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

Zoom camera only takes JPEG

Personally I wouldn't find no raw support on a small sensor like that to be much of an issue, there's not much latitude in the raw files to improve them so on the original Mavic Pro I mostly worked with jpeg anyway. With 1in and larger sensors there's a lot more you can pull out of the raw files so that's what I primarily use on the M2P.

That said I'm not sure how useful I'd find the telephoto camera anyway and it's the 4/3 camera that has my interest,
 
  • Like
Reactions: CanadaDrone
Maybe it will turn out that most M3 users won't even use the zoom camera.
If and when I can afford the M3 - the zoom for camera is one of the items I like about it that I CAN use.

The 'intelligence" being built into the M3 seems to be pretty nice. Features like the RTH figures out a way to come home best route - not the way it flew and the OA looks to be upgraded A LOT, so hopefully far less crashes with all the sensors and drone doing it's best to not hit something.

We'll see on the long battery life and with OC3 - the range is phenomenal but meaningless to the majority of us who will buy it.
 
I would say that the micro 4/3 sensor is a good improvement for noise / low light photos etc. But yeah doubt I will upgrade at this price.
It is barely much improvement from the 1 inch, hardly much bigger, I looked at some sample DNGs taken with it and I am not impressed with the lens quality either.
 
It is barely much improvement from the 1 inch, hardly much bigger, I looked at some sample DNGs taken with it and I am not impressed with the lens quality either.

It’s also still 20MP but an even wider lens vs the MP2 so you would need to crop some shots more. It’s a simple decision for me to not buy it.
 
It is barely much improvement from the 1 inch, hardly much bigger, I looked at some sample DNGs taken with it and I am not impressed with the lens quality either.


There is about one stop more dynamic range, a bit more than that at the higher ISO values, and DLog is shot at either ISO 400 or 800. M3 DLog looks worlds better than the M2P from what I have seen so far.

The M2P used subsampling to get half-baked 4K footage in it's FOV mode (28mm). Quality was OK but nothing to write home about. The M3 appears to be using proper oversampled 5.1K video for it's 4K output, which even if it still used a 1" sensor, would be an enormous difference in quality compared to how the M2P achieves 4K.

A 4/3 sensor has about 2X the sensing area as the 1" sensor in the M2P, I would say that having double the area is quite significant.

I haven't seen many comments on the lens yet, other than that none of the reviews I have seen so far have mentioned anything negative about it, only praise for the overall video image quality.

Once more people get their hands on it, everything will become more clear but there are some things that we know for sure already.
 
  • Like
Reactions: justcruisin
There is about one stop more dynamic range, a bit more than that at the higher ISO values, and DLog is shot at either ISO 400 or 800. M3 DLog looks worlds better than the M2P from what I have seen so far.
Is that just video you're meaning? For stills the 13 stop figure from DJI would only give the drone a half stop more dynamic range than the 1in BSI sensors which are around 12.5 stops range, the stacked versions are a bit worse but they're not the version the M2P uses.
 
Which figures are those based on? The 13 stop figure from DJI would give the drone a half stop more dynamic range than the 1in BSI sensors which are around 12.5 stops range, the stacked versions are a bit worse but they're not the version the M2P uses.

The photographic dynamic range charts I posted in prior discussion from Bill Claff:


The difference is smaller at base ISO, but up around ISO 400 and 800 where DLog is, the difference is greater.

That is the best measurement of DR available that I am aware of, and what most of the photographic community goes by.

Every manufacturer measures their own DR differently, so the values they provide are of limited usefulness. Canon used to be famous for grossly overstating their DR, just as one example.
 
The photographic dynamic range charts I posted in prior discussion from Bill Claff:


The difference is smaller at base ISO, but up around ISO 400 and 800 where DLog is, the difference is greater.

That is the best measurement of DR available that I am aware of, and what most of the photographic community goes by.

Every manufacturer measures their own DR differently, so the values they provide are of limited usefulness. Canon used to be famous for grossly overstating their DR, just as one example.
You've mentioned them before but I those numbers clearly have no relevance to actual sensor performance, you can see they're hugely overrating the DR on the 4/3 sensor by adding the A6300 - it's almost identical to the GH5 which is nothing like reality, the APS-C sensor has a good bit more dynamic range. It's not a criticism of the site as I realise putting numbers to dynamic range is difficult and varies hugely.

So instead let's look at the actuaI images and I think you mentioned this comparison yourself, on the DPR comparison scene you can see the Panasonic GH5 II has nothing like a full stop more DR while the A6300 manages a good bit more exactly as would be expected:


In theory the Mavic 3 could be using a different 4/3 sensor but DJI have claimed 13 stops and the same sites that measure the 4/3 sensors at 13 stops (which the graph above does not) measure the 1in sensors at around 12.5 stops.

I don't know about iso and video as I'm mainly interested in stills.
 
Double the area is all great but doesn't provide that much of an advantage and makes lens design within this small size harder.

From the photo samples DP review have taken the lens is not as good as The Mavic 2 Pro in the corners and sides, it is soft which may show up on 4k video, the centre is very sharp though, it will probably be ok in crop mode unless you have a particularly bad lens which is possible with DJI quality control.

You can look at samples here - DJI Mavic 3 Cine sample gallery

I use the drone for photography so quality is paramount. I tested 20 different Mavic 2 pros to find the 2 I have now that had the best centered lens. The quality control of the optics from DJI is terrible and by the looks of it The Mavic 3 will be no different.

From those DP review tests The Mavic 2 Pros I have now are far better than The Mavic 3 for photography due to the sharper sides and corners. If only they had stuck with a 28mm lens or even 35mm, as designing a 24mm for a larger sensor in such a small package is clearly too challenging. The noise on images doesn't seem all that much better either.

Getting great 4k is easy from The Mavic 2 Pro, just shoot crop mode, everything looks more cinematic anyway with a close FOV, which you will have to do anyway with The Mavic 3 if you want sharp sides/corners.

I wouldn't expect youtube reviews to pick up on bad lens quality.

Looking a the raw files I also see something odd going on with the sensor. I attach a screenshot. I can see the pattern of the pixels. Is this some kind of Quad Bayer sensor? Odd.




There is about one stop more dynamic range, a bit more than that at the higher ISO values, and DLog is shot at either ISO 400 or 800. M3 DLog looks worlds better than the M2P from what I have seen so far.

The M2P used subsampling to get half-baked 4K footage in it's FOV mode (28mm). Quality was OK but nothing to write home about. The M3 appears to be using proper oversampled 5.1K video for it's 4K output, which even if it still used a 1" sensor, would be an enormous difference in quality compared to how the M2P achieves 4K.

A 4/3 sensor has about 2X the sensing area as the 1" sensor in the M2P, I would say that having double the area is quite significant.

I haven't seen many comments on the lens yet, other than that none of the reviews I have seen so far have mentioned anything negative about it, only praise for the overall video image quality.

Once more people get their hands on it, everything will become more clear but there are some things that we know for sure already.
 

Attachments

  • Screenshot-2021-11-05-at-22.14.36.jpg
    Screenshot-2021-11-05-at-22.14.36.jpg
    251.9 KB · Views: 6
Last edited:
  • Like
Reactions: KS-6
You've mentioned them before but I those numbers clearly have no relevance to actual sensor performance, you can see they're hugely overrating the DR on the 4/3 sensor by adding the A6300 - it's almost identical to the GH5 which is nothing like reality, the APS-C sensor has a good bit more dynamic range. It's not a criticism of the site as I realise putting numbers to dynamic range is difficult and varies hugely.

So instead let's look at the actuaI images and I think you mentioned this comparison yourself, on the DPR comparison scene you can see the Panasonic GH5 II has nothing like a full stop more DR while the A6300 manages a good bit more exactly as would be expected:


In theory the Mavic 3 could be using a different 4/3 sensor but DJI have claimed 13 stops and the same sites that measure the 4/3 sensors at 13 stops (which the graph above does not) measure the 1in sensors at around 12.5 stops.

I don't know about iso and video as I'm mainly interested in stills.
It looks like the Sensor is a Sony Quad Bayer one, which means the pixel size will actually be smaller than in The Mavic 2 pro, but hopefully there is some advantage to combining those 4 smaller pixels.
 
Early rumors had the M3 releasing in January 2022. The finished firmware, now targeted for January 22. 2022, would seem to confirm that. This early release is DJI reacting to Autel announcements so they can retain customers and keep 'defections' to Autel at bay.
Since Tesla is selling cars with autopilot support in the feature. since 2015 and those people are still waiting until is approved, many companies like DJI now are selling products with "promises" features capabilities without having it actually. I guess this is a lack of respect for the customers selling a product that is not finished or ready.
 
Lycus Tech Mavic Air 3 Case

DJI Drone Deals

New Threads

Forum statistics

Threads
131,534
Messages
1,563,995
Members
160,436
Latest member
GillesdeColombes