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

Joined
Apr 26, 2014
Messages
17
Reaction score
18
I have done some lab comparisons between the Mavic-2-Pro, Mavic-2 Zoom and Phantom 4 Pro and have come up with some shocking results.

In my video, I theorize that DJI is doing a "line skipping" and/or "pixel binning" technique that significantly degrades it's image quality in 4k video mode. I have clearly demonstrated this and I and even theorized that this was done and an emergency measure to solve a sensor over-heating problem stemming from a possible design issue of the new Hasselblad 1inch-type sensor camera.

Here are some of the points I discuss in this video:

  • Mavic-2 Pro has a 1 inch-type sensor....but does NOT use that full sensor surface area in either "HQ" or FOV mode. (customers are not "truly" using a full 1inch-type sensor in video mode like the Phantom 4 Pro does)
  • Mavic-2 Pro "line skips" and/or "pixel binns" as a very crude scaling solution to reduce a potential over-heating problem of this sensor? No other explanation for it aside from marketing cripple trick.
  • Mavic-2 Pro does NOT use the full sensor area in "HQ" to again, solve an a possible overheating design flaw on the camera body?
  • Mavic-2 Pro's "4K" FOV wide mode has no more real, actual resolution than it's 2.7k recording mode! (demonstrated in stationary, identical scene recordings with zoom-in examination)
  • I love the Mavic-2 Pro and I am NOT "trashing" it, but I do feel it has been heavily "crippled".

To view this content we will need your consent to set third party cookies.
For more detailed information, see our cookies page.


There is also a full length video on my channel right next to this one.

What do you guys think? 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.

CT
 
Last edited:
Very interesting, MANY thanks for the test, literally an eye-opener! I recently bought my P4P (1st gen) and wanted to do the same - probably I'll still do it, alongside my proven sony rx100iv, because there might be differences in the two p4p generations.

This line skipping story turns the entire m2p vs m2z question upside down, I am certain, that there is no line skipping on the 12MP sensor of the m2z, thus in reality it may use almost the same sensor area as the m2p...
 
Very interesting, MANY thanks for the test, literally an eye-opener! I recently bought my P4P (1st gen) and wanted to do the same - probably I'll still do it, alongside my proven sony rx100iv, because there might be differences in the two p4p generations.

This line skipping story turns the entire m2p vs m2z question upside down, I am certain, that there is no line skipping on the 12MP sensor of the m2z, thus in reality it may use almost the same sensor area as the m2p...
Think about this logically and the simple conclusion must be that to the extent 4K on the M2 depicts the same horizontal FOV as that obtained from a still image there must be some processing to down sample the video. There doesn't have to be line skipping, the full sensor readout may be subject to post processing. In any case there are other P4P and M2P comparisons that suggest very small differences between the two cameras in processed footage. This might be like comparing DLOG footage that has been subject to an in camera profile. The M2 10 bit dynamic range has to be an advantage for grading.
 
Think about this logically and the simple conclusion must be that to the extent 4K on the M2 depicts the same horizontal FOV as that obtained from a still image there must be some processing to down sample the video. There doesn't have to be line skipping, the full sensor readout may be subject to post processing. In any case there are other P4P and M2P comparisons that suggest very small differences between the two cameras in processed footage. This might be like comparing DLOG footage that has been subject to an in camera profile. The M2 10 bit dynamic range has to be an advantage for grading.

Just checked for you, my P4P gives 4% narrower FOV in video compared to stills. This means it uses substantially (more than 2x) more than the necessary 8MP. A bad downsampling strategy would show up in decreased resolution, more noise and worse Moiré as we've seen here.
Any other sites compared low-light performance and Moiré? (I know, low-light wasn't a topic here, but it's an obvious follow-up as low-light performance is even more crippled if pixels are omitted. Also, dynamic range suffers, which might eventually render the 10bit mode of the m2p useless)
 
Last edited:
Just checked for you, my P4P gives 4% narrower FOV in video compared to stills. This means it uses substantially (more than 2x) more than the necessary 6MP. A bad downsampling strategy would show up in decreased resolution, more noise and worse Moiré as we've seen here.
Any other sites compared low-light performance and Moiré? (I know, low-light wasn't a topic here, but it's an obvious follow-up as low-light performance is even more crippled if pixels are omitted. Also, dynamic range suffers, which might eventually render the 10bit mode of the m2p useless)
You might find this interesting for some low light comparisons
To view this content we will need your consent to set third party cookies.
For more detailed information, see our cookies page.
. You can download the actual files and have a really good look if your excited about it.
 
  • Like
Reactions: FlyingUser
Just checked for you, my P4P gives 4% narrower FOV in video compared to stills. This means it uses substantially (more than 2x) more than the necessary 6MP. A bad downsampling strategy would show up in decreased resolution, more noise and worse Moiré as we've seen here.
Any other sites compared low-light performance and Moiré? (I know, low-light wasn't a topic here, but it's an obvious follow-up as low-light performance is even more crippled if pixels are omitted. Also, dynamic range suffers, which might eventually render the 10bit mode of the m2p useless)
Your numbers aren't making sense. Where do you get the "necessary 6MP" from? 4K = 8.8MP. If we take your 4% crop as being correct we potentailly have around 30% more pixels in the sensor readout than required to create the 4K stream, not 2 times more.

As to increased noise in low light downsizing resolution almost always provides for cleaner images. Employing less of the sensor in the image acquisition will probably have the opposite effect for a given chip which will be most obvious in higher resolution chips where the individual photo sites are smaller.
 
You might find this interesting for some low light comparisons. You can download the actual files and have a really good look if your excited about it.

Thx!
Corrected my late-night maths, of course ~9MP for 4k, but the 16:10 crop of the 1" sensor would be significantly more. The 4% crop you can check yourself. However, I compared the video to jpegs, which are already cropped by the dji lens profile.
 
I just cannot believe the M2P is as bad as it is in your video. The footage from the test labs is just **** quality.

Feels more like some firmware fault than actual crippling of the sensor. And it is not like the sensor is new and have caught DJI off guard. It works fine on the P4P which is two years old. Why should it start to over heat on the Mavic 2 Pro?

To be honest I am more inclined to believe you have pulled this off to serve as click bait to get youtube views, rather than the Mavic 2 Pro having a bad sensor. Have studied raw footage from TomsTechTime which looks very good, nothing like that horrific quality in your videos..

Will need more people saying this to believe this.

Edit: After seeing the other video posted up here I can tell that you are not the only one experiencing this. And that you’re not a liar. Let’s hope it is some firmware fault that can be corrected.
 
Last edited:
I just cannot believe the M2P is as bad as it is in your video. The footage from the test labs is just **** quality.

Feels more like some firmware fault than actual crippling of the sensor. And it is not like the sensor is new and have caught DJI off guard. It works fine on the P4P which is two years old. Why should it start to over heat on the Mavic 2 Pro?

To be honest I am more inclined to believe you have pulled this off to serve as click bait to get youtube views, rather than the Mavic 2 Pro having a bad sensor. Have studied raw footage from TomsTechTime which looks very good, nothing like that horrific quality in your videos..

Will need more people saying this to believe this.

Edit: After seeing the other video posted up here I can tell that you are not the only one experiencing this. And that you’re not a liar. Let’s hope it is some firmware fault that can be corrected.
By other video are you referring to the wetalkuav one? The footage seemed good there. Considering it was at 1600ISO- they came up pretty good with a bit of post. The P4 certainly didn’t pull the pants down on the M2.
 
By other video are you referring to the wetalkuav one? The footage seemed good there. Considering it was at 1600ISO- they came up pretty good with a bit of post. The P4 certainly didn’t pull the pants down on the M2.
Yup that one.

It didnt pull the pants down on it but the M2 have this smudgy look. Like it is a tad off focus. The P4P have (has?) got more detail in it. Not only in the low light footage but in the footage from the beach as well. I was suprised to see it because in the files from toms tech time, photos and footage is superb.
 
Did you ever get hands on a p4p2.0 to compare to the p4p? i've been doing tests of the p4p various framerates, and i'm seeing a sizeable difference between regular and slow motion framerates. 24 and 30 appear OK, but anything above that, 48/60, must be binning/skipping, i can see stairstepping and and overall mushier image. i'm wondering if the p4p2.0 had improved on this
 
Did you ever get hands on a p4p2.0 to compare to the p4p? i've been doing tests of the p4p various framerates, and i'm seeing a sizeable difference between regular and slow motion framerates. 24 and 30 appear OK, but anything above that, 48/60, must be binning/skipping, i can see stairstepping and and overall mushier image. i'm wondering if the p4p2.0 had improved on this

Not surprising, when a frame has half the data budget compared to 24/30.
 

Members online

Forum statistics

Threads
143,066
Messages
1,467,358
Members
104,936
Latest member
hirehackers