible with higher resolution. However, that observation has no practical meaning. That is why all DR measurements occur at the output level.
 
Remember when comparing sharpness and noise you need to normalise resolutions to compare like for like.

In other words, the 48mp needs to be reduced to 12mp to judge side by side.
For me its the Bayer artefacts on the 48mp that make it unusable - the same artefacts arent present on the 12mp (lines on the road go purple etc).
 
  • Like
Reactions: globetrotterdrone
Funny for me I find the 48mp images much better compared to resizing the 12mp images in Photoshop ("Preserve Details 2.0" mode).

12mp:
View attachment 151650

48mp:
View attachment 151651

My biggest gripe with the 48mp photos is that they tend to be darker. I find the 12mp usually (but not always) tends to look better out of the box
This example is similar to my experience. The 48mp has artifacts like purple fringing. For me, having RAW is usually more important than megapixels.
 
Remember when comparing sharpness and noise you need to normalise resolutions to compare like for like.

In other words, the 48mp needs to be reduced to 12mp to judge side by side.
For me its the Bayer artefacts on the 48mp that make it unusable - the same artefacts arent present on the 12mp (lines on the road go purple etc).
Does "Remove Chromatic Aberration" help with the purple fringing you observe?
 
Does "Remove Chromatic Aberration" help with the purple fringing you observe?
Not very well as its not chromatic aberration we're seeing. The de-bayer is producing colour shifts where you wouldn't expect to see CA so the eyedropper style fixes cant fully address it.

It helps a little but depending on scene, can still leave lots.
 
After a few test shots with the 48mp compared with the 12mp versions... (plus seeing reviewers and photographers on youtube saying the same) I am now only shooting 12mp jpeg and raw. To me the 48mp images don't look better or indeed worse at times. I found the same on the hi-res settings on an Olympus camera - underwhelming. Maybe for some situations it really works? but I really can't see the point. It seems like a gimmick to me but maybe I am missing something?
Exactly my findings too.
 
Pretty clear to me the 48MP version has more noise. It is also less sharp compared to the 12MP version. Since it is using a "Quad Bayer" sensor it is not a true or native 48MP sensor, which is why it does not look better. I think it is really more of a marketing gimmick than anything.
What do you mean by, "true or native 48MP sensor"? How many photosites are on this sensor, and what is their pattern? How is the color filter constructed? How many each of red, green, and blue filter sites are there, and what is the pattern?

Not trying to be antagonistic here. I'm sincerely very curious. Reason is only sensors I'm aware of universally use a bayer filter to separate color channels and an interpolation algorithm to calculate the other two channels at each pixel location. "True" or "native" w.r.t. resolution are not terms I've encountered in image sensor technology.
 
In an initial test with Siemens Star chart I am seeing better results with 48MP than with 12MP. The IQ difference is not that big, but 12MP has strange dot artifacts. Could be a bug.
I would like to see artifacts with 48MP that do not appear with 12MP. I have not noticed them yet.
Easy.

Go to a straight road with white dashed striping down the middle. Ascend high enough that you can see a dozen or so dashes in the frame diagonally from one corner to the opposite. Snap a 48MP photo.

When viewing it, there will be obvious color errors on the stripes, which should be 100% white, but aren't 😲
 
Only a Year into my Drone life but over 60 as a photographer.
Sincerely confused: Over 60 years as a photographer, or over 60 years old as a photographer?

I suspect the latter but if the former, props man!
 
@Eagle Eye 62 has not clarified whe he means by a "true or native 48MP sensor" vs. a quad-bayer sensor, that presumably produces better images than a 48MP quad-bayer sensor.

Anyone else know what he's talking about?
 
I’m not sure what he means, but put very simply: the Mini 3 has a 48mp sensor. It includes a quad-bayer filter which “bins” pixels 4 to 1. In general, this is done because the 48 mega pixels are tiny and may suffer in low light.

Last I checked however, the Mini 3 will not do RAW at 48mp (probably because file size and processing requirements would be prohibitive). I find that having RAW and dynamic range is more important than having 48mp, so I shoot 12mp.
 
Last I checked however, the Mini 3 will not do RAW at 48mp (probably because file size and processing requirements would be prohibitive).
It can not do 48MP RAW because the colour filters in front of the sensor each covers 4 pixels (or photosites to be precise). And since a RAW file is an unproceesed file "straight from the sensor", it would not be a proper 48 MP RAW-file, but 12.

"A Quad Bayer filter is a bit of a misnomer as it’s actually the same as a regular Bayer filter. What really changes is not the filter but the sensor behind it – these new sensors put four pixels behind each color square instead of just one.
So, really these 48MP Quad Bayer sensors can’t offer much more detail than a 12MP sensor."

The above is a quote from this website: Quad Bayer sensors: what they are and what they are not
 
It can not do 48MP RAW because the colour filters in front of the sensor each covers 4 pixels (or photosites to be precise). And since a RAW file is an unproceesed file "straight from the sensor", it would not be a proper 48 MP RAW-file, but 12.

"A Quad Bayer filter is a bit of a misnomer as it’s actually the same as a regular Bayer filter. What really changes is not the filter but the sensor behind it – these new sensors put four pixels behind each color square instead of just one.
So, really these 48MP Quad Bayer sensors can’t offer much more detail than a 12MP sensor."

The above is a quote from this website: Quad Bayer sensors: what they are and what they are not
Nice bit of clarification. Thanks.

Bottom line, for mightypilot2000 it depends on what you want to achieve. If pure resolution matters, try 48mp, but check your results and decide for yourself what is best. That's what I did and I found 12mp works best for my photography work.
 
  • Like
Reactions: waynorth
I’m not sure what he means, but put very simply: the Mini 3 has a 48mp sensor. It includes a quad-bayer filter which “bins” pixels 4 to 1. In general, this is done because the 48 mega pixels are tiny and may suffer in low light.

Last I checked however, the Mini 3 will not do RAW at 48mp (probably because file size and processing requirements would be prohibitive). I find that having RAW and dynamic range is more important than having 48mp, so I shoot 12mp.
Thanks for taking a stab at it, @abruzzopat. Still unclear about a few things.

Quad-bayer filters act as traditional Bayer-pattern filters when binning (combining) pixels in the high S/N configuration, in the case of the Mini3P, the 12MP mode. The value averaging of each 4-pixel color block is done on-chip, and requires no processing by the GPU. The sensor reads out raw @ 12MP, each pixel location being the 4-diode average for that part of the filter.

This filter pattern is more or less standard Bayer, and the GPU then performs a demosaicing algorithm to interpolate the missing 2 color channels at each pixel. The result is the JPG you download from the drone.

48MP mode is the same with a (big) exception: Neighboring pixels with information on the other two color channels are farther away, resulting in larger errors in interpolation and therefore more artifacts. This results in a loss of resolution where the errors are large, not uniformly across the entire image.

It's important to note that a 12MP image requires demosaicing, introducing interpolation errors, and just like the 48MP image, has artifacts and loss of resolution localized and proportional to the degree of error.

So, the sensor is every bit as "true" a 48MP sensor as a similar sensor with a high-density traditional Bayer pattern.

This is why RAW is so important. Everything's in the demosaicing algorithm. On-board the drone, the processing power is limited, and is doing a lot in real time. Simple neighbor-averaging is about all the GPU can handle in real-time, so the 48MP jpgs off the drone are going to be the worst possible outcome – and are rather good for non-professional applications.

However, taking RAW files from the 48MP Quad-bayer output and applying sophisticated, compute intensive demosaicing processing can produce stunning, near fully-realized resolution, color accuracy, and detail from the very same capture.

I suspect the lack of RAW for the 48MP mode is a file size issue. Each image would output 144MB, a challenge in many ways, not the least of which is simply the time to write to the SD card would complicate many things (burst shots, AEB, etc.).

I wish we had it, though, and I hope it gets added. DJI's customers could really make the Mini3P rock for sills.
 
  • Like
Reactions: scro
Lycus Tech Mavic Air 3 Case

DJI Drone Deals

New Threads

Members online

Forum statistics

Threads
131,616
Messages
1,564,581
Members
160,494
Latest member
tansh