During my first flight of my Mavic Pro I noticed a radial discoloration in both stills images and video.  The footage was in snow covered forest.  It appears that this is a systemic problem with the camera.  DJI references the angle of incident, which may or may not be the case.  I would expect the sensor to be on par with an iphone at the very least.  I decided to do my own very non-scientific experiment to determine if I could reproduce at room temperature under incandenscent lighting.  The video below is my attempt.  I must admit that I cannot make a solid determination here.  If anything I can say that my field results are very inconsitent compared to my studio test.  Looking at the still imagery it is clear that the DJI raw files are being rendered as expected under any color profile they should be the same, and in this case they are.  It is also clear that their jpg engine is cooking the images as expected and based on the jpg images they have clearly identified the problem and are correcting with jpg.  That is great, unless you happen to have purchased the system to be able to utilize a raw image.  Note the D810 raw image is much more true to life, understanding that the D810 is a superior camera and a larger sensor, it is also a Sony sensor.   

At the end of the day, I have not clue what is going on.  I suspect colder temps have some effect, and it is possible that the angle of incidence is the cause.  I do submit that the excuse that it is systemic with all small sensors is preposterous, due to evidence from gopro footage that does not exhibit this hostility with the images.  

Forgive the audio, I am not set up for this type of work.



Note pronounced circle at :34

http://forum.dji.com/forum.php?m ... &fromuid=429792

thehippoz Posted at 2017-2-11 13:12
http://forum.dji.com/forum.php?mod=redirect&goto=findpost&ptid=81762&pid=687817&fromuid=429792

I cannot accept the response that DJI is putting out..  They have failed to release a "ready" product.  We should expect an evenly lit  image in any condition.  Perhaps they should have put out a 1080 camera on this unit with solid tech rather tahan attempt to put a 4K sensor in this "consumer, toy" camera, and then oh by the way call it "pro"..   They can do better.  I just hope they correct the issue in the next firmware.  

Turkk71 Posted at 2017-2-11 14:54
I cannot accept the response that DJI is putting out..  They have failed to release a "ready" product.  We should expect an evenly lit  image in any condition.  Perhaps they should have put out a 1080 camera on this unit with solid tech rather tahan attempt to put a 4K sensor in this "consumer, toy" camera, and then oh by the way call it "pro"..   They can do better.  I just hope they correct the issue in the next firmware.

Yeah it is what it is. At least got a answer from the camera guy! I guess all short lens cameras suffer from it and it's corrected in software. Agree though, it would have been nice to have hired programmers to finish the dng correction.

Excellent job on the video.  I too come from photography so I am doing thorough testing and its very disappointing.  On top of this, I am seeing a big problem with my white balance and blues.  Not sure if this is related to the centre yellow/pink issue which then causes the camera to have messed up white balance, but for shots taken during dusk, the blues are very heavily saturated, and when using DNG files, they are turning out much different than JPEG files.  The Mavic is clearly trying to compensate with JPEG processing, but the JPEGs are unusable from a resolution point of view, and the DNG files are unusable because of the red/pink centre, and for me, white balances issues.

The sad thing is that since its a design flaw of the sensor/lens, I'm not sure how much a firmware upgrade will help.  A higher quality sensor/lens unit seems like the only real solution.  

thehippoz Posted at 2017-2-11 17:29
Yeah it is what it is. At least got a answer from the camera guy! I guess all short lens cameras suffer from it and it's corrected in software. Agree though, it would have been nice to have hired programmers to finish the dng correction.

You are correct thehippoz at least we know they are aware of it.  I truly blame myself here because I was so amazed by the size handling of the drone, I didn't research the optics like I would have done on any other camera.  At the end of the day it is one amazing piece of tech on it's own.  Credit where credit is due.  

Stewie Griffin Posted at 2017-2-11 20:42
Excellent job on the video.  I too come from photography so I am doing thorough testing and its very disappointing.  On top of this, I am seeing a big problem with my white balance and blues.  Not sure if this is related to the centre yellow/pink issue which then causes the camera to have messed up white balance, but for shots taken during dusk, the blues are very heavily saturated, and when using DNG files, they are turning out much different than JPEG files.  The Mavic is clearly trying to compensate with JPEG processing, but the JPEGs are unusable from a resolution point of view, and the DNG files are unusable because of the red/pink centre, and for me, white balances issues.

The sad thing is that since its a design flaw of the sensor/lens, I'm not sure how much a firmware upgrade will help.  A higher quality sensor/lens unit seems like the only real solution.

Thank you, I don't normally do videos but enjoyed working on that one.  Like I mentioned to another I am now realizing that because of the wiz bang factor of the drone itself, and oh by the way I can shove it in my daypack, I didn't research the camera like I would have done with a stand alone camera.  Cool little drone no doubt.  

"Note pronounced circle at :34"
Looks like it is caused by the reflection of the sun by the snow crystals on the ground at specific angles which changes both the brightness and the colour depending on the angle.   Very often get the same thing with grass when the sun is directly behind the aircraft, but in that case with a much smaller circle with the shadow of the aircraft in the centre.  The answer is not to film with the sun directly behind the lens - move sideways or up/down to get a different angle and the effect disappears.

You also have an issue on the video following that with the snow being over exposed a lot of the time causing it to go pure white which also removes the issue except around the trees where it is darker.

Nigel_ Posted at 2017-2-12 08:11
"Note pronounced circle at :34"
Looks like it is caused by the reflection of the sun by the snow crystals on the ground at specific angles which changes both the brightness and the colour depending on the angle.   Very often get the same thing with grass when the sun is directly behind the aircraft, but in that case with a much smaller circle with the shadow of the aircraft in the centre.  The answer is not to film with the sun directly behind the lens - move sideways or up/down to get a different angle and the effect disappears.

I cannot agree with he sun behind the lens in this case the light was flat and no point light exists.  The angle of incidence would not cause what you are suggesting at least in this case.  Further I have other images and video that demonstrate the issue in sunlight, not in snow, and with the sun not behind the camera.    Other portions of the video were deliberately blown out for the reason you mentioned.   I do appreciate your input, however I disagree.   

Some technical details about the camera limitation: https://www.dropbox.com/s/hf8qau ... %20Shading.pdf?dl=0

Ex Machina Posted at 2017-2-12 18:11
Some technical details about the camera limitation: https://www.dropbox.com/s/hf8qauswt8fggei/Mavic%20Pro%20Color%20Shading.pdf?dl=0

THanks for the link!

Seems some professional photographers need to take more technical lessons