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

[Cliff_622]

We've known for a while now that the M2P uses the Ambarella H3 unless there is another processor in the M2P with a giant "H3" logo written on it:

There are many factors that contribute to heat, not just the processor.

I have seen this photo before. However, Im 100% sure its real or not. It could be but I just have a weird feeling that it might not be. If the M2-Pro uses Ambarella's most powerfull processor than that's excellent news. "If" there really is a problem with the camera sensor getting too hot, then, that would be very embarassing for DJI's engineering and design team. If their original goal was to install an extremely powerfull image processor and then they got trapped into NOT being able to use its full readout and scaling engine because of bad sensor thermal oversights? Oh crap,...that would be horrible for DJI's R&D team.

I dont know....I really am mystified at why this happened this way.

 

[Lon Denard]

I have seen this photo before. However, Im 100% sure its real or not. It could be but I just have a weird feeling that it might not be. If the M2-Pro uses Ambarella's most powerfull processor than that's excellent news. "If" there really is a problem with the camera sensor getting too hot, then, that would be very embarassing for DJI's engineering and design team. If their original goal was to install an extremely powerfull image processor and then they got trapped into NOT being able to use its full readout and scaling engine because of bad sensor thermal oversights? Oh crap,...that would be horrible for DJI's R&D team.

I dont know....I really am mystified at why this happened this way.

Well, obviously there's a bottleneck somewhere. It really can only be one of two things, hardware or software. Some have suggested the hardware is the problem, either the chip or the sensor can't handle the loaf. Others have suggested that it's the software, as in DJI is crippling the camera on purpose to protect the P4P. Of course, then you have idiots like myself that don't really see that big of a problem in the first place.

Again, this has been a very interesting discussion for sure. What strikes me as a problem for DJI is that this is a no win situation for them. If they did somehow overcome this "problem" with an update, they will be accused of holding back in the first place. If they can't then it's an embarrassment to the company. Especially if someone else does it.

 

[Category 5]

Well, obviously there's a bottleneck somewhere. It really can only be one of two things, hardware or software. Some have suggested the hardware is the problem, either the chip or the sensor can't handle the loaf. Others have suggested that it's the software, as in DJI is crippling the camera on purpose to protect the P4P. Of course, then you have idiots like myself that don't really see that big of a problem in the first place.

Again, this has been a very interesting discussion for sure. What strikes me as a problem for DJI is that this is a no win situation for them. If they did somehow overcome this "problem" with an update, they will be accused of holding back in the first place. If they can't then it's an embarrassment to the company. Especially if someone else does it.

I'm not so sure of this. I have been looking into dumping my Canon DSLR gear in favor of a mirrorless setup that can do 4K video too, and with the exception of the A7iii and the GH5 almost all of the cameras use a significant crop to get decent UHD video. Even the A7Riii looks like crap when doing a full sensor readout, but looks great in crop mode. Apparently it's not as simple as we think. Of all imaging companies I'd have expected Canon to pull it off, and their implementation is one of the worst (1.8 crop on their new mirrorless gem) . With all of the cameras on the market, and all of the imaging horsepower attached to them only 2 cameras manage to get a non crop UHD image that's worth a ****.

After a few days of reading I am coming away more impressed with the results DJI are getting from these small cameras, despite their shortcomings. I'd certainly take 24/25 fps 10 bit and 30 8 bit selectable modes if they could improve sharpness, and I do think that with time they can find more detail in the pro. Time will tell.

I still think the overall image (sharpness aside) of the M2P looks the most cinematic and professional of the consumer drones I've seen. For example, Zenmuse x7 footage looks just breathtaking, and holds that characteristic aside from its sharpness. There is something more refined in the M2P output too, that looks more cinematic, and less video-camera-like to me. This is aside from the sharpness of FFOV mode, which I agree could look better compared to the P4P and the RX100 series.

 

[prothorin]

There is no one who would disassemble the Mavic 2 to check what kind of chip it is in?

 

[prothorin]

Theory: Why so easy change camera on Mavic 2?

 

[Dronefriend]

Theory: Why so easy change camera on Mavic 2?

That's a question, not a theory.

 

[WithTheBirds]

That's a question, not a theory.

Astounding observation. Not everyone has English as a first language. Apologies @prothorin if my assumption is off here.

My take on this is it’s a suggestion that perhaps in making the Zoom and Pro gimbals function with common electronics inside the fuselage a commoromise needed to be made.

It’s as good a theory as many offered to date.

 

[Dronefriend]

Astounding observation. Not everyone has English as a first language. Apologies @prothorin if my assumption is off here.

My take on this is it’s a suggestion that perhaps in making the Zoom and Pro gimbals function with common electronics inside the fuselage a commoromise needed to be made.

It’s as good a theory as many offered to date.

Not trying to make him feel bad about choosing the wrong word. It's just very short info to be a theory.

 

[Lon Denard]

Astounding observation. Not everyone has English as a first language. Apologies @prothorin if my assumption is off here.

My take on this is it’s a suggestion that perhaps in making the Zoom and Pro gimbals function with common electronics inside the fuselage a commoromise needed to be made.

It’s as good a theory as many offered to date.

Well as someone that endeavors to keep a positive attitude about DJI's motives for doing things, I'm hoping that it's to future proof the drone. You know, like Autel promised but never delivered with the X Star. Or... We could, just this once, take their word for it, that it's so the cameras can be swapped out. Or... It's a "trickle down" feature from the Enterprise model that is designed to have things added and taken away. I'm willing to bet good money that the Mavic 2 Enterprise will have a version of the Dual camera available.

 

[WithTheBirds]

Well as someone that endeavors to keep a positive attitude about DJI's motives for doing things, I'm hoping that it's to future proof the drone. You know, like Autel promised but never delivered with the X Star. Or... We could, just this once, take their word for it, that it's so the cameras can be swapped out. Or... It's a "trickle down" feature from the Enterprise model that is designed to have things added and taken away. I'm willing to bet good money that the Mavic 2 Enterprise will have a version of the Dual camera available.

I don’t think the FOV issue that concerns many has anything to do with the fact both cameras are supported. This could be simply managed by a digital ID in the camera read by the image processing circuitry on initialisation.

I was simply trying to guess what the theory might be.

 

[Dronefriend]

Is it possible to rule out the SoC specs, whichever it is, as the reason for the binning using this logic:

A) SoC treats incoming pixels as incoming pixels wherever they're from
B) The Zoom version can handle 4K fine using same SoC
C) Because of this the weak part of the chain has to be the Pro camera.

 

[WithTheBirds]

Is it possible to rule out the SoC specs, whichever it is, as the reason for the binning using this logic:

A) SoC treats incoming pixels as incoming pixels wherever they're from
B) The Zoom version can handle 4K fine using same SoC
C) Because of this the weak part of the chain has to be the Pro camera.

I think the answer lies in the supported readout modes of the sensors.

The 1/2.3” sensor in the zoom supports full readout (no line addition) at 17:9 59.97 fps. The 1” sensor can’t do this. Perhaps all we might be dirty on DJI for is not employing a higher spec sensor. We probably can’t fairly allege they have deliberately crippled the hardware.

 

[Dronefriend]

How come the sensor can do it on the P4P and not the M2P?

 

[peros550]

Probably the guy wanted to say:

"Conspiracy Theory:Why so easy change camera on Mavic 2? "

which I hope it's true actually. If the issue is on the sensor, and if that theory is not just a theory, then in the future DJI could announce a new camera module, and we could just upgrade the camera module instead of the whole drone.

If the price was reasonable, I would prefer that!

 

[WithTheBirds]

How come the sensor can do it on the P4P and not the M2P?

Can it? Is it just additional software sharpening?

 

[sar104]

I think the answer lies in the supported readout modes of the sensors.

The 1/2.3” sensor in the zoom supports full readout (no line addition) at 17:9 59.97 fps. The 1” sensor can’t do this. Perhaps all we might be dirty on DJI for is not employing a higher spec sensor. We probably can’t fairly allege they have deliberately crippled the hardware.

The Sony RX100, which has a similar (identical?) sensor, will only do 4k video for 5 minutes due to heating issues. Not at all surprising that in the Mavic there would be similar problems.

 

[Dronefriend]

Sony would be the one to address that question.

Yes but lets speculate. Older version of sensor on P4P? Is it the new 10bit mode on the new version that does not support this? I seem to recall I read that somewhere.

 

[sar104]

How come the sensor can do it on the P4P and not the M2P?

Is it the same sensor?

 

[WithTheBirds]

The Sony RX100, which has a similar (identical?) sensor, will only do 4k video for 5 minutes due to heating issues. Not at all surprising that in the Mavic there would be similar problems.

Ok- so I would say it’s fair to assume then DJI is faced with a sensor limitation in the implementation. Not a deliberate attempt to cripple the device spec or poor thermal engineering on their part.

 

[Dronefriend]

Personally I have never believed them to cripple it on purpose. That would be just plain weird to do.

A "non-crippled" M2P could never compete with Inspire 2 anyways and since P5 is coming up, and P5 will be better than M2P, is makes no sense at all.


Why would DJI go Hmm.... the M2P is a little bit better than the P4P.. You can barely notice it but it is... we cannot allow this... we must make it little bit worse instead.... yet, you can barely notice it... Yes, this makes sense... lets spend a couple of thousands of dollars of R&D on this crippling.... even though there is barely any difference...

No, not at all. It is not intentional, that would be insane.