Mavic Pro-2 vs Zoom and P4P - Line Skipping + Sensor Heat?

[yysc]

Ah, that‘s not the same: LX10 has lower 4K resolution than RX100V since it‘s also only „native“ cropping the sensor in 4K so you get 36mm instead of 24mm for photo.
RX100V uses the full width with oversampling like the P4pro.

Image comparison: Digital Photography Review

GH5 is a complete different animal with a M43 sensor and a big body that can dissipate temperature better.
Question is what happens with heat in the P4pro in 4K oversampling video when not cooled by the props when filming on the ground.
I have mine not here (I’m on vacation) to test.

Speaking about GH5, that and the G9 share the sensor and both downsample equally from 5.5K to 4K but the G9 has a 4K limit due to smaller heatsinks. I believe the biggest source of heat is the SOC as basically all the downsample calculations, perspective adjustment, encoding, etc are done here.

Now what the RX100 doesn't have and both the P4P and M2P have is active cooling. Their little fans which in the case of the M2P sounds louder, spins faster. The H1 SOC is manufactured with a 28nm node, whereas the H3 is done with a 14nm node which is the best Intel for example can offer today. Smaller process means less heat.

This combined with the airflow in the camera, and the metallic heatsink shaped camera body in the M2P makes me question if thermals are really the issue.

 

[chunky]

What makes us really that sure that the M2P uses an Ambarella H3, what that it uses a Sony IMX183xxx? That‘s all nothing but guessing, not??

We only KNOW that the P4pro has an Ambarella H1 (CLEARLY to see in teardown photos) and that DJI states themself that it has a Sony 1“sensor.

For the M2P „I“ only saw a teardown photo with a chip that had a large H3 on it - but that can be everything.

But why does that even matter?!
They could have included an unknown Samsung sensor and a homebrew SoC, who cares!?

But maybe because of that things become harder and even more complex for DJI because one thing is fact - they CLEARLY state 4K VIDEORESOLUTION (not imagesize):

- in the specs
(FullFOV isn‘t specified differently)


- the FAQs


- and the manual


Nobody of the reviewers, of the „testers“ here and either not DJI could prove reliable yet that FullFOV mode delivers real (or close) 4K resolution (image detail/-information) instead of only 4K imagesize and so it is only promoted but not complied!

„We“ only found that FullFOV looks somewhere close to 2.7K imageresolution from our tests what makes sense because this would be the result of pixelbinning the 5.5K sensors image data.

As long as nobody and even not DJI proves that also FullFOV delivers the promoted 4K „resolution“ the product is „out of specs“!
We may get a (small) kind of „Dieselgate“ here.

I also still don‘t find any „official“ DJI statement about the FullFOV mode being „one class worse“, being 2.7K instead of the promised 4K, do you?!
I had expected one class better, 5.5K oversampling to 4K down scale as described and as it was possible with the P4pro.


BTW: with other copters before the specs where not so „clear“.
With the M1P you only spoke of „imagesize“ and „video recording modes“ - nothing anybody could „count“ on.

 

[chunky]

Videoresolution != imagesize is nothing to uncommon (where oversampling cameras usally come very close in lumaresolution), but 4K specified resolution with only ~2.7K effective resolution is unusual and „overrated“!

 

[mrg]

The video resolution of the M2Pro is further reduced when using more than 30 FPS, which looks like line skipping and DJI doesn't say anything about this problem anywhere. With the M2Zoom the problem doesn't exist and 2.7K 60p has the same resolution as 2.7K 30p.

 

[chunky]

The video resolution is further reduced when using more than 30 FPS, which looks like line skipping and DJI doesn't say anything about this problem anywhere. With the Zoom the problem doesn't exist and 2.7K p60 has the same resolution as 2.7K p30.

Interesting, didn‘t check this myself yet.
The P4pro also had lower resolution in 4K60p than in 4K30p but as far as I remember it was even higher than the M2P FullFOV resolution.

 

[WithTheBirds]

What makes us really that sure that the M2P uses an Ambarella H3, what that it uses a Sony IMX183xxx? That‘s all nothing but guessing, not??

We only KNOW that the P4pro has an Ambarella H1 (CLEARLY to see in teardown photos) and that DJI states themself that it has a Sony 1“sensor.

For the M2P „I“ only saw a teardown photo with a chip that had a large H3 on it - but that can be everything.

But why does that even matter?!
They could have included an unknown Samsung sensor and a homebrew SoC, who cares!?

But maybe because of that things become harder and even more complex for DJI because one thing is fact - they CLEARLY state 4K VIDEORESOLUTION (not imagesize):

- in the specs
(FullFOV isn‘t specified differently)
View attachment 47065

- the FAQs
View attachment 47066

- and the manual
View attachment 47067

Nobody of the reviewers, of the „testers“ here and either not DJI could prove reliable yet that FullFOV mode delivers real (or close) 4K resolution (image detail/-information) instead of only 4K imagesize and so it is only promoted but not complied!

„We“ only found that FullFOV looks somewhere close to 2.7K imageresolution from our tests what makes sense because this would be the result of pixelbinning the 5.5K sensors image data.

As long as nobody and even not DJI proves that also FullFOV delivers the promoted 4K „resolution“ the product is „out of specs“!
We may get a (small) kind of „Dieselgate“ here.

I also still don‘t find any „official“ DJI statement about the FullFOV mode being „one class worse“, being 2.7K instead of the promised 4K, do you?!
I had expected one class better, 5.5K oversampling to 4K down scale as described and as it was possible with the P4pro.


BTW: with other copters before the specs where not so „clear“.
With the M1P you only spoke of „imagesize“ and „video recording modes“ - nothing anybody could „count“ on.

Its probably unlikely that Sony released a new 1" sensor without a public announcement- DJI might be a big customer however supplying them with exclusive silicon probably isn't a likely scenario.

From the Sony specifications it is apparent that the 1" sensor only provides one readout mode where horizontal and/or vertical line addition isn't performed as depicted in the table below. All other readout modes use pixel summing. It is well known that CMOS imaging sensors do not perform pixel binning at the sensor level, neighbouring pixel summing is what is ordinarily perfumed. Whether the image processing SOC might be performing line/column skipping or binning is what remains to be established.

 

[chunky]

Interesting, didn‘t check this myself yet.
The P4pro also had lower resolution in 4K60p than in 4K30p.

Its probably unlikely that Sony released a new 1" sensor without a public announcement- DJI might be a big customer however supplying them with exclusive silicon probably isn't a likely scenario.

From the Sony specifications it is apparent that the 1" sensor only provides one readout mode where horizontal and/or vertical line addition isn't performed as depicted in the table below. All other readout modes use pixel summing. It is well known that CMOS imaging sensors do not perform pixel binning at the sensor level, neighbouring pixel summing is what is ordinarily perfumed. Whether the image processing SOC might be performing line/column skipping or binning is what remains to be established.

View attachment 47069

What do you expect to be the difference between pixel summing vs. pixel binning?

In the end nothing changes, 4K resolution offered, only ~2.7K delivered.

 

[Synephrine]

Im losing faith this can be fixed at this point. This is sounding more and more like a hardware limitation that is significantly misaligned with the printed 4k res specification + what was advertised.

Question about a refund. If purchased on day 1, would a refund by dji be considered at this point with this misadvertised / subperformance / customer unsatisfaction?

 

[WithTheBirds]

What do you expect to be the difference between pixel summing vs. pixel binning?

In the end nothing changes, 4K resolution offered, only ~2.7K delivered.

One gives better noise performance- both give less than the native sensor resolution which is undoubtably the problem we are hoping doesn’t exist.

 

[chunky]

One gives better noise performance- both give less than the native sensor resolution which is undoubtably the problem we are hoping doesn’t exist.

I hardly see any difference since I can not imagine there is even any - technically.
Could someone explain what the technical difference between „summing“ and „binning“ should be, with references??

BTW: „binning“ is done on the sensor - otherwise it would be worthless.
It‘s done for reading the demanded sensor area with all included pixels (for better SNR) with high frame rates at the possible MIPI or LVDS clock speed over the available readout channels.

 

[WithTheBirds]

I hardly see any difference since I can not imagine there is even any - technically.
Could someone explain what the technical difference between „summing“ and „binning“ should be, with references??

BTW: „binning“ is done on the sensor - otherwise it would be worthless.
It‘s done for reading the demanded sensor area with all included pixels (for better SNR) with high frame rates at the possible MIPI or LVDS clock speed over the available readout channels.

Technically there is a significant difference. A CMOS sensor can only combine pixels by averaging in the digital domain whereas CCD devices can combine electrical charges prior to AD conversion (true binning). Both can give reduced noise due to averaging and reduced data but only true hardware binning can provide increased SNR due to reduced read noise.

 

[chunky]

Technically there is a significant difference. A CMOS sensor can only combine pixels by averaging in the digital domain whereas CCD devices can combine electrical charges prior to AD conversion (true binning). Both can give reduced noise due to averaging and reduced data but only true hardware binning can provide increased SNR due to reduced read noise.

A CCD is not present in the M2P, no question ;-)!
The combine/binning process kicks in on the sensor before the A/D conversion (on sensor with Sony, off sensor with many others since „Exmor“ is a Sony patent).

 

[WithTheBirds]

A CCD is not present in the M2P, no question ;-)!

Yep.... Like everyone I am trying to understand what DJI might have tried to do to us.

My point re the binning really is whatever games are being played they will be post sensor readout.

I find it difficult to accept it is a heat issue, unless the engineers are grossly incompetent that sort of problem would have been caught early in the design process.

 

[chunky]

Many of the current sensors, especially Sony Exmor can do the binning on the sensor itself (Exmor also contains the ADC).

Here is a nice explanation for Quad HDR adaptive binning on a Sony chip that also happens on the sensor before readout and further processing in the SoC:
IMX294CJK | Sony Semiconductor Solutions

There was also a nice explanation for on sensor binning I think on FRAMOS but I don’t find it anymore.

Whatever happened, the current result for the FullFOV mode is a whatever flaw and the explanation of the potential DJI guy is it even more.
I can do nothing with „RAW subsampling“, for
me it sounds like binning that is even an advantage over line skipping because of the better SNR.
If „RAW subsampling“ is simple on sensor binning for saving readout efforts then it‘s simply „binning“ to 2.7K with no excuse - bad since the P4pro was capable of much more.
If they read the full sensor and do their „RAW subsampling“ in their „processor“ - then it‘s unclear why this „quality reduction“ happens anyway since every needed information is already present in the pipe.

We can break our heads but actually it is like it is, we bought a 4K drone with only the power of 2.7K in FullFOV.

Even worse it seems the „native“ 2.7K modes become even less powerful in fps‘es over 30p (haven‘t checked that myself yet).

 

[bluesgeek]

…Whatever happened, the current result for the FullFOV mode is a whatever flaw and the explanation of the potential DJI guy is it even more.
I can do nothing with „RAW subsampling“, for
me it sounds like binning that is even an advantage over line skipping because of the better SNR.
If „RAW subsampling“ is simple on sensor binning for saving readout efforts then it‘s simply „binning“ to 2.7K with no excuse - bad since the P4pro was capable of much more.
If they read the full sensor and do their „RAW subsampling“ in their „processor“ - then it‘s unclear why this „quality reduction“ happens anyway since every needed information is already present in the pipe.

Wouldn't the term RAW subsampling imply pre-processor?

 

[chunky]

Many of the current sensors, especially Sony Exmor can do the binning on the sensor itself (Exmor also contains the ADC).

Here is a nice explanation for Quad HDR adaptive binning on a Sony chip that also happens on the sensor before readout and further processing in the SoC:
IMX294CJK | Sony Semiconductor Solutions

There was also a nice explanation for on sensor binning I think on FRAMOS but I don’t find it anymore.

Wouldn't the term RAW subsampling imply pre-processor?

...means „what“??

In the end it‘s all about the result that‘s only „mediocre“ when viewed on a 4K TV.

 

[Dronefriend]

What should „RAW subsampling“ be ;-)?!
This answear is a joke!
If the complete 5.5K sensor is read we would have nothing else than with the P4Pro that downsamples the 5.5K image to 4K output.
As many tests state the 4K FFOV mode only gives resolution of close to 2.7K real image information, horizontal as vertical.
With simple maths this is half the 5.5K resolution of the sensor, so the sensor resolution is divided by two - maybe by lineskipping or pixelbinning (I expect the later since SNR seems to improve a bit in FFOV).

We have some people here that are somewhat deeper in the techs, don‘t expect everyone to be plain stupid.

... this HQ mode.. is just ********. The HQ mode does not use enough area to be brandished as a 1 inch sensor. It is more like FOV mode with 2x digital zoom, all the time.

This fake FOV 4K mode though. Help me understand it if you will. Is it comparable to P4P recording in 2.7K?

 

[Sockeye]

As just posted in the RC-Groups I am starting to think we really have "FOV-Gate" developing. In my test chart pics, it is so obvious that the artefacts in the details in 2,7K and 4K FOV are nearly identical. Having 5472×3648px I presume they do the following:

Take the full sensor (cropped to 16:9)

4K = full sensor with 2x2 pixel binning resulting in 2,7K and let the SOC inflate it to 4K
2,7K full sensor with 2x2 pixel binning resulting in 2,7K
1080 full sensor with 4x4 pixel binning resulting in 1080

This should be extremely fast, quick and dirty to implement and would explain my test results and provide the reason for 2,7K and 1080p being in FOV by default..
What do you guys think?

 

[mozso]

... this HQ mode.. is just ********. The HQ mode does not use enough area to be brandished as a 1 inch sensor. It is more like FOV mode with 2x digital zoom, all the time.

This fake FOV 4K mode though. Help me understand it if you will. Is it comparable to P4P recording in 2.7K?

Test images here for you question:
Mavic Pro-2 vs Zoom and P4P - Line Skipping + Sensor Heat?

 

[yysc]

DJI double screwed customers. For early adopters (like me) who purchased on announcement day, we were promised "superior 10bit 4K" powered by Hasselblad, and we got what we got, the crappy 4K FOV.

Then for those who waited a bit, now both the M2P and the Fly More Combo are an extra 110 euro combined, and the issues are known making it tougher to make the purchase.

I really hope sales take a hit and they're forced to fix 4K FOV. Just use the algorithms used in the P4P, the hardware is the same or better (the SoC). As of today 4K FOV it is basically 2.7K FOV resized.