After my first test flight and a couple of test shots (albeit in good lighting conditions), it seems pretty real to me

Not all 48mpx are equal watch half chrome's vid here
https://youtu.be/1zrJSMehR8o it's a nice to have and bodes well for what the Mavic 3 feature set might hold. I 30mpx 1" would be nice.

BaynhamPhoto Posted at 5-23 09:47
Not all 48mpx are equal watch half chrome's vid here
https://youtu.be/1zrJSMehR8o it's a nice to have and bodes well for what the Mavic 3 feature set might hold. I 30mpx 1" would be nice.

Yeah, don’t really care about that tbh Is it a significant improvement over the standard 12MP? If so, it’s irrelevant that a much larger sensor (say 1”) of 48MP would get even better detail.

Wouldn't say it's an improvement unless you are referring to the earlier mavic pro/mini sensor. As I said its a nice to have feature but I will stick to the 12 with better DR and stich for larger image size if needed but again it depends if you process raw or just take the jpeg. The sensor is a good compromise to keep the aircraft size down a factor for me or I would have gone MP2 possibly waited for the 3. As long as you like the output that's good enough

BaynhamPhoto Posted at 5-23 11:42
Wouldn't say it's an improvement unless you are referring to the earlier mavic pro/mini sensor. As I said its a nice to have feature but I will stick to the 12 with better DR and stich for larger image size if needed but again it depends if you process raw or just take the jpeg. The sensor is a good compromise to keep the aircraft size down a factor for me or I would have gone MP2 possibly waited for the 3. As long as you like the output that's good enough

Yes, not saying it’s good for any situation by any means.

Ice_2k Posted at 5-23 11:25
Yeah, don’t really care about that tbh  Is it a significant improvement over the standard 12MP? If so, it’s irrelevant that a much larger sensor (say 1”) of 48MP would get even better detail.

well the sensor size is all that it counts..it is not irrelevant, the bigger the better...

it is and isn't a real 48MP.  it uses quad bayer filtering on sensor which in most basic of terms means its sampling 2x on each axis, so more like its taking 2x 24mp images and pixel binning them together in sensor.  so lives in this inbetween relm of real and not real...   
but dont forget, MP isnt everything, lot more goes into it..   

sbonev Posted at 5-23 13:19
well the sensor size is all that it counts..it is not irrelevant, the bigger the better...

it is irrelevant imho because I'm not comparing this sensor to a bigger sensor. I'm comparing what this sensor can do in its 12MP form and whether bumping it up to 48MP using the quad bayer filter actually provides any benefits. And it seems it does.

Ice_2k Posted at 5-23 14:02
it is irrelevant imho because I'm not comparing this sensor to a bigger sensor. I'm comparing what this sensor can do in its 12MP form and whether bumping it up to 48MP using the quad bayer filter actually provides any benefits. And it seems it does.

well actually the relevant stuff is something else....because the file size is quite disturbing to say the least. I have been looking at some samples of exact same scenes 12 mp dng - 25 mb, 48 mp dng - 100mb. After processing the differences were so little in matter of quality but the quantity difference in matter of size was blowing my mind...there is no justifying that, i cannot see making more than a couple of pictures in 48 mp and still it has to be a 100% scene which is difficult...and when i take my compact with 1' that takes 20 mp raw file with a size of 24 mp and the quality is simply not comparable... so now i see why some reviewers were skeptical for the usage of the 48 mp. it is good, but not great and definitely not something that can actually can be justified for regular use.

sbonev Posted at 5-23 14:56
well actually the relevant stuff is something else....because the file size is quite disturbing to say the least. I have been looking at some samples of exact same scenes 12 mp dng - 25 mb, 48 mp dng - 100mb. After processing the differences were so little in matter of quality but the quantity difference in matter of size was blowing my mind...there is no justifying that, i cannot see making more than a couple of pictures in 48 mp and still it has to be a 100% scene which is difficult...and when i take my compact with 1' that takes 20 mp raw file with a size of 24 mp and the quality is simply not comparable... so now i see why some reviewers were skeptical for the usage of the 48 mp. it is good, but not great and definitely not something that can actually can be justified for regular use.

i honestly don't understand why we keep talking about bigger sensors

Ice_2k Posted at 5-23 15:20
i honestly don't understand why we keep talking about bigger sensors

no, i am talking about 20 mb vs 100 mb difference if you decide to shoot 48 mp...for that small quality improvement is not justified...the way i see it the most useful is the 12 mp hdr and 12mp raw, 48 is good until you realize it is more waste of space than it is worth to bother with

It is 4 times the size (96mb vs 24mb) but I’m not too bothered about that, I have a large SD in the drone and I won’t be storing it long term as a 48MP RAW so it’s not really an issue for me. If I can get better quality in a certain situation, I will take it. The sample photo I posted at the beginning of this topic i think it’s a great example where “wasting” that extra space on the SD was worth it.

Some background info which might clear the confusion: The photo sites on the sensors can only detect the intensity of the light, not it's color; so essentially they can only see in "black and white". A checkered color filter is applied in front, passing only red, green or blue light in any given area (normally each a pixel size) and these individual readings are then combined to interpret color.

The standard layout is to have a square of 4 pixels in each "color zone", two diagonal ones measuring green light only (as human eye is much more sensitive to that) and remaining two measuring red and blue spectrum only. So each pixel actually contains 1/3rd of the needed color information (since green is repeated twice). This is called a "Bayer Filter".  Then the full color of each pixel is derived by borrowing the lacking information from neighboring pixels as you need all of the red, green and blue values to form full color. This is called "demosaicing". I know I am oversimplifying, but this is the gist of it.

A "Quad Bayer" filter means the color resolution is reduced even further, down to 1/4th of what it normally is exactly. So it uses a 16x16 pixel square zone, but every 4x4 pixel square gets a single color filter, instead of having their individual ones. For every 16 pixels, there is a single red, single blue, and 2 green readouts. A "Quad Bayer" 48Mp sensor is basically a 48MP sensor with the color filter from a 12MP one.

It is normal that you get more detail at 48Mp, as a quad bayer 48MP sensor stil records 48 million pixels. It is not "fake" or "interpolated" (well, apart from color) . But can it compete with a 48MP sensor with a regular, non "Quad/Crippled" bayer filter? I would say no

And before and above all this conversation, there is the matter of diffraction limit; which is a product of any particular sensor/lens combination. It can also be oversimplified as the "actual resolution of the light" hitting the sensor. Sometimes the spots of light (Airy disks) hitting the sensor are so big, they cover multiple pixels rendering their readouts identical and thus, reducing the true resolution of the output. The smaller the sensor, the higher the megapixel, the more likely this is to happen.

Hope this helps

Great writeup!

Good comparison.
This is the only way to really check the difference.

The 48mp definitely. looks better. Nice test

CemAygun Posted at 5-23 17:22
Some background info which might clear the confusion: The photo sites on the sensors can only detect the intensity of the light, not it's color; so essentially they can only see in "black and white". A checkered color filter is applied in front, passing only red, green or blue light in any given area (normally each a pixel size) and these individual readings are then combined to interpret color.

The standard layout is to have a square of 4 pixels in each "color zone", two diagonal ones measuring green light only (as human eye is much more sensitive to that) and remaining two measuring red and blue spectrum only. So each pixel actually contains 1/3rd of the needed color information (since green is repeated twice). This is called a "Bayer Filter".  Then the full color of each pixel is derived by borrowing the lacking information from neighboring pixels as you need all of the red, green and blue values to form full color. This is called "demosaicing". I know I am oversimplifying, but this is the gist of it.

I think we can talk about the mumbo jumbo all day, but first shooting 48mp with Air2 is basically free and if it’s an improvement (and I will say light here is a big factor) then by all means shoot in 48mp, I know I will when the opportunity arises. As somebody who regularly shoots landscapes with a really good crop sensor camera big sensor more MP and great quality photographs. I have noticed particularly over the last two years whenever I show my photos that 90% of people yes 90% are attracted and find visually attractive dramatic and have the real wow factor are Aerial shots whether shot with my P4P M2 or even my Mavic mini.
So apart from the pixel peepers “mainly Photographers “ people are interested in the photograph not the file size, and I believe these people are looking for interesting photographs taken by talented photographers, not graphic designers. It’s just my opinion but it’s influenced by the people I take the photographs for .

What Photo program did you use for 400% and 800% views of 48MP vs. 12MP?   

Asking because 48MP when zoomed in appears like a Artistic filter was used.  The bush above shed looks like it is swirl painted.

HedgeTrimmer Posted at 5-25 11:28
What Photo program did you use for 400% and 800% views of 48MP vs. 12MP?   

Asking because 48MP when zoomed in appears like a Artistic filter was used.  The bush above shed looks like it is swirl painted.

Photoshop, nothing else done on the image except zoom. Folks on mavicpilots.com were saying those are the artefacts created by the quad bayer filter

Ice_2k Posted at 5-25 13:11
Photoshop, nothing else done on the image except zoom. Folks on mavicpilots.com were saying those are the artefacts created by the quad bayer filter

Thanks, thought it might be effect of quad bayer filter.  
Interesting look.  Similar to effect of Impresso has on images viewed 1:1.

Ice_2k Posted at 5-25 13:11
Photoshop, nothing else done on the image except zoom. Folks on mavicpilots.com were saying those are the artefacts created by the quad bayer filter

Makes sense... Also bear in mind, even if there was no "quad" bayer filter; when the output is not binned down to 12MP and the pixels are used individually, the pixel size would still have been a bit too small. So it is only normal that the output has to be processed heavily.

One way or another, I believe the choice of sensor in MA2 was a big step forward. Some of the phones get incredible results with these...