Received Mavic 3 today. Liking it for the most part, but optics arent a huge improvement over my 2 pro

[Gemini8026]

Flies great. Great battery life. Tried a few sample photos and it looked fine. However Im a pixel peeper and trying to verify my lens is centered decently. Anyone have good tips for testing it on this guy?

I had sent back a 1st run Mavic 2 pro when they came out because my left side was a mess in the corners. Im not sure how much QC has been put into these cameras. I shoot my ground stuff with a D850 and Nikon lenses so I am spoiled by that level of quality.

Anyhow. Im more of a photographer then a movie guy so Im a little partially disappointed that it doesnt appear to take sharper photos then the 2 pro. Anyone else notice this in comparing? Anyone else find their lens decentered at all?!? I think mine is ok, but trying to verify the left side isnt softer then the right and need daytime light (dark gloomy day here in Sk, Canada) to properly test.

My expectations werent anything ridiculous but for the money I was hoping the optics would be a step up. Everything else appears to be.

Any thoughts/comments?

Jeff

 

[CactusJackSlade]

Watching!...

 

[Jfvdenning]

Fly up 70-100m and shoot straight down at some with with detail across the frame, shoot at f2.8, f4 and f5.6, that will tell you how good your lens is.

The Mavic 2 Pros has terrible quality control with the optics, I tested 20 to get the 2 good ones I have now. From looking at reviews the Mavic 3 seems to have the same issues, but possibly worse since it is a wider lens and the corners will always be soft at f2.8.

 

[globetrotterdrone]

@Gemini8026
I think, that a top down shot with variation on aperture on something with a lot of texture and no moving parts is the easiest way. Or the famous brick wall.

Besides the bigger sensor and having the upper hand (in theory) with DNR, I am also concerned, if the QA and the soft spots are something, that DJI worked on. But I doubt that. So, my thought is, you get an advantage in low light but I wouldn't assume much of a deal at daylight conditions. Resolution is the same and more FOV is often harder to correct (and you need mostly to step down to f 4 now with the increased image circle). I pretty much guess, most can be done in post.

At least try to exclude you have some gherkin aka lens problem, but I would expect similar results at least with the Air 2S, if you want to compare with this and a neglectable differences with the M2P.

 

[Jfvdenning]

@Gemini8026
I think, that a top down shot with variation on aperture on something with a lot of texture and no moving parts is the easiest way. Or the famous brick wall.

Besides the bigger sensor and having the upper hand (in theory) with DNR, I am also concerned, if the QA and the soft spots are something, that DJI worked on. But I doubt that. So, my thought is, you get an advantage in low light but I wouldn't assume much of a deal at daylight conditions. Resolution is the same and more FOV is often harder to correct (and you need mostly to step down to f 4 now with the increased image circle). I pretty much guess, most can be done in post.

At least try to exclude you have some gherkin aka lens problem, but I would expect similar results at least with the Air 2S, if you want to compare with this and a neglectable differences with the M2P.

A brick wall isn't a good test for a drone as it is very close and is not that easy to see flaws, and you will most likely be much farther away in real world shooting. Optics can behave differently depending on where it is focussed. 70-100m away with fine detail is best. Above lots of houses or trees.

I shoot my tests like this.



Whilst the larger sensor will help with low light, creating a high quality lens for it in a tiny package will be harder adding that it is also wider, and so you may have to stop down more for a sharper image with the M3, meaning for night shots there is little benefit over the M2P since I can get edge to edge sharpness at f2.8 with my M2P.
I can't imagine the 4k footage looking great if you get a flawed lens.

 

[globetrotterdrone]

I would expand my original thought by: a brick will should not be the only test, right. For checking if the focus may work up close and with other top down shots for infinity, it may still be handy tbh.

Everything else I am completely fine to second.

 

[Saladshooter]

A good way to check the optics on a drone is to take multiple pictures of the same object placed left, center, and right in the frame then compare. This will only give you a basic idea of distortions and varying softness but will show up issues better than trying to subjectively compare the left and right sides of a picture that has significant differences. On the ground you can try using a printed target or object that fills the entire frame. All tests will be pretty rudimentary but will show gross problems in the lens.

 

[Surge]

We need to differentiate between the lens design and production quality control.
No lens will be sharp from edge to edge, especially wide open! I shoot Leica M lenses semi-professionally - they make the best lenses; but even their $10k+ monster lenses are not perfectly sharp edge to edge.

Variation in production/QC is another matter.

Also, the larger sensor in the Mavic 3 will provide a lot more benefit than just low light performance. Look for dynamic range improvements (i.e., the transition between dark to light).

 

[vindibona1]

Here's something that I've not seen mentioned on this forum: The LENS quality. You can put whatever sensor you want in a camera but if the lens isn't up to the task of taking advantage of the sensor, you can only get the best of the weakest link.

While technology might have improved things, but only a few years ago it was said that most good DSLR lenses wouldn't make much with a megapixel count greater than 15mp. While I am sure that the advanced GLASS lenses have improved things I have to wonder where the law of diminishing return is for sensors? And what happens to the optical resolution when you use a plastic lens and then put a plastic filter on top of it????

If I could draw a parallel example... When early in my career I went straight to medium format, even in the 80's and early 90's each (fixed focal length) lens alone could cost up to $4000 (in 1980's dollars) to make sure the optics took full advantage of the larger film format's resolution and enlargement capabilities...

... So in the case of the M3, we have a larger 4/3 sensor, with (presumably based on the FOV differences) essentially the same lens as that of the M2. While we know the sensor will be an improvement in low light I have to question how much improvement there is in imaging on a practical level? Just something to think about.

 

[Ralph thompson]

I have an Inspire 1 Pro, it has a micro 4/3” sensor & uses good quality panasonic lens’. The differance with the M2P is visible although I’d say acedemic. The lens makes a big difference, the M2P & M3 have small lens & this will reduce the quality slightly over the I1P. I’m sure Hasselblad did an amazing job but the physics limits quality in a small lens. However the only way to determine if there is an issue with quality & quantify it, whether, design, manufacturing or assembly is to use a test card. See the B&H video. Just photographing trees from above or brick walls is kids play, subjective, there are multiple factors that will effect quality, not least would be leaves fluttering in the breeze.

How to Test Your Lens

The goal of this article is to give you some pointers on how you can do some basic lens testing at home.

www.bhphotovideo.com

 

[CanadaDrone]

So far I haven't seen any controlled testing done with multiple samples of both drones, so until that happens we don't really know much of anything. I am very curious to see that though.

It's the same thing in the 'traditional' photography world - especially with lenses, test methodology is extremely important to get usable results. This is why the most reliable results always come from rental companies, who have the unique ability to test a wide range of samples in a controlled manner - something the average user simply cannot reasonably do. Occasionally you get a dud, and occasionally you get a lens that performs significantly better than the norm - that will always be the case.

Things like lens decentering are normal to a certain degree - every lens I have ever bought over the last 20+ years has not been perfect in that regard. It's also more common on wide angle lenses. $17,000 DSLR lenses are not immune to it either and are not as good in the corners as they are in the center of the frame. If there is a major problem, obviously that needs to be addressed, but when someone says they have gone through dozens of samples to get a good one, that is far more likely to be an error in testing methodology rather than some sort of major widespread QC issue. I've owned a M2P, examined files from many others, and I've owned an original Air, and none had lens issues - am I just incredibly lucky? Probably not. If you are able to see tiny flaws at 200% magnification or whatever, that is not an issue. Also, as others have touched on, of course there had to be some compromises made to get a 4/3 sensor on the M3 without using a comparatively huge M43 mount like on the Inspire line. I don't doubt that an Inspire 2 with an X5S and high-end M43 lens produces a better image, but size and cost are in a different league. Everything is a compromise in the photography world, and it's rare that something gets miniaturized and performs as well as the 'full size' version.

The other issue I see come up all the time in the 'traditional' photography world is setting up comically extreme testing scenarios to try and coax an issue out of the product (not saying that necessarily what is happening here). When that happens, of course you are going to find issues, and then when the warranty department tells those people there is nothing wrong, they get mad and take to the internet. Two things that come to mind that I have seen many times in the past are people taking 30 minute exposures of the inside of their lens caps at maximum ISO to claim issues with hot pixels, and people taking images of white paper at F32, and then maxing out the clarity slider in photoshop to claim their camera has an issue with dust or debris on the sensor. Neither scenario would result in any issues whatsoever during normal usage, and that sort of thing is a major source of frustration for warranty departments.

Anyway, I look forward to seeing some proper testing of the two.

 

[CanadaDrone]

Here's something that I've not seen mentioned on this forum: The LENS quality. You can put whatever sensor you want in a camera but if the lens isn't up to the task of taking advantage of the sensor, you can only get the best of the weakest link.

This is true to a point, as a lens only affects sharpness - it does not affect other properties of an image sensor such as DR, readout speed, or ISO performance, all of which contribute to it's image quality.

While technology might have improved things, but only a few years ago it was said that most good DSLR lenses wouldn't make much with a megapixel count greater than 15mp. While I am sure that the advanced GLASS lenses have improved things I have to wonder where the law of diminishing return is for sensors? And what happens to the optical resolution when you use a plastic lens and then put a plastic filter on top of it????

You might be surprised how well sensor resolution can overcome an average quality lens. For example if we took a 45MP camera and a 24MP camera, all else equal, the 45MP image will always look either the same or better than the 24MP image (never worse). And if you were to print both images, you still have a 45MP file even if the lens wasn't the best, and it would produce a better print than the 24MP image, all else equal. The advantages of using the best possible lens are the same regardless of sensor resolution. Obviously if you put total garbage in front of the sensor, that is a different story, but that isn't a realistic scenario and it would affect a lower resolution sensor in exactly the same way.

With mainstream cameras reaching 45, 50, and even 60MP these days, and even higher in the Medium Format world, we have seen time and time again how higher resolution has a wide range of image quality benefits even when using old film lenses or lenses that were not designed with such sensors in mind.

If I could draw a parallel example... When early in my career I went straight to medium format, even in the 80's and early 90's each (fixed focal length) lens alone could cost up to $4000 (in 1980's dollars) to make sure the optics took full advantage of the larger film format's resolution and enlargement capabilities...

The very best lenses will always be too large and/or too expensive to be practical for the average user. If you want a lens with the fewest optical compromises, it will always be comparatively larger, heavier, and more expensive. Let's say you want to get to 600mm on your DSLR, you can do so for ~$1,500 or ~$15,000.

... So in the case of the M3, we have a larger 4/3 sensor, with (presumably based on the FOV differences) essentially the same lens as that of the M2. While we know the sensor will be an improvement in low light I have to question how much improvement there is in imaging on a practical level? Just something to think about.

The M3 lens is a completely different focal length and is designed around a completely different sensor and gimbal, so it's unlikely they share many similarities at all. The sensor's aspect ratio isn't even the same as the M2P. Every sensor performs well in ideal conditions, where you will notice the biggest difference is in more challenging shooting scenarios or in more extreme post processing scenarios.

 

[Ralph thompson]

Agree with most above but with respect, MP is not a defining parameter of quality. Sorry a couple of niggly points. I have an old 12MP Nikon D3 (a 2008 era pro full format DSLR), my go to workhorse and a 45MP Nikon D810 (full format DSLR). The 12MP D3 is miles ahead of the 45 MP D810, with the same f2.8 lens, in low light conditions with only a small increase above native ISO. Its also argued that the colour reproduction on the D3 is also superior, thats a whole discussion unto itself. The reason is that cramming more pixels into the same sized sensor introduces more noise through the close proximity of sensor connections. Studies show that quality is directly related to sensor size (assuming all other factors are the same), large sensors (not number of pixels) produce better quality images. Also, a factor in lens price is not just quality but speed i.e. aperture size, a 600mm f4 full format lens is a honking big lens... but all somewhat remote to the M3 discussion ). The lens and its coatings are unquestionably a significant factor in quality. But Hassalblad is arguably a world leader in camera technology, they know what they're doing. I have little doubt they've done an outstanding job on the M3 considering the incredible constraints. Otherwise totally agree with CanadDrone, thank you for taking the time to join the discussion.

 

[CanadaDrone]

Agree with most above but with respect, MP is not a defining parameter of quality. Sorry a couple of niggly points. I have an old 12MP Nikon D3 (a 2008 era pro full format DSLR), my go to workhorse and a 45MP Nikon D810 (full format DSLR). The 12MP D3 is miles ahead of the 45 MP D810, with the same f2.8 lens, in low light conditions with only a small increase above native ISO. Its also argued that the colour reproduction on the D3 is also superior, thats a whole discussion unto itself. The reason is that cramming more pixels into the same sized sensor introduces more noise through the close proximity of sensor connections. Studies show that quality is directly related to sensor size (assuming all other factors are the same), large sensors (not number of pixels) produce better quality images. Also, a factor in lens price is not just quality but speed i.e. aperture size, a 600mm f4 full format lens is a honking big lens... but all somewhat remote to the M3 discussion ). The lens and its coatings are unquestionably a significant factor in quality. But Hassalblad is arguably a world leader in camera technology, they know what they're doing. I have little doubt they've done an outstanding job on the M3 considering the incredible constraints. Otherwise totally agree with CanadDrone, thank you for taking the time to join the discussion.

Those higher MP bodies downsampled to 12MP or compared at an otherwise equivalent resolution do not look significantly different than the lower MP bodies at higher ISOs. At base or close to base ISO, the higher MP bodies become so much better than something like a D3 that it isn't even really comparable, as that is where they are specifically designed to perform their best and also why their base ISOs are set at 64 rather than 100. The sensors in the flagship D(X) bodies are specifically optimized for a combination of speed and high ISO shooting rather than maximum DR and resolving power.

The D810 is 36MP and not 45MP (the D850 is 45MP) - not sure which one you have (I will assume D850) but here is a resolution-normalized comparison with the oldest camera they have in the comparison list (D4), which has better ISO performance than a D3:

Studio shot comparison: Digital Photography Review

Expert news, reviews and videos of the latest digital cameras, lenses, accessories, and phones. Get answers to your questions in our photography forums.

www.dpreview.com

Looking at the above, the D850 appears to be doing the best job of the bunch, or at the very least, not worse than the D4 - therein lies the beauty of high resolution sensors, so long as other constraints such as (usually) lower FPS and larger file sizes are not major issues for your work, you more/less get the best of both worlds. Dynamic range and ISO performance hit their practical limits about 10 years ago, and it will remain that way until there is a major change with regards to current sensor design. They way manufacturers improve upon sensors these days is by achieving approximately the same performance while increasing resolution and/or readout speed. Readout speed is particularly important for some applications, especially as we move into a world of shutterless professional cameras such as the Nikon Z9, where they can now operate as fast as around 1/270sec, matching or surpassing most mechanical shutters. A camera like that also removes any speed or file size limitations of the past, and truly becomes a "do everything" camera.

Regarding color reproduction, the D850 can capture a greater color depth than the D3, so any differences you are seeing there would be processing or RAW converter related rather than a physical limitation of the camera. The D3 was a phenomenal camera, and along with the D700 and D300 of the same era, was arguably responsible for much of Nikon's success in the digital world, but sensor technology has come quite a long way since 2007-8. For example back side illuminated sensors with gapless microlenses (like the D850 has) moves virtually all of that noisy circuitry behind the imaging layer of the sensor, allowing for much more efficient light collection and faster readout speeds.

 

[Surge]

Just had my first flight with the M3. I'm very impressed with the image quality, considerably better than the M2P.
When I shoot for work, it's with a Leica M. I tried Nikon glass with a D800, but the glass just isn't as good as Leica (in my view).
Anyway, I never used the M2P for work because the image quality was just not good enough when compared to the Leica M glass, so clients hated the drone shots because they didn't look "pro" enough.
The M3 is the first compact drone that comes close to the level of quality that you can get from a Nikon/Canon/Leica DSLR.

I'm not going to do any back to back shots with the M2P and the M3, as I just don't have the time, and I can immediately tell the quality (sharpness, dynamic range, contrast) is better.

 

[Ralph thompson]

Sorry yr right, D810 is 36MP. But the D3 & D3s do not downsample and all these have 14 bit RAW, effectively the same bit depth. D810/D850 very little difference in practice. All camera manufacturers stopped focusing on stills improvements more than a decade ago, all their effort is on video. Anyway this is off topic. I have little doubt the M3 produces better images than the M2P but whether that would be observable to anyone but a super critical eye, I doubt it. I've been using drones professionally since 2012 using the Phantom 0 with a Go Pro and have never had a client complain about quality. I'm sure the M3 cine would also be an improvement, most notably with ProRes 442 video in advertising and movies but it will not get me $6K worth of additional work in the next year so will be passing for now. If Gemini8026 can see flaws in the still images, then in all fairness, he/she should take them to DJI who have an amazing support service and I'm sure would want to know if they have a quality issue. But I would get a good quality test card and send photographs of the test pattern to DJI, not trees or brick walls. I saw a video of DJI's quality tests (can't find it at this moment) but the tests are incredible and thorough.

 

[CanadaDrone]

Sorry yr right, D810 is 36MP. But the D3 & D3s do not downsample and all these have 14 bit RAW, effectively the same bit depth. D810/D850 very little difference in practice. All camera manufacturers stopped focusing on stills improvements more than a decade ago, all their effort is on video. Anyway this is off topic. I have little doubt the M3 produces better images than the M2P but whether that would be observable to anyone but a super critical eye, I doubt it. I've been using drones professionally since 2012 using the Phantom 0 with a Go Pro and have never had a client complain about quality. I'm sure the M3 cine would also be an improvement, most notably with ProRes 442 video in advertising and movies but it will not get me $6K worth of additional work in the next year so will be passing for now. If Gemini8026 can see flaws in the still images, then in all fairness, he/she should take them to DJI who have an amazing support service and I'm sure would want to know if they have a quality issue. But I would get a good quality test card and send photographs of the test pattern to DJI, not trees or brick walls. I saw a video of DJI's quality tests (can't find it at this moment) but the tests are incredible and thorough.

You wouldn't want to donwsample the D3 file, you would end up with even lower resolution than the 12MP, however if you did, the same principles would apply if comparing to a lower resolution camera of say 6MP or 8MP (the D3 would not look worse). What I am saying is that if you have a D810 or D850, you can downsize that image to 12MP to match the D3, and you will get similar, often superior, image quality (as shown in the comparison above). Because you cannot increase the resolution of the D3 to match the D8XX series in the opposite case, that is a huge advantage to higher resolution sensors. You get the best of both worlds, from an image quality standpoint. The main takeaway is all else equal, the higher MP sensors are never any worse than the lower MP sensors when resolutions are normalized, yet they retain all the benefits of a higher resolution sensor because it is one. With a D3 you're stuck at 12MP, with a DXX body, you get 36 or 45MP if you want it, or you can simply downsample those images to 12MP and they will not be worse than a native D3 image. I hope I am explaining that clearly. The flagship D(X) body advantages are almost entirely related to higher FPS, smaller file sizes, ergonomics, and the ability to shoot at ridiculously high ISOs like 3,000,000, which is more useful than it sounds for things like surveillance applications (i.e. reading a license plate in the dark).

The D8XX bodies can simply capture more color depth than a D3, that is measurable. The fact that they both are capable of 14bit shooting is not the whole story there. Regardless, if you are having issues with color fidelity, it can be corrected on the processing side because the information is all there in the NEF.

Back on topic haha - regarding the M2 vs M3 image quality, one simply needs to compare Sony's 1" 20MP sensor with Sony's 4/3" 20MP sensor which both exist in several cameras today and those are the sensors used in the M2P and M3 respectively. While not a perfect comparison, the ISO and DR advantages are visible and are in line with samples from the M2 and M3 that people have posted so far. In ideal conditions, the difference between the two sensors will be very little, but in challenging conditions or with aggressive post processing, the differences become more noticeable.

 

[Jet skier]

Flies great. Great battery life. Tried a few sample photos and it looked fine. However Im a pixel peeper and trying to verify my lens is centered decently. Anyone have good tips for testing it on this guy?

I had sent back a 1st run Mavic 2 pro when they came out because my left side was a mess in the corners. Im not sure how much QC has been put into these cameras. I shoot my ground stuff with a D850 and Nikon lenses so I am spoiled by that level of quality.

Anyhow. Im more of a photographer then a movie guy so Im a little partially disappointed that it doesnt appear to take sharper photos then the 2 pro. Anyone else notice this in comparing? Anyone else find their lens decentered at all?!? I think mine is ok, but trying to verify the left side isnt softer then the right and need daytime light (dark gloomy day here in Sk, Canada) to properly test.

My expectations werent anything ridiculous but for the money I was hoping the optics would be a step up. Everything else appears to be.

Any thoughts/comments?

Jeff

Did they Finally get rid of the gimbal jump ?

 

[Jfvdenning]

My Mavic 2 Pro is sharp edge to edge wide open, as is my Sigma 65mm f2 lens, and 100-1400mm. These are all cheap lenses.

We need to differentiate between the lens design and production quality control.
No lens will be sharp from edge to edge, especially wide open! I shoot Leica M lenses semi-professionally - they make the best lenses; but even their $10k+ monster lenses are not perfectly sharp edge to edge.

Variation in production/QC is another matter.

Also, the larger sensor in the Mavic 3 will provide a lot more benefit than just low light performance. Look for dynamic range improvements (i.e., the transition between dark to light).

 

[Jfvdenning]

So far I haven't seen any controlled testing done with multiple samples of both drones, so until that happens we don't really know much of anything. I am very curious to see that though.

It's the same thing in the 'traditional' photography world - especially with lenses, test methodology is extremely important to get usable results. This is why the most reliable results always come from rental companies, who have the unique ability to test a wide range of samples in a controlled manner - something the average user simply cannot reasonably do. Occasionally you get a dud, and occasionally you get a lens that performs significantly better than the norm - that will always be the case.

Things like lens decentering are normal to a certain degree - every lens I have ever bought over the last 20+ years has not been perfect in that regard. It's also more common on wide angle lenses. $17,000 DSLR lenses are not immune to it either and are not as good in the corners as they are in the center of the frame. If there is a major problem, obviously that needs to be addressed, but when someone says they have gone through dozens of samples to get a good one, that is far more likely to be an error in testing methodology rather than some sort of major widespread QC issue. I've owned a M2P, examined files from many others, and I've owned an original Air, and none had lens issues - am I just incredibly lucky? Probably not. If you are able to see tiny flaws at 200% magnification or whatever, that is not an issue. Also, as others have touched on, of course there had to be some compromises made to get a 4/3 sensor on the M3 without using a comparatively huge M43 mount like on the Inspire line. I don't doubt that an Inspire 2 with an X5S and high-end M43 lens produces a better image, but size and cost are in a different league. Everything is a compromise in the photography world, and it's rare that something gets miniaturized and performs as well as the 'full size' version.

The other issue I see come up all the time in the 'traditional' photography world is setting up comically extreme testing scenarios to try and coax an issue out of the product (not saying that necessarily what is happening here). When that happens, of course you are going to find issues, and then when the warranty department tells those people there is nothing wrong, they get mad and take to the internet. Two things that come to mind that I have seen many times in the past are people taking 30 minute exposures of the inside of their lens caps at maximum ISO to claim issues with hot pixels, and people taking images of white paper at F32, and then maxing out the clarity slider in photoshop to claim their camera has an issue with dust or debris on the sensor. Neither scenario would result in any issues whatsoever during normal usage, and that sort of thing is a major source of frustration for warranty departments.

Anyway, I look forward to seeing some proper testing of the two.

I have been through 20 Mavic 2 pro drones to find 3 good ones, it was not an error in testing. I know how to test. My first M2P was perfect, I was lucky, my second was dreadful. Decentering is not normal, it is bad lens production or from dropping a lens. I have plenty of lenses that are not decentered, and that I buy that are get sent straight back.
I bet I can find lens issues in your M2P!