Explained: Bands of colored "particles" in sky photos [digital camera artifact]

Trailblazer

Moderator
Staff member
Last edited by a moderator:

cloudspotter

Senior Member.
It's possible, but I think it may be a weird camera artifact, perhaps overloading of the sensor in brightly sunlit regions of the contrails?

But not at the bottom of the photos where the sun is brightest. I think Ros is a professional photographer or an experienced amateur so I can't believe this is accidental
 

Trailblazer

Moderator
Staff member
Yes, her Flickr page disables downloading which also means you can't access the EXIF data from the originals.
 

solrey

Senior Member.
Could be an artifact produced by the lens coating. She said 3 of the same model camera have all captured the effect.


Modern multicoated lens. Note the typical green and magenta colors.

http://www.kenrockwell.com/tech/lenstech.htm

The only chromatic aberration seen commonly today is secondary lateral chromatic aberration. Lay people call this "purple fringing" for the green/magenta artifacts seen at the corners of images on bright, contrasty things.
Content from External Source
 

Mick West

Administrator
Staff member
It's some digital artifact. The noise just occurs at a particular brightness point in the image. You you add a gradient map you can see it follow a region of equal brightness, and the shape just matched the shape of the brightness, nothing to do with the trees.

(Move the slider in the image below)


I suspect either a faulty sensor, or some firmware bug. Possible to do with low light combine with bright sun. ISO 64 is very low ISO.
 

Mick West

Administrator
Staff member
The exif data says


"Sunset, 10, 10" suggests some in-camera processing was done.

The sensible things to do would be to take a photo of a candle in a dark room with the exact same settings. See if you get the same effects.

It's very obviously just a camera artifact though.
 

Mick West

Administrator
Staff member
On the attached original image, you can see the noise in the trees, again following the brightness contours.
20141016-100650-z5teh.jpg
 

Attachments

  • 11137982716_83c11890c9_o.jpg
    11137982716_83c11890c9_o.jpg
    181.7 KB · Views: 484
Last edited:

Mick West

Administrator
Staff member
It's very easy to duplicate the effect digitally in Photoshop:

  1. Add a solid colored layer
  2. Set transparency to 50%
  3. Edit the blend options and move the two end points of "Underlying layer" close together.


Now I don't think these image are faked, but this just demonstrates it's a purely digital thing, and not something in the air.
 

Trailblazer

Moderator
Staff member
So @Mick West , in the example above what is the blend doing? Replacing pixels with those of the underlying layer if they match a specific brightness/luminance range? My Photoshop skills aren't too advanced.

It seemed to me that the camera was somehow corrupting pixels of a particular range, which seems to be borne out by your example there.
 
Last edited:

Mick West

Administrator
Staff member
So @Mick West , in the example above what is the blend doing? Replacing pixels with those of the underlying layer if they match a specific brightness/luminance range? My Photoshop skills aren't too advanced.]

Not exactly, it's blending the red pixels with the underlying layer if the underlying layer is within a certain range of brightness. Otherwise it ignores them.

It seemed to me that the camera was somehow corrupting pixels of a particular range, which seems to be borne out by your example there.

Yes, that's exactly what it is. I'd suspect a software bug, but then I'd also expect to have seen this type of corruption elsewhere. So possibly a bad sensor.
 

Trailblazer

Moderator
Staff member
Not exactly, it's blending the red pixels with the underlying layer if the underlying layer is within a certain range of brightness. Otherwise it ignores them.
Of course, the "underlying layer" in this case is the photo. Got it.

I searched for any discussion online of this kind of corruption from this make of camera. Nothing.

The closest thing I came up with was this green fringing caused by sensor overload (and in a different make of camera).

upload_2014-10-16_19-26-14.png

http://mintaka.sdsu.edu/GF/observing/artifacts.html
 

Mick West

Administrator
Staff member
Of course, the "underlying layer" in this case is the photo. Got it.

I searched for any discussion online of this kind of corruption from this make of camera. Nothing.

The closest thing I came up with was this green fringing caused by sensor overload (and in a different make of camera).

upload_2014-10-16_19-26-14.png

http://mintaka.sdsu.edu/GF/observing/artifacts.html

A bit late, but this photo made me think that may it's an HDR (High Dynamic Range) compositing thing. HDR in-camera photos take two of three images in quick succession, and then blend them together, so you get more detail in dark and bright areas. So it would blend in the artifact on what looks like a good exposure.

The EXIF picture style says "Sunset 10,10", and HDR can work well for sunsets, so might be the setting the camera uses. Example HDR sunset
 

Trailblazer

Moderator
Staff member
I have spotted a very similar effect in Google Street View imagery of Europe. I didn't think to save it but I might be able to find it again. I was looking at views somewhere in Germany or Poland, I think, and there was a thin fringe of colour in places that looked just the same.
 

Trailblazer

Moderator
Staff member
I have spotted a very similar effect in Google Street View imagery of Europe. I didn't think to save it but I might be able to find it again. I was looking at views somewhere in Germany or Poland, I think, and there was a thin fringe of colour in places that looked just the same.

This was the effect I was referring to on Google Street View. Example from the Czech Republic here. The green banding in the sky is very similar to the effect in the contrail pics, but there are also other coloured bands.

upload_2015-1-23_14-37-3.png
 

Mick West

Administrator
Staff member
The fact that it only occurs in the front facing image suggests it's a hardware issue with that particular camera.


The street view cars have eight camera in a circle, each covering 45° of the view, and the banding here is just on the 45° that one camera would cover.
 
Top