First video quality comparison - Mavic 2 Pro v.s. Mavic 3

[SkywalkerFeng]

Found this Mavic 3 review (Chinese language only) on Youtube. This is the first review I found that gives a direct comparison of the video quality (actual comparison start at 4:05) between Mavic 2 Pro and Mavic 3.

 

[Cymruflyer]

Need to brush up on my Mandarin...

 

[Dale D]

Found this Mavic 3 review (Chinese language only) on Youtube. This is the first review I found that gives a direct comparison of the video quality (actual comparison start at 4:05) between Mavic 2 Pro and Mavic 3.

No English no watchee

 

[Robert Prior]

Interesting. Thanks for posting this.

Video looks noticeably better than Mavic 2 Pro. Obstacle avoidance also looks to be improved (if I understood correctly, which I might not).

 

[manowar_gub]

You need to compare a2s, not m2p
m2p is old and worse to a2s in everything, except stills and adjustable aperture

 

[IMU]

Wooow!!
I have been very skeptical of this camera's footage until now.
The detail in the city shot is just fantastic!
I now believe that the problem I was focused on, is actually caused by the larger sensor being able capture more light in real time.
Meaning that where the Mavic 2 pro needs an ND 16 filter, this camera will use an ND 32.
Which makes sense once you think about it.
The sensor is larger and captures in a much higher transfer rate, giving it the ability to capture larger amounts of light along with more detail.
Looks to me that the trick is to use a darker filter (than you would use on the Mavic 2 Pro) to calm down the highlights.
That city footage is awesome!!
Attaboy DJI

 

[bkushner]

You need to compare a2s, not m2p
m2p is old and worse to a2s in everything, except stills and adjustable aperture

DJI used the SAME CAMERA from the M2P in the M3.

 

[sar104]

DJI used the SAME CAMERA from the M2P in the M3.

As pointed out in at least one other thread where you posted this - no - it's not even close to being the same camera.

 

[adcimagery]

Wooow!!
I have been very skeptical of this camera's footage until now.
The detail in the city shot is just fantastic!
I now believe that the problem I was focused on, is actually caused by the larger sensor being able capture more light in real time.
Meaning that where the Mavic 2 pro needs an ND 16 filter, this camera will use an ND 32.
Which makes sense once you think about it.
The sensor is larger and captures in a much higher transfer rate, giving it the ability to capture larger amounts of light along with more detail.
Looks to me that the trick is to use a darker filter (than you would use on the Mavic 2 Pro) to calm down the highlights.
That city footage is awesome!!
Attaboy DJI

Sensor size isn’t going to have an effect on exposure like that, just for reference. The smaller sensor may have lower dynamic range, but it’s not capturing more or less light in the way that you’re referring to.

 

[manowar_gub]

DJI used the SAME CAMERA from the M2P in the M3.

Nope, its different sensor

 

[Brojon]

The still images clearly show the improved dynamic range.
I will note that they specifically state Cine mode which has lower contrast similar to Dlog.
I wonder if this is a true apples to apples compare.

 

[canadian drone videos]

Sensor size isn’t going to have an effect on exposure like that, just for reference. The smaller sensor may have lower dynamic range, but it’s not capturing more or less light in the way that you’re referring to.

The pixels in the M3 are twice the size of the M2 pixels. This allows twice as many photons to land on a pixel, yielding better dynamic range and low light performance.

Think of a camera sensor as a telescope (my other hobby is astrophotography) the bigger the scope the more photons enter. A 10” scope will perform much better and faster then a 5” scope because more light is hitting the sensor. But in the case of a drone the aperture size doesn’t change, you’ve increased the surface area of the sensor. This allows more detail in the image, doesn’t make the camera faster.

This is pretty hard to explain especially on a Friday night. If you have a two inch sensor it’s surface area is receiving and recording more photons per second, rendering more detail in any given image under any given conditions as the speed of light is constant. This is why you can blow up a 2 inch sensors image on a big screen and still see excellent results.

 

[sar104]

The pixels in the M3 are twice the size of the M2 pixels. This allows twice as many photons to land on a pixel, yielding better dynamic range and low light performance.

The number of photons collected is determined by the lens and the pixel (photosite) size, but larger pixels have deeper wells and thus better signal-to-noise ratio, which is what improves dynamic range and low-light images.

Think of a camera sensor as a telescope (my other hobby is astrophotography) the bigger the scope the more photons enter. A 10” scope will perform much better and faster then a 5” scope because more light is hitting the sensor. But in the case of a drone the aperture size doesn’t change, you’ve increased the surface area of the sensor. This allows more detail in the image, doesn’t make the camera faster.

That's incorrect. The M3 has a larger physical aperture than the M2 since they are both ƒ 2.8 but the M3 has a longer focal length (12 mm vs 10.3 mm). In terms of incident light per unit solid angle of view the M3 captures around 1.43 times more light. Since the field of view of the M3 is larger with the same number of pixels, it's technically capturing less detail, although the larger pixels mean that you can go to higher ƒ numbers without hitting the diffraction limit.

This is pretty hard to explain especially on a Friday night. If you have a two inch sensor it’s surface area is receiving and recording more photons per second, rendering more detail in any given image under any given conditions as the speed of light is constant. This is why you can blow up a 2 inch sensors image on a big screen and still see excellent results.

The speed of light, eh? Now there's a physical constant that doesn't apply to any of these calculations. Must have been a good Friday night.

 

[canadian drone videos]

The number of photons collected is determined by the lens and the pixel (photosite) size, but larger pixels have deeper wells and thus better signal-to-noise ratio, which is what improves dynamic range and low-light images.

That's incorrect. The M3 has a larger physical aperture than the M2 since they are both ƒ 2.8 but the M3 has a longer focal length (12 mm vs 10.3 mm). In terms of incident light per unit solid angle of view the M3 captures around 1.43 times more light. Since the field of view of the M3 is larger with the same number of pixels, it's technically capturing less detail, although the larger pixels mean that you can go to higher ƒ numbers without hitting the diffraction limit.

The speed of light, eh? Now there's a physical constant that doesn't apply to any of these calculations. Must have been a good Friday night.

If you increase the surface area of a pixel and increase the number of photons the pixel is recording then how is there less detail? We know there is a flaw somewhere in your logic because we know an M3 photo does have more detail, not less as your suggesting. We know this by simply looking at the file size of a photo. The more detail the more data, the more photons were recorded, end result is a photo containing more information.

All characteristics of light are very important when discussing photography and sensors, but I didn’t mention any calculations. Its the aperture setting that determines the amount of light being aloud to hit the sensor, not the lens or pixel size. The sensor on its own doesn’t decide how much light it gets to record.

 

[sar104]

If you increase the surface area of a pixel and increase the number of photons the pixel is recording then how is there less detail?

Because, in the example under discussion (M2 and M3), each pixel in the M3 receives light from a larger solid angle. You are confusing signal-to-noise ratio with detail.

We know there is a flaw somewhere in your logic because we know an M3 photo does have more detail, not less as your suggesting.

No - you don't understand the logic, or optics apparently - that doesn't make it flawed. The M3 has less magnification than the M2, and so you need to be careful to define what you mean by detail. For a given subject at the same distance the M3 has fewer pixels per linear distance at the subject. If you mean resolution - that is typically measured by the ability to resolve line pairs. The number of photons recorded is not the limiting factor here - for an ideal system in focus it is the diffraction limit.

We know this by simply looking at the file size of a photo. The more detail the more data, the more photons were recorded, end result is a photo containing more information.

No - that's entirely incorrect. The level of detail is unrelated to the number of photos detected except when the number is low enough that the noise floor becomes significant relative to the dynamic range.

All characteristics of light are very important when discussing photography and sensors, but I didn’t mention any calculations. Its the aperture setting that determines the amount of light being aloud to hit the sensor, not the lens or pixel size. The sensor on its own doesn’t decide how much light it gets to record.

You wrote:

"If you have a two inch sensor it’s surface area is receiving and recording more photons per second, rendering more detail in any given image under any given conditions as the speed of light is constant."

You can't have that one back. The level of detail has nothing to do with the speed of light.

 

[canadian drone videos]

Because, in the example under discussion (M2 and M3), each pixel in the M3 receives light from a larger solid angle. You are confusing signal-to-noise ratio with detail.

No - you don't understand the logic, or optics apparently - that doesn't make it flawed. The M3 has less magnification than the M2, and so you need to be careful to define what you mean by detail. For a given subject at the same distance the M3 has fewer pixels per linear distance at the subject. If you mean resolution - that is typically measured by the ability to resolve line pairs. The number of photons recorded is not the limiting factor here - for an ideal system in focus it is the diffraction limit.

No - that's entirely incorrect. The level of detail is unrelated to the number of photos detected except when the number is low enough that the noise floor becomes significant relative to the dynamic range.

You wrote:

"If you have a two inch sensor it’s surface area is receiving and recording more photons per second, rendering more detail in any given image under any given conditions as the speed of light is constant."

You can't have that one back. The level of detail has nothing to do with the speed of light.

Your kinda missunderstanding me, after all I did specify it was Friday night and the topic is hard to explain. The only way the 4/3rds sensor will render less detail is if it records less photons, suggesting the speed of light wouldn’t be constant, but we know it is, as a result the mandatory outcome is more detail.

Your trying to teach us that somehow a M2 photo has more detail then an M3 photo, and if that’s not what your trying to teach then I’m confused.

The level of detail is absolutely higher, regardless of the noise floor. That’s why they don’t use 1” sensors to take HD billboard pictures or shoot video meant for the big screen. When you blow it up there isn’t enough detail and it looks bad. A picture taken under perfect conditions with a 1” sensor verses a picture taken under perfect conditions with a 2” are very different, one can be blown up way more then the other, as one has more detail then the other.

Put it this way, my P4P raw photo is 20mb, my x5s raw photo is 40, both are 20mp cameras. The difference is sensor size. If I took the same photo in the same conditions with each camera one photo will be twice the size as the other. The easiest way to judge the amount of detail in a photo is by its file size. The larger the sensor the bigger the file size gets. My iPad mini renders 5mb photo files, with a very small sensor, my Panasonic G7 4/34s is around 30mb. Both photos look fine when I look at them on my phone, when I look at them on a 40” screen there is a huge difference.

I can blow up an M3 photo well beyond the magnification of the M2 and still have better detail. Again, your trying to teach us the M2 photo has more detail then the M3, I can’t think of any way that could possibly be true? You can’t blow up a M2 photo nearly as much as an M3 photo.

At any rate this discussion has run its course, take care.

 

[sar104]

Your kinda missunderstanding me, after all I did specify it was Friday night and the topic is hard to explain. The only way the 4/3rds sensor will render less detail is if it records less photons, suggesting the speed of light wouldn’t be constant, but we know it is, as a result the mandatory outcome is more detail.

The speed of light has nothing to do with how many photons hit the sensor or the individual pixels. Please stop digging.

Your trying to teach us that somehow a M2 photo has more detail then an M3 photo, and if that’s not what your trying to teach then I’m confused.

Yes, you are very confused.

The level of detail is absolutely higher, regardless of the noise floor. That’s why they don’t use 1” sensors to take HD billboard pictures or shoot video meant for the big screen. When you blow it up there isn’t enough detail and it looks bad. A picture taken under perfect conditions with a 1” sensor verses a picture taken under perfect conditions with a 2” are very different, one can be blown up way more then the other, as one has more detail then the other.

Continually just repeating "more detail" is not an argument. As I said, in good light that's only true if the resolution on the smaller sensor is diffraction limited. The gains in sensor size are low-light signal to noise, small-aperture resolution (the diffraction limit issue), and often that the larger sensors can simply accommodate more photosites without the previous problems manifesting.

Put it this way, my P4P raw photo is 20mb, my x5s raw photo is 40, both are 20mp cameras. The difference is sensor size. If I took the same photo in the same conditions with each camera one photo will be twice the size as the other. The easiest way to judge the amount of detail in a photo is by its file size. The larger the sensor the bigger the file size gets. My iPad mini renders 5mb photo files, with a very small sensor, my Panasonic G7 4/34s is around 30mb. Both photos look fine when I look at them on my phone, when I look at them on a 40” screen there is a huge difference.

No, the difference is not sensor size, it's bit depth. That represents a dynamic range difference, not a detail difference.

I can blow up an M3 photo well beyond the magnification of the M2 and still have better detail. Again, your trying to teach us the M2 photo has more detail then the M3, I can’t think of any way that could possibly be true? You can’t blow up a M2 photo nearly as much as an M3 photo.

Again - only subject to the mechanisms above. And don't misunderstand - I completely agree that the M3 4/3 sensor has measurable advantages over the M2 1" sensor. It's just that your explanations for that are almost completely incorrect.

 

[SkywalkerFeng]

Continually just repeating "more detail" is not an argument. As I said, in good light that's only true if the resolution on the smaller sensor is diffraction limited. The gains in sensor size are low-light signal to noise, small-aperture resolution (the diffraction limit issue), and often that the larger sensors can simply accommodate more photosites without the previous problems manifesting.

No, the difference is not sensor size, it's bit depth. That represents a dynamic range difference, not a detail difference.

All things equal, a bigger sensor with better dynamic range can capture more information in both highlight and shadow area. That capability does translate into more details in final images captured.

 

[sar104]

All things equal, a bigger sensor with better dynamic range can capture more information in both highlight and shadow area. That capability does translate into more details in final images captured.

It can, depending on light levels, lead to more detail across a wider dynamic range. I already pointed that out.

 

[Robert Prior]

A picture taken under perfect conditions with a 1” sensor verses a picture taken under perfect conditions with a 2” are very different, one can be blown up way more then the other, as one has more detail then the other.

Not necessarily.

Suppose you have a 1" sensor that is 3000 pixels wide — you can blow it up to a 10" image at 300 dpi. Now suppose you have a 2" sensor that is only 1500 pixels wide. The resulting image will only be 5" at the same print resolution. The larger sensor has much larger pixel sites, and assuming the same technology there would be far more dynamic range and less noise, but you would see pixellation at much smaller sizes than with the smaller sensor.