Please select Into the mobile phone version | Continue to access the computer ver.
Mavic 2 Pro Sensor Issues
1234Next >
15751 132 2018-9-15
Uploading and Loding Picture ...(0/1)
o(^-^)o
Ahrimaan
lvl.4
Flight distance : 141972 ft
  • >>>
Germany
Offline

So , DJI lied to us ?




2018-9-15
Use props
HedgeTrimmer
Captain
United States
Offline

The resusts need to be indepdently confirmed.  Similar test repeated between Phantom-4 Pro and Mavic-2 Pro.
2018-9-15
Use props
Zatx
First Officer
Flight distance : 158166 ft
United States
Offline

Wow... if this checks out...
2018-9-15
Use props
DJI Stephen
Super Moderator

Offline

Hi there, thank you for reaching out and bringing this up. I will forward this concern to our developer for use to have the feedback and answer about this, Once I have the answer will get back here, Thanks.
2018-9-15
Use props
castormalin
Second Officer
Flight distance : 638907 ft
France
Offline

Thx for this test. Really interesting, indeed.
2018-9-15
Use props
sinloi
lvl.4
United States
Offline

Wow! Wonder how Hasselblad feels about this - their reputation at stake.
2018-9-15
Use props
SMacDrone
lvl.4
  • >>>
Offline

Glad to see your thoughts,  I was thinking something was up with the output size of the photos on Mav 2 Pro compared to P4P.  The P4P ones were larger.  I really think you might be on to something.
2018-9-15
Use props
HedgeTrimmer
Captain
United States
Offline

WingSpeed Posted at 2018-9-15 17:41
DJI bought Hasselblad. It is sad if DJI is ruining the reputation of this iconic brand by crippling the M2Pro.

Before going down that road.  The test done in video needs to be replicated.
2018-9-15
Use props
Eagle_Eye
First Officer
Flight distance : 4137720 ft
United States
Offline

Still on the fence w/ M2P.
2018-9-15
Use props
thehippoz
Captain
Flight distance : 23 ft
United States
Offline

WingSpeed Posted at 2018-9-15 18:39
Already done and reported by others in RC Groups. Wondering if DJI plans on fixing this...

You have a link wing? I would like to see it. I just did a test, the fov mode is worse than hq but it's nothing like in that guys video
2018-9-15
Use props
HedgeTrimmer
Captain
United States
Offline

WingSpeed Posted at 2018-9-15 18:39
Already done and reported by others in RC Groups. Wondering if DJI plans on fixing this...

I haven't seen any separate tests done.  I have seen threads / posts pointing to same video.

Only thing that adds any credibility to video is this statement by CT:
   I did give DJI 1 week to watch this privately and give me their feedback. There were 76 views all within DJI internally but they declined to comment.
2018-9-15
Use props
EdisonW1979
First Officer
Flight distance : 1366555 ft
Canada
Offline

HedgeTrimmer Posted at 2018-9-15 18:00
Before going down that road.  The test done in video needs to be replicated.

This guy lost me the moment he showed the resolution chart for the M2P 4K FOV mode, that wasn’t the result of pixel binning or line skipping, that was a focusing issue!

I’ve already seen other videos where the M2P seems to have issues with it’s auto focus like its predecessor, and many people have shown, with video evidence, that setting the focus manually to the infinity point fixes it, and image clarity is restored.

It’s quite funny because We Talk UAV just did their M2P vs P4P2 comparison, and when they applied an unsharp mask to the M2P footage, the zoomed in detail levels between the two were nearly identical, with the exception of a bit more ISO noise evident in the M2P footage...

Like you said, this needs to be independently verified, because to me that M2P 4K FOV focus / resolution chart footage screams out-of-focus and not any sort of pixel binning / line skipping...
2018-9-15
Use props
HedgeTrimmer
Captain
United States
Offline

EdisonW1979 Posted at 2018-9-15 20:17
This guy lost me the moment he showed the resolution chart for the M2P 4K FOV mode, that wasn’t the result of pixel binning or line skipping, that was a focusing issue!

I’ve already seen other videos where the M2P seems to have issues with it’s auto focus like its predecessor, and many people have shown, with video evidence, that setting the focus manually to the infinity point fixes it, and image clarity is restored.

It may even be a case the guy has a bad copy of MP2.  It's not like there hasn't been bad copies of MPp, RC, Cs-Ub, or ...  


2018-9-15
Use props
jffry7
lvl.4
Flight distance : 27566 ft
United States
Offline

EdisonW1979 Posted at 2018-9-15 20:17
This guy lost me the moment he showed the resolution chart for the M2P 4K FOV mode, that wasn’t the result of pixel binning or line skipping, that was a focusing issue!

I’ve already seen other videos where the M2P seems to have issues with it’s auto focus like its predecessor, and many people have shown, with video evidence, that setting the focus manually to the infinity point fixes it, and image clarity is restored.

I agree with your observation, in fact I lose interest when he started saying he will compare 2 diff FOV and present them (zoom  in) with similar FOV, even moving the subject distance will have different result. This could cause soft/out of focus result. I think the test is very flawed to begin with.

However, I dont think the M2P is not without flaw and an independent test with more proper setup and well lit setup and attention to focus would need to be done. Tho I would still think the P4P V2 would still win but not as how the guy presented it, if not maybe very close.

Or maybe I got it completely wrong since I am no expert.
2018-9-15
Use props
FlyingUser
lvl.2

United Kingdom
Offline

EdisonW1979 Posted at 2018-9-15 20:17
This guy lost me the moment he showed the resolution chart for the M2P 4K FOV mode, that wasn’t the result of pixel binning or line skipping, that was a focusing issue!

I’ve already seen other videos where the M2P seems to have issues with it’s auto focus like its predecessor, and many people have shown, with video evidence, that setting the focus manually to the infinity point fixes it, and image clarity is restored.

Unlikely, you won't get Moiré patterns with bad focusing.BTW, line skipping is still quite common, unfortunately. Full-sensor readout and smart downsizing to 4k needs quite significant processing power and proper cooling of the image processor. Where is the image processor located for the m2p and the p4p? (camera head right behind the sensor vs in the aircraft?)
2018-9-16
Use props
EdisonW1979
First Officer
Flight distance : 1366555 ft
Canada
Offline

FlyingUser Posted at 2018-9-16 07:28
Unlikely, you won't get Moiré patterns with bad focusing.BTW, line skipping is still quite common, unfortunately. Full-sensor readout and smart downsizing to 4k needs quite significant processing power and proper cooling of the image processor. Where is the image processor located for the m2p and the p4p? (camera head right behind the sensor vs in the aircraft?)

I must disagree...

Whilst yes I can identify Moire patterns in the footage in question of the M2P, pay close attention to the King card on the left side; that doesn’t show Moire distortion, but is rather completely out of focus. I’ve seen how some sensors, especially Sony’s, which the M2P uses, can create Moire-like distortions in out-of-focus scenarios, or at certain ISO levels.

As for where the actual ISP is for the camera, there’s no way to know for sure, as no schematics have been made available, and I don’t recall seeing the ISP on the tear down image of the M2P that was posted here for a while. I would like to think this chip, which likely generates a lot of heat, is located somewhere in the main chassis to allow the primary cooling system to govern its temperature.

I guess the bottom line is, we can’t know anything for certain at this point, at least, not with this video and how the guy presented it.

I went onto DPREVIEW.COM, and they state they’ll be doing an in-depth review of the M2P/Z, so it’ll be interesting to see what they come up with, as they are VERY good in doing these analysis’.

Until then, most of this is idle speculation...
2018-9-16
Use props
EdisonW1979
First Officer
Flight distance : 1366555 ft
Canada
Offline

HedgeTrimmer Posted at 2018-9-15 20:34
It may even be a case the guy has a bad copy of MP2.  It's not like there hasn't been bad copies of MPp, RC, Cs-Ub, or ...

Quite possible! I mean, look at even the Mavic Air, how many of those were affected by soft lenses? Even mine has a slight softening effect just off-centre to the upper-right of the image, but on mine it’s not nearly as strong as I’ve seen on other videos posted by pilots here...

The guy makes these claims based on his one unit, so it’s not a fair statement to make that all M2P’s might have this issue! Like the unit We Talk UAV put against the P4P2, those were neck-and-neck in detail, so how would this guy explain that?

This needs more investigating for sure...
2018-9-16
Use props
FlyingUser
lvl.2

United Kingdom
Offline

EdisonW1979 Posted at 2018-9-16 10:29
I must disagree...

Whilst yes I can identify Moire patterns in the footage in question of the M2P, pay close attention to the King card on the left side; that doesn’t show Moire distortion, but is rather completely out of focus. I’ve seen how some sensors, especially Sony’s, which the M2P uses, can create Moire-like distortions in out-of-focus scenarios, or at certain ISO levels.

The card "doesn't attract" the Moiré plague, but simple line skipping may have a negative effect on overall detail/resolution as well. First and obvious effect: if you simply forget the extra pixels, instead of doing e.g. a median averaging, you loose a powerful noise reduction tool. This is not that significant at everyday ISO100 shots, but much more important when you shoot higher iso's or you need an extended exposure latitude in a high-dynamic range environment (dusk or dawn...).

Found another review here, original files can be downloaded:



To the question "has DJI lied?" Nope, they promised an upgrade to the Mavic pro, which they indeed delivered! Their intention wasn't to beat the P4P, which maybe wasn't possible in the smaller space and power envelope.
2018-9-16
Use props
EdisonW1979
First Officer
Flight distance : 1366555 ft
Canada
Offline

FlyingUser Posted at 2018-9-16 12:23
The card "doesn't attract" the Moiré plague, but simple line skipping may have a negative effect on overall detail/resolution as well. First and obvious effect: if you simply forget the extra pixels, instead of doing e.g. a median averaging, you loose a powerful noise reduction tool. This is not that significant at everyday ISO100 shots, but much more important when you shoot higher iso's or you need an extended exposure latitude in a high-dynamic range environment (dusk or dawn...).

Found another review here, original files can be downloaded:

Yes, this was the We Talk UAV video I was talking about... In this video, especially with unsharp masks applied to the M2P footage, it really looks comparable to the P4P2 in most scenarios, but in low light the P4P2 still wins by a bit.

And you're right, DJI in this case delivered on their promise of an enhanced Mavic Pro. And whilst it's not 100% of what the P4P2 is (remember M2P still cannot do 4k/60fps), it's very close it, and I think for many can replace the P4P2 for travel.
2018-9-16
Use props
lawrenceB
lvl.1

United States
Offline

re: Did they lie?  
2018-9-16
Use props
lawrenceB
lvl.1

United States
Offline

Sorry, messed up that last post with a quick reply.

Regarding whether 'DJI lied':

I think it comes down to how they've advertised the FOV and HQ modes, and to the details of what they're doing, capture and processing-wise.

If they're calling those modes '4k', but it turns out they're upscaling 2.7k captures/input to 4k output res, I think they're in deep (legal) trouble. and had better issue a fix, pronto.

Imagine if you captured at some absurdly low res, say 512x288, then up res'd to '4k'--no one would agree that that is '4k'.  If they're upscaling 2.7k, it's the same situation, just less extreme. You can't create detail that is not there, so everything is bound by the initial capture/input res. That's your _best_ case for detail. If you start with less than 4k, what you end up with is worse than true 4k. Full stop, end of story.
2018-9-16
Use props
thehippoz
Captain
Flight distance : 23 ft
United States
Offline

lawrenceB Posted at 2018-9-16 18:03
Sorry, messed up that last post with a quick reply.

Regarding whether 'DJI lied':

I really don't think that's the case. If you have a 4k monitor look at my test last night https://bit.ly/2NgKO2Z

You can actually see the pixels in my 32" monitor in 4k hq normal. I didn't move between modes, so the fov change (from 55 to 77) makes chart 30 a 22 in full (you can measure this yourself). The pixels are gone though in full but chart 22 is 30 if measured.

Full is a processed mode though not 1:1 crop so you would never see pixels. And I guess that's the mode up for debate. 4k HQ is not 2k upsampled, if it was, you wouldn't see every single pixel like you do. I have a 40" monitor too but it's curved, not good for this kind of test.

I could try and see what I get matching the size of the card in 77 fov. But I already know what you'd see pretty much. No pixels because it's processed (I maybe wrong) and the 30 chart would match.
2018-9-16
Use props
Ahrimaan
lvl.4
Flight distance : 141972 ft
  • >>>
Germany
Offline

@DJI Stephen anything new ?
There are rumors, that the 1 inch seonsr is overheating so DJI is only reading every second line of the sensor and scaling up to 4K (2,7K up to 5.5 and down to 4K)
hopefully this is just a bug, if not ... i smell many many refunds ;)
2018-9-18
Use props
gnirtS
Captain
Flight distance : 2993497 ft
United Kingdom
Offline

EdisonW1979 Posted at 2018-9-15 20:17
This guy lost me the moment he showed the resolution chart for the M2P 4K FOV mode, that wasn’t the result of pixel binning or line skipping, that was a focusing issue!

I’ve already seen other videos where the M2P seems to have issues with it’s auto focus like its predecessor, and many people have shown, with video evidence, that setting the focus manually to the infinity point fixes it, and image clarity is restored.

Not just that, the optics are optimised for infinity focus distances typical of what you'd get in the air.
Using a dodgy test chart WAY too close is never going to be a fair, controlled test.
2018-9-18
Use props
FlyingUser
lvl.2

United Kingdom
Offline

gnirtS Posted at 2018-9-18 01:03
Not just that, the optics are optimised for infinity focus distances typical of what you'd get in the air.
Using a dodgy test chart WAY too close is never going to be a fair, controlled test.

The one I linked above in #21 was neither a test chart nor a close-up testing. Still the same conclusion...
@Ahrimaan, DJI is only reading every second line of the sensor and scaling up to 4K (2,7K up to 5.5 and down to 4K)
hopefully this is just a bug, if not ... i smell many many refunds ;) - many camera makers do that, no reason for refund.

2018-9-18
Use props
simis
First Officer
Flight distance : 344551 ft
Latvia
Offline

They have NOT ENOUGH POWER to fix lens distortion, how you imagine they going to fix THIS????
2018-9-18
Use props
Mikex
lvl.4
Ukraine
Offline

oh, i was waiting for mavic two, thinking that they will fix video issues, and now with 1 inch sensor mavic pro quality about the same as in air...

Thinking about buying phantom instead of mavic...
2018-9-18
Use props
FlyingUser
lvl.2

United Kingdom
Offline

Mikex Posted at 2018-9-18 03:18
oh, i was waiting for mavic two, thinking that they will fix video issues, and now with 1 inch sensor mavic pro quality about the same as in air...

Thinking about buying phantom instead of mavic...

It's better than the air, almost P4P quality with the mobility of a Mavic, that's the m2p story.
2018-9-18
Use props
Ales Bajc
lvl.1
Flight distance : 215289 ft
  • >>>
Slovenia
Offline

I have same problem, and I had all of the DJI drones...I even think on SUNNY day with no dark areas the MAVIC 1 PRO has better image video. Anyone agrees ?
2018-9-18
Use props
gnirtS
Captain
Flight distance : 2993497 ft
United Kingdom
Offline

Ales Bajc Posted at 2018-9-18 03:59
I have same problem, and I had all of the DJI drones...I even think on SUNNY day with no dark areas the MAVIC 1 PRO has better image video. Anyone agrees ?

Nope. M2P is massively better than m1 for me.
2018-9-18
Use props
trica de
First Officer
Flight distance : 729199 ft
Germany
Offline

if thats true - it would be a shame from dji!
2018-9-18
Use props
FlyingUser
lvl.2

United Kingdom
Offline

trica de Posted at 2018-9-18 04:42
if thats true - it would be a shame from dji!

I do not recall DJI claiming the m2p being better than the P4P. Line skipping is a common strategy in digital video, many cameras do that if resources are limited. And yes, they are more limited than in a Phantom.
The same should go to those forumers who advocated ditching the p4p and run to the stores for a new m2p.
2018-9-18
Use props
Mikex
lvl.4
Ukraine
Offline

FlyingUser Posted at 2018-9-18 03:32
It's better than the air, almost P4P quality with the mobility of a Mavic, that's the m2p story.

sure, you're right here, as a drone - Mavic great, but as a camera with 1' sensor, not good as it can be.
And I almost don't see picture quality difference between Air and MP2, don't forget that camera with 1' sensor should be a lot.. a lot better than 1/2.3.

I don't believe that this will fixed with firmware updates, 'cause we have tiny camera in MP2 and there is no space to put some sort of cooler to reduce sensor overheat.

So we have the same story with camera as with MP1, I think it can be fixed only in next generation...
2018-9-18
Use props
castormalin
Second Officer
Flight distance : 638907 ft
United Kingdom
Offline

The question for DJI/HB is simple:
Does DJI/HB  use 100% of the sensor pixels for the 4k Full FOV mode generation ?
No answer means = NO
Is it an issue ?
For the photo, no. People have just to manage the distorsion correction and a slight loss on image quality on the edges. But that's fine.
In video. I'd like to know if the 'resolution' issue exists in all Full FOV modes, or only in D-Log 10 bits like the distorsion


2018-9-18
Use props
simis
First Officer
Flight distance : 344551 ft
Latvia
Offline

FlyingUser Posted at 2018-9-18 05:14
I do not recall DJI claiming the m2p being better than the P4P. Line skipping is a common strategy in digital video, many cameras do that if resources are limited. And yes, they are more limited than in a Phantom.
The same should go to those forumers who advocated ditching the p4p and run to the stores for a new m2p.

I do not recall DJI claiming the m2p 1" sensor being worth than the P4P ether
2018-9-18
Use props
AForceOne
lvl.3
Flight distance : 5339846 ft
United States
Offline

Would love to hear DJI's response. I have been wondering why 4K HQ has to be cropped and cannot use the full sensor and why the FOV mode re-samples footage and degrades the quality.
2018-9-18
Use props
Peach
lvl.1
Norway
Offline

Why don't DJI just officially confirm how they use the sensor?
It is ok for me if they use line skipping, and if they can't fix the distortion, as long as the final (edited) image quality is good.
But why keep this info as a secret?
2018-9-18
Use props
fanscb6385d2
lvl.3
Flight distance : 281283 ft
Offline

They call it "Raw subsampling ", whatever that is. But they surely need to optimize the algorithm...

Summary.jpg
2018-9-18
Use props
castormalin
Second Officer
Flight distance : 638907 ft
United Kingdom
Offline

Yeah.. Subsampling, undersampling whatever it's binning with digital LPF to limit aliasing and limit the noise.
It's better than line skipping, but the result is quite the same in terms of resolution
The P4P makes oversampling and has a 4K resolution better than a 12MP
The MP2 in 'cropped HQ' is a 12MP so has the same resolution
The MP2 in 'Full FOV' is a binned one with a resol lower than a 12 MP (eq. 2.7 K)
I like the answer of the guy on RCG : 'it's visible only on test chards at 400% , not on real images (I guess, on the sky)' .
2018-9-18
Use props
fanscb6385d2
lvl.3
Flight distance : 281283 ft
Offline

Downscale the image by 4x and you will still see it... 400% just makes it also visible for iPhone users... ;-)
2018-9-18
Use props
1234Next >
Advanced
You need to log in before you can reply Login | Register now

Credit Rules