I think that is a internal interpolation from the original 12MP Picture.
In a silmple side by side test you see that a 12MP in ACR interpolated to 48MP DNG looks better than the new 48MP picture.

A made such a simple test and you can see if you zoom in, that the DJI interpolating artefacts are looking much more present and destructive than the ACR one.
Bad and useless!

Thanks and regards

Yeah, I wouldn't buy this drone if looking for 48 megapixels.  It is the 4k 60p and the HDR video that is awesome at that price.

No point in closing the barn door after the horse has bolted, did you really think you were buying a full 48mp camera on a drone with such a tiny sensor considering Fuji sells it’s 50mp camera without a drone for 10k , you’ve made a serious discovery here.

I wouldn't say that it is a fake 48mp resolution but reading this might shine some light on your experience.

https://www.gsmarena.com/quad_ba ... ined-news-37459.php

But I think it arguable and I don't notice as much improvement myself, then again a photo is looked upon from corner to corner and not zoomed in ... but that is a completely other discussion.

I knew what I was getting after looking at the output of the EVO II 8K which uses the same camera.  Needless to say, it is still an improvement over the old 1/2.3 sensors that most birds in this price class use.

BobWinNV Posted at 5-17 13:10
I knew what I was getting after looking at the output of the EVO II 8K which uses the same camera.  Needless to say, it is still an improvement over the old 1/2.3 sensors that most birds in this price class use.

From what I've seen of that 1/2 sensor, completely agree.

BobWinNV Posted at 5-17 13:10
I knew what I was getting after looking at the output of the EVO II 8K which uses the same camera.  Needless to say, it is still an improvement over the old 1/2.3 sensors that most birds in this price class use.

I fully agree and I’m quite impressed with it, as they say cameras make files, photographers make pictures.

It’s pointless to look at the 48mp quality in anything but plenty of light. That’s the whole point of this quad bayer tech, it can only effectively shoot 48mp when there’s so much light that the tiny pixels are not a huge drawback anymore. If you force it into 48mp in low light, it’ll be crap.

so this camera is no better than the original mavic pro. That sucks. I did my research, but all youtube unboxers said it was indeed a 45 mpx camera.  I could happen. But it didn't.

djiuser_M3tUeKlVTQn7 Posted at 5-17 15:17
so this camera is no better than the original mavic pro. That sucks. I did my research, but all youtube unboxers said it was indeed a 45 mpx camera.  I could happen. But it didn't.

No it has a much better camera than M1.

Thank you for posting sample photos.

A while back I tried to find out more about the 48MP mode of MA2.  The assumption is that it has the Sony IMX586 quad bayer sensor.  So I read Sony's press release and pixel peeped the demonstration photos which showed maybe marginally better detail and obvious sharpening applied.  But I also saw an amazing zooming example of a 48MP photo from the MA2 from a Youtuber.  Now I'm thinking this amazing zoom detail photo was not taken with a MA2.  I'll look at it again later.


But I would still like to see more comparason shots of 12MP and 48MP mode.  Ideally the shots would be ithe same well-lit scene.

Thanks again.  This is an interesting topic for me.

Actually it's not fake at all, you just need good lighting conditions for 48MP to shine. I took this yesterday over the empty Hollywood Bowl and it was when the whole bowl was in a shadow at sunset so lighting wasn't optimal but you can see the difference between the 12MP (left) and the 48MP (right) when they are both at the same crop size. The detail in the 48 is much higher and sharper than the 12. Your example is very low lighting and looks to be quite upclose. Try a real world scenario, particularly in bright conditions and you'll see the difference for sure.

Links to the two full size JPGs so you can inspect them yourself under the thumbnail



12MP https://ibb.co/hcHMj46
48MP https://ibb.co/8dWrT5t

If any one could post comparison 100% photos of same scene taken with Mini, Air 2 12mb, Air 2 48mb and MP2 would be so useful.  Thanks

melodicskys Posted at 5-17 16:49
Links to the two full size JPGs so you can inspect them yourself under the thumbnail

[view_image]

Great example and nice improvement in detail.  Thanks!

The Duck Posted at 5-17 17:51
Great example and nice improvement in detail.  Thanks!

Glad it helped! The 48MP can show more noise in the shadows, but when there's enough light the detail improvement is worth using it.

first of all, it isnt true true 48MP if what you count as true 48MP is sampling sensor as it is 1x for a single shot..  infact the 48MP images are made up of 2x 24MP images captured on same sensor thanks to way quad bayer filtering works.  2x sampling more sampling on each axis.   

there is benefits and drawbacks to this..    benefits are more flexibility with sampling methods and computational photography techniques way more then a sensor just straight up 12MP or 48MP..  downsides though because of the nature of quad layering, the results may lack accuracy and reference quality also likely will struggle in very tighter detailed and highly contracted situations where got of dark darks and bright brights between small sensor and filtering tech, mavic 2 pro would handle HDR situation better, i mean mavic air 2 te shadows already fall off a cliff in HDR while mavic 2 pro's shadows are very stable.  doesn't mean it less valuable drone and camera arrangement by any means, still a great price for all the camera and flight tech offered..  my only 2 gripes about mavic air 2 is weird controller vs previous mavics and lack os side and top sensors..could have at-least added IR based sensor on top and sides for little more protection but thats knit picking, prob one of the first drones that thus far i dont really have much to b*tch about..

HereForTheBeer Posted at 5-17 20:21
first of all, it isnt true true 48MP if what you count as true 48MP is sampling sensor as it is 1x for a single shot..  infact the 48MP images are made up of 2x 24MP images captured on same sensor thanks to way quad bayer filtering works.  2x sampling more sampling on each axis.   

there is benefits and drawbacks to this..    benefits are more flexibility with sampling methods and computational photography techniques way more then a sensor just straight up 12MP or 48MP..  downsides though because of the nature of quad layering, the results may lack accuracy and reference quality also likely will struggle in very tighter detailed and highly contracted situations where got of dark darks and bright brights between small sensor and filtering tech, mavic 2 pro would handle HDR situation better, i mean mavic air 2 te shadows already fall off a cliff in HDR while mavic 2 pro's shadows are very stable.  doesn't mean it less valuable drone and camera arrangement by any means, still a great price for all the camera and flight tech offered..  my only 2 gripes about mavic air 2 is weird controller vs previous mavics and lack os side and top sensors..could have at-least added IR based sensor on top and sides for little more protection but thats knit picking, prob one of the first drones that thus far i dont really have much to b*tch about..

A massive print, also a very honest review and certainly for me he gets it just about right.

Hy guys, thank you for the big repost.
I know what you talking about, i am a photographer since 25years, and what i not mean, is a comparison between both resolution as a jpg, while jpg has a huge postprocessing in the Drone itself.

You must have to look in a non processed DNG.
The tip with the light is nice, but this shot is in direct sunlight with shadow behind, so there is enough light on the flower.

The other thing is, that you have to zoom in, while the cam has the nearest focus distance at one meter, so something that are in focus, is whide away.
The point is also not the sharpness, it is only the interpolation artefact, they are more visible and stronger in the 48mp pic.

In summary, for us pros or semis this goody makes no sense, while the native 12mp pic is much cleaner and has no internal interpolation process.
If i use this in Photoshop to upscale to 48mp, the result is a better picture, and this has nothing to do with light, while the artefacts are not from noise, its causing from interpolation.
So a shot in the dark, has the same interpolation artefacts, but plus chroma and luma noise.

I don`t say that it is not a good function for some, but in the end it is only a tiny process with a huge marketing hype.
Regards and stay healthy. :-)

hallmark007 Posted at 5-17 22:07
A massive print, also a very honest review and certainly for me he gets it just about right.

OK!!
And now whats good on the man with a short chair?

What a useless comparison with 2 different resolutions in print.
It is clear that a greater print resolution gives a better and greater print, but thats not the case, if he interpolates the 12mp, the print was cleaner than his org 48mp print.

And all other points in this video are reading facts, not more, nothing special, no looking in depth from own tests.

I can´t see something that makes me feel better after looking this short man. ;-)

Because no one of you clicked on the pics to zoom in and see the artefacts, i do it for you, so you clearly see the difference!

Lookup quad bayer sensor.  It should explain it.

DAFlys Posted at 5-18 01:02
Lookup quad bayer sensor.  It should explain it.

thats not the case, the bayer sensor is in both pictures the same. :-=
It is nice to see that you know something about sensors, but that is not the case here.

MavGuido Posted at 5-18 01:08
thats not the case, the bayer sensor is in both pictures the same. :-=
It is nice to see that you know something about sensors, but that is not the case here.

But the data is sampled differently between 12mp and 48mp.

MavGuido Posted at 5-17 23:33
OK!!
And now whats good on the man with a short chair?

He’s looking for a pleasing picture to look at, you’re treating this like an exam, that’s the difference, he’s a photographer, you’re school teacher.
Photography is about pictures not pixels. The one thing you seemed interested in, you completely botched. You bought 48mp camera for a few hundred quid, and expected to get same quality of a camera that should cost 5/10k. Now i don’t know if you know the story of jack and the beanstalk, well this reminds me of that.

MavGuido Posted at 5-18 01:08
thats not the case, the bayer sensor is in both pictures the same. :-=
It is nice to see that you know something about sensors, but that is not the case here.

https://www.gsmarena.com/quad_bayer_sensors_explained-news-37459.php

MavGuido Posted at 5-17 23:17
Hy guys, thank you for the big repost.
I know what you talking about, i am a photographer since 25years, and what i not mean, is a comparison between both resolution as a jpg, while jpg has a huge postprocessing in the Drone itself.

Do you want me to upload the two DNG files for 12MP and 48MP that I took of the Hollywood Bowl? No post-processing?

melodicskys your picture didn't have much color information and was not as typical picture imho.  

You might want to try some well exposed shot.
Now you get a lot of noise.
I'm not sure about the 48MP, it seems to be a technical way of storing the pixels.
They can also combine it to 12MP

I think this video sums it up well.

Blellow Posted at 5-19 09:08
I think this video sums it up well.

https://youtu.be/1zrJSMehR8o

Good video nice find.

I still don't get why people keep tripping up on this so much.

It's a 48MP sensor. It has 48M sites on the die.
However, the arrangement of the RGB filter on top of it is "clumped" - the arrangement is the same as it would be on a 12 MP sensor.

This means that for all intents and purposes, you can and should treat it as a 12MP sensor with *a slight* edge in resolving power, provided there's enough light*
That's all there is to it.

* when light is low, it will be forced to bin the extra sites on the sensor to prevent excessive shot noise.

Hello folks,
i know the details, the physics and the techs and you are right, but i am speaking of something different!

This Drone has not an native 48MP Sensor, it has a 12MP Sensor and by this there is an prozess that cuts the pixel size in half, that called interpolation and by this the spot size is twice as great as on the 12MP Picture.

So the logical fact here is, that an rezised 12MP DNG to an 48MP DNG within ACR has both a spot size of 2.3 pixels, but an 48mp DNG out of the Cam has an size of 4.6, what gives you interpolation artefacts, while the interpolation process in the Cam is not so professional like in ACR.

So you must have know, that everything like light, color, motion and so on, is not nessesary in this case, while that interpolation artefacts stay there without all that.
You can do what you want, the artefacts stays in any shooting conditions. If you choose jpg over DNG, the effect is not identicals while the postprocessing in jpg is different over DNG.

So summary a smaler but better and lesser processed image could give you a better result in ACR postprocess, than an in camera postprocess.
It is the same like with the missing sharpening settings, if you use an oversharped footage like yet, is half as good as if you can zero the sharpness and make it professionaly in Resolve.

Stay save and healthy. :-)