Mavic 3 color shifting in white balance / color tint

[JustName]

Just take a look at this posting.

Mavic 3 color shifting in white balance / color tint

These kind of defects always show up in low contrast situations, that's logical. Caused by the fact that the contrast in these scenarios is low, and this low contrast is 'boosted' to get a pleasant looking image. When contrast gets boosted, color shifts get boosted to (become more contrasty =...

mavicpilots.com

 

[flyliedutch]

I haven't noticed this issue with my M3.

Is there a specific test I can do to test this out? What I mean is, taking a photo of a white wall and then applying filter/effect XYZ on the .dng file...

Something of that nature spelled out step by step so we can all test our M3's out and perhaps we can come to a better understanding just how widespread this issue is. Maybe reference it with production date and general location of purchase.

Thoughts?

Make some pictures on a very cloudy day in different directions with at least 1/2 of the image filled with the sky. Upload the DNG files and let’s see.

When you set the value of vibrance to +100 in Lightroom or Photoshop you will see the magenta colors on the left and right side.

But it depends on the lighting conditions. Some pictures on cloudy days are perfect on my unit. Another cloudy day it’s awful. Even testing inside don’t give the exact same results.

Personally I hope DJI will bring a software fix through a firmware update. But let’s see…

 

[ianwood]

Some good posts here. My thoughts for the cheap seats:

An M43 sensor with a compact permanent lens using plastic elements is going to have a decent amount of color shifts, vignetting and CA. That's just physics. After all, the primary reason you have a large lens on a camera is to minimize all the optical issues of a smaller lens. And the larger the sensor, the larger the challenge.

It could be that there were issues / needed optimization in the opcodes for DNG photos which the new firmware may have partially fixed.

This doesn't preclude issues with manufacturing tolerances and maybe that is something to evaluated further. I would guess those tolerances decrease with sensor size meaning they need to be that much more exact compared to say an M2P.

 

[Paradigm]

Make some pictures on a very cloudy day in different directions with at least 1/2 of the image filled with the sky. Upload the DNG files and let’s see.

When you set the value of vibrance to +100 in Lightroom or Photoshop you will see the magenta colors on the left and right side.

But it depends on the lighting conditions. Some pictures on cloudy days are perfect on my unit. Another cloudy day it’s awful. Even testing inside don’t give the exact same results.

Personally I hope DJI will bring a software fix through a firmware update. But let’s see…

I just tried with some photos I have and I don't see it. The lack of consistency is both worrying and alleviating, since it leads me to believe that it MAY be a software issue.

Some good posts here. My thoughts for the cheap seats:

An M43 sensor with a compact permanent lens using plastic elements is going to have a decent amount of color shifts, vignetting and CA. That's just physics. After all, the primary reason you have a large lens on a camera is to minimize all the optical issues of a smaller lens. And the larger the sensor, the larger the challenge.

It could be that there were issues / needed optimization in the opcodes for DNG photos which the new firmware may have partially fixed.

This doesn't preclude issues with manufacturing tolerances and maybe that is something to evaluated further. I would guess those tolerances decrease with sensor size meaning they need to be that much more exact compared to say an M2P.

I was gonna ask about the M2P, but you answered it with your last question.

I'd love to see more people these this out with units from all over the world...

 

[JustName]

I just tried with some photos I have and I don't see it. The lack of consistency is both worrying and alleviating, since it leads me to believe that it MAY be a software issue.

Could you please upload some of them? In best case DNG-Files and with bigger amount of white surfaces?

 

[Paradigm]

Could you please upload some of them? In best case DNG-Files and with bigger amount of white surfaces?

Well, I'm not going to upload until I figure out how to get rid of EXIF data such as GPS.

But, I just flew the M3 earlier and it's fantastically cloudy. I then uploaded some DNG's into Lightroom and the only setting I touched was the vibrance, which I turned all the way up to 100.

https://imgur.com/a/RYHBgod

I've been keeping an eye on this issue since before I purchased the drone and have yet to see it, even when flying in cloudy conditions more often than not.

It was all shot in auto camera mode via Raw + JPG. I also took some JPG photos with the telephoto lens, but have yet to run them in Lightroom.

My next test is to run the Therapee (sp) app mentioned earlier to see if I can find something with that.

Also, I'm curious what part of the world people that have the color tint are located in?

 

[JustName]

You could use this application

Metadata Converter

***Metadata Editor + Metadata Viewer + Renamer + Metadata Exporter*** ***GENERAL - DATE/TIME - CAMERA - GPS - IPTC - XMP*** --Remove private metadata & add own copyright-- Metadata Converter is an intuitive batch photos metadata editor, viewer and renamer. Correct EXIF/IPTC/XMP metadata tags in…

apps.apple.com

 

[mischakeijser]

Well, I'm not going to upload until I figure out how to get rid of EXIF data such as GPS.

But, I just flew the M3 earlier and it's fantastically cloudy. I then uploaded some DNG's into Lightroom and the only setting I touched was the vibrance, which I turned all the way up to 100.

https://imgur.com/a/RYHBgod

I've been keeping an eye on this issue since before I purchased the drone and have yet to see it, even when flying in cloudy conditions more often than not.

It was all shot in auto camera mode via Raw + JPG. I also took some JPG photos with the telephoto lens, but have yet to run them in Lightroom.

My next test is to run the Therapee (sp) app mentioned earlier to see if I can find something with that.

Also, I'm curious what part of the world people that have the color tint are located in?

The Netherlands

 

[mischakeijser]

Today I tested my M3 after returning from DJI (via Droneland), with the latest firmware. The effect seems somewhat reduced, but is still visible en unusable. I've added two mavic 2 samples; all 4 photos have had the same settings setting in lightroom; reduce haze +90. No problem for the M2 files.

 

[Point Zero]

Today I tested my M3 after returning from DJI (via Droneland), with the latest firmware. The effect seems somewhat reduced, but is still visible en unusable. I've added two mavic 2 samples; all 4 photos have had the same settings setting in lightroom; reduce haze +90. No problem for the M2 files.View attachment 140614

Sorry to see this!

It looks like all your photos are labelled as Mavic 2 and not Mavic 3's in the magenta images?

 

[GtoJon]

Looking at my 3rd M3, it has the Magenta issue .... very slight though. Arrrrg

Cheers, Jon

 

[mischakeijser]

Sorry to see this!

It looks like all your photos are labelled as Mavic 2 and not Mavic 3's in the magenta images?

Oh, that's weird. Something gone wrong there. The ones with the trees are the M3 files.

 

[mischakeijser]

Looking at my 3rd M3, it has the Magenta issue .... very slight though. Arrrrg

Cheers, Jon

Looking at my 3rd M3, it has the Magenta issue .... very slight though. Arrrrg

Cheers, Jon

But definitely visible, Jon.

 

[flyliedutch]

So replacing units don’t seem to be the solution. My third unit has also this issue when it’s foggy / cloudy.

 

[Paul2660]

What you are seeing is lens color cast, it's a problem with any lens that is close to a sensor. The closer your lens is to the sensor, the more prone you will see this. With Medium format Tech cameras, the problem is fixed with a LCC (lens cast correction) process. This involves taking a off white card and placing it in front of the lens, and taking a series of images at all the various apertures. The in raw software you create a LCC and use it each time. You only need to create the series of LCC's one time unless shift is involved. Software like Capture One and LR both allow for creation of the LCC and storing the file. I have used LCC's on the M2 Pro since the beginning and they will make a difference in the overall image quality. The LCC will correct both the light fall off and Magenta color cast issues. DJI will probably keep working on firmware fixes to eventually make this issue less noticeable, however the best solution for a raw with a sky or other light areas where the magenta shows up is to use the LCC. Adds one more step to the process, but gives a much better file. The firmware fix will work better on the jpgs than the raw files IMO. DJI has never been very focused on still image quality, much more on video.

The closer a lens is to the sensor, the more prone this problem is. Thus with a Medium format camera, you tend to see it only with a wide angle lens, not a telephoto, wider the lens, closer it is to the sensor.

With the M2 Pro, you can see slight magenta issues on the raws, the jpgs are corrected in the drone. Same with the original Mavic line with the 12MP camera.

If you want to learn more on the LCC process, Capture One, etc. and the effects of color shift with a sensor close to the lens, there are tons of great articles on the web.

Paul

 

[flyliedutch]

Thanks Paul. But the strange thing is that only in low contrast situations as foggy or cloudy weather, the issue is visible (at my unit).

I will look into LCC profiles but I’m hoping Adobe or DJI will fix this with a lens profile in software or a firmware update.

 

[Paul2660]

It will be most visible in those type of situations, at least from my experience, where you have a slighter sky, or white sky due to fog or clouds. It will also show up on the corners on a blue sky at times, very dependent on the light and angle of sun. On the M3 raws I pulled down from various web sites a while back the effect was clear to see, however current versions of Capture One do not support the M3 raws. I tried a few exif changes and got C1 to open them but never could get a LCC to process out. Knowing Capture one and how slow they are to add new cameras, especially Hasselblad, it may be a while. I have not tried the LR correct (I believe it's called flat field in LR),

Hopefully DJI will address this with a firmware fix in the future, or per some of the later posts, they may have already.

Paul

 

[mischakeijser]

It will be most visible in those type of situations, at least from my experience, where you have a slighter sky, or white sky due to fog or clouds. It will also show up on the corners on a blue sky at times, very dependent on the light and angle of sun. On the M3 raws I pulled down from various web sites a while back the effect was clear to see, however current versions of Capture One do not support the M3 raws. I tried a few exif changes and got C1 to open them but never could get a LCC to process out. Knowing Capture one and how slow they are to add new cameras, especially Hasselblad, it may be a while. I have not tried the LR correct (I believe it's called flat field in LR),

Hopefully DJI will address this with a firmware fix in the future, or per some of the later posts, they may have already.

Paul

Hi Paul, I'm familiar with the problem you discribe on for example cambo wide angle technical camera's. But what keeps me puzzled, is the 'randomness' of magenta and green area's. They are gradually appearing towards the edges, in a circular way, as you would see with vignetting. But just 'blobs'. Can you explain this?

 

[mischakeijser]

Hi Paul, I'm familiar with the problem you discribe on for example cambo wide angle technical camera's. But what keeps me puzzled, is the 'randomness' of magenta and green area's. They are gradually appearing towards the edges, in a circular way, as you would see with vignetting. But just 'blobs'. Can you explain this?

They are not gradually appearing

 

[Paradigm]

I'm curious why some units are affected, or more affected, while others are not. Could it be the environment they're shooting in, or is it an actual issue with the lens of the affected unit.