Local retailer. They were not happy because of the extra paperwork and that they basicaly got an open box unit in their inventory and either try to sell it for less as such or return it to DJI. At that time they did not have stock to simply exchange it. I had to prove or demonstrate to them what was wrong by showing them prints highlighting the blurry areas in the frame etc. It was a nightmare and it took weeks.. The one I have now is the forth unit. The first 2 were Cine combos this one is normal M3.
Mavic 3 Pro lens test
- Thread starter Meta4
- Start date
Meta 4-These are excellent results.I bought aM2P 4+ years ago,was happy with it and always longed for a larger sensor.With the examples you've shown I'm now in full blown G.A.S.As always thanks for your continuing problem solving(though this may create a small problem at home).Mostly,another drone?Thought you liked yours
To induce long term suffering from selfinflicted anxiety syndrom?
Last edited:
No… I already have that! Lol.To induce long term suffering from selfinflicted anxiety syndrom?
I remember the good old days when lenses for cameras were tested scientifically, not just subjectively. I'm not sure how doable that would be with a camera from a drone. I am sure there must be methods, but they might not be the same.
I can't see why such test could technically not be done with a drone.
The problem is that variation in IQ between units is not negligible. If such test and resulting MTF charts would be released by DJI that would be a benchmark by which every new owner would be measuring his/her new drone camera performance. Can you imagine the headache this would cause to DJI?
Lens tests on camera lenses have been done for years by independent labs and published in photo magazines, maybe on web sites. I've never seen a mfr. release such data. Therefore they aren't beholden to it.
Such tests can be both revealing, useful and go a long way toward removing subjectivity (personal opinion) from the evaluation process. They can provide useful comparisons between comparable lenses about their characteristics such as resolving power (sharpness), distortion, contrast, etc.
Such tests can be both revealing, useful and go a long way toward removing subjectivity (personal opinion) from the evaluation process. They can provide useful comparisons between comparable lenses about their characteristics such as resolving power (sharpness), distortion, contrast, etc.
Such test is only useful when lenses have consistent optical qualities. From my experience DJI cameras and lenses do not. Therefore such test would be not worth much to consumers.
Understood. I had not considered there might be such wide variability in lenses.
What exactly in your experience leads you to believe that such variability exists? Have you had the opportunity to bench test a number of the same model of lenses, or is it something else that has led you to your conclusion?
What exactly in your experience leads you to believe that such variability exists? Have you had the opportunity to bench test a number of the same model of lenses, or is it something else that has led you to your conclusion?
A number of us on this forum and the DJI one had to go through not 1 or 2 but up to 4 Mavic 3 drones to get a good main camera that was not plagued by badly decentered lenses or had odd asymmetric zones of smearing that made it impossible to either crop out or adjust in post. I had to go through 3x $5K Mavic 3 Cine‘s to get the one I have now.
It’s hard to say how many were affected but the fact that just a handful of us had to return multiple drones to get decent image quality speaks to far too much sample variation to benchmark the lenses as designed. DJI would never admit that so many were sub par only offering up the very best for an official test.
And after 35+ years of using optics professionally as a photographer, I know that the one I have in hand as it performs in my application is the only real test that produces the clear distinction of it is working for me or not. I doubt that ever changes.
Skyscope
Well-Known Member
- Joined
- Oct 16, 2016
- Messages
- 89
- Reactions
- 123
That's my experience too, and the reason, why the first thing I do with every new drone is my own check of the camera(s), if their images are evenly focussed.
With a drone it's quite simple. At a spot with a large evenly textured area like a field I get the drone in the air and shoot some 90° top down images. Flaws like distortion or a decentered lens or very mushy edges will get visible by processing these images in an app that doesn't process opcodes like Lightroom oder Photoshop does, for example in Affinity Photo, Filter "Detect Edges", or in RAWtherapee, which has a dedicated focus mask function.
Here an example of the 24mm camera of my Mavic 3 Pro:
It has a very slight falloff to the right, that's because of a slight shift in exposure due to sunset at shooting time, not ideal for the test shots. So in general it's OK.
And of course, the sharpness falls off a bit towards the edges, but evenly, and that is completely normal with these wide angle camera modules, as is the donut pattern, see Mavic 3 color shifting in white balance / color tint
You don't see this with the 70mm oder 166mm images.
Last edited:
KS-6 and Skyscope summed it up well. FWIW I returned 2 Cine and 1 non cine before getting a decent one and have seen samples from other owners which were as bad as from the three crappy M3 I initially got.
I only add to this lens testing discussion the aspect called Manufacturer lens profile. Software such as LR won't let you see the real projection of the lens. It automatically applies the software correction created by DJI and baked into the DNG and JPEG processing pipeline. CaptureOne (and some other Apps) on the other hand allows this profile to be disabled and I can tell you that without the "software correction" the 24mm lens on the main camera is very ordinary. It has substantial barrel distortion, only central area is really sharp, peripheral areas get soft pretty quickly and corners are just terrible.
This lens without software correction would be basically unusable. So my question to the OP is, how would you like the test to be performed? With or without the software correction? And what such test would be telling you? How bad the lens optically is or how good the software correction is?
I only add to this lens testing discussion the aspect called Manufacturer lens profile. Software such as LR won't let you see the real projection of the lens. It automatically applies the software correction created by DJI and baked into the DNG and JPEG processing pipeline. CaptureOne (and some other Apps) on the other hand allows this profile to be disabled and I can tell you that without the "software correction" the 24mm lens on the main camera is very ordinary. It has substantial barrel distortion, only central area is really sharp, peripheral areas get soft pretty quickly and corners are just terrible.
This lens without software correction would be basically unusable. So my question to the OP is, how would you like the test to be performed? With or without the software correction? And what such test would be telling you? How bad the lens optically is or how good the software correction is?
Last edited:
The experience you and others had sounds both remarkable and shocking. In my 60 years as a photographer, many of them as a pro, I have never heard of such an abysmal quality control issue. Sounds like something was remarkably wrong. Is it possible that whatever it was has since been corrected?
I would agree, that in the end it is the subjective discretion of the user that determines the suitability of a lens, camera or system for the task to which it is applied, but objective data never hurts in forming one's opinion.
With erratic quality control such as that you describe at DJI, bench testing a lens would seem as you suggest a rather useless exercise, but it's difficult for me to imagine that except for the cheapest of drones such problems would be widespread for very long before being reigned in. While I admit I haven't paid much attention of the goings on within the world of drones lately, I for one have not heard of persistent manufacturing issues in regard to DJI optics. Have I missed something or do you think the problems you described were perhaps more of an isolated and temporary nature?
No one will ever know if those issues are systemic or isolated or if they are things of the past affecting only M3 production or if they are still present in M3P and other drones production. DJI will never admit to such things. But the idea of putting such a small lens in front of m43 size sensor was daring and in order this to work the manufacturing tolerances when putting these cameras together would have to be spot on. The fact that many of those drone suffered from bad IQ proves otherwise. But hey this is not Leica and with the price tag to match
These drone are still amazing pieces of engineering. Let's not fotget that!!
Last edited:
I honestly think these were exceptional circumstances exacerbated by the onset of COVID and its effect on manpower.
My new Mavic 3 pro is far better in terms of optical performance across the board so I suspect Q/C and manufacturing practices have improved a fair bit since 2020.
I wouldn't expect an admission from DJI necessarily, but rather that issues such as this would come out in the drone community, here in this discussion group and other drone forums as well as YouTube. While I certainly have not done much of anything anything along the lines of research, I did do a cursory search on YouTube, and the only thing that I turned up in any number had to do with a foggy lens issue on some DJI Minis. Do you have the sense a great many drones/people were effected or do you think the number perhaps relatively modest? It didn't sound by what you wrote previously you thought the issue widespread but perhaps I misconstrued your remarks.No one will ever know if those issues are systemic or isolated or if they are things of the past affecting only M3 production or if they are still present in M3P and other drones production. DJI will never admit to such things. But the idea of putting such a small lens in front of m43 size sensor was daring and in order this to work the manufacturing tolerances when putting these cameras together would have to be spot on. The fact that many of those drone suffered from bad IQ proves otherwise. But hey this is not Leica and with the price tag to match
I do not have enough evidence to be sure one way or another. There could be two factors at play here, I suppose. One, that those receiving the duds were very unlucky few and two, that the majority of users are not professional photographers and are not subjecting their photos to such scrutiny as me and a few others on this forum. No disrespect intended to anyone.
Rob Dog
Well-Known Member
Similar threads
DJI Drone Deals
New Threads
-
-
-
-
Air 3s DJI Air 3S vs DJI Neo vs DJI Mavic 3 Pro Video Quality Test- Started by IlluminationsfromtheAttic
- Replies: 3
-
Forum statistics
