I've been looking at some sample photos from different mobile phones and I've noticed some of them have this weird feature: edges where contrasting colors meet have an additional glow. Presumably this effect is unwanted?

Normal photo at the top, abnormal at the bottom:

Concentric circles in the middle of the ISO 12233 test chart

Similar effect with colors. On the bottom photo there are additional highlights around each grid cell. I can kind of see them on the top photo too, but they are barely noticeable:

Colour chart of the ISO 12233 test chart

What is this effect called and what causes it? I think it has something to do with post-processing?

I thought it could be a halo, but all the examples I've seen are much more pronounced.

(this is the source of the photos)

  • 4
    \$\begingroup\$ The 2 images you shared are PNG files but it looks like they were first saved with lossy compression, most likely JPG. Does this answer your question? photo.stackexchange.com/questions/19270/… \$\endgroup\$
    – MrUpsidown
    Sep 26, 2022 at 8:15
  • \$\begingroup\$ @MrUpsidown You are right, these images were originally saved as JPEGs because they are intended to represent "normal" images taken by these two smartphones in controlled conditions. But top halves don't have these anomalies despite also being JPEGs, and probably saved with the same or higher JPEG compression (because they come from an older phone). So I don't think it's a JPEG thing \$\endgroup\$
    – gronostaj
    Sep 26, 2022 at 8:48
  • \$\begingroup\$ Regarding the "what is it called"-part of the question: Can you elaborate why you think "halo" does not fit? Because that's exactly what I would call it. \$\endgroup\$
    – luator
    Sep 26, 2022 at 16:07
  • \$\begingroup\$ @luator I guess you could classify it as halo, but the examples of halo I saw were cases where HDR created transitions over a larger area like this, so I thought there's maybe a more precise term for this particular defect \$\endgroup\$
    – gronostaj
    Sep 27, 2022 at 6:11

3 Answers 3


Looks like a processing effect to me. Part of a sharpening or local contrast effect - particularly unsharp masking.

Mobile telephones - and to a lesser extent in-camera processed jpegs - are devils for throwing processing that's deemed to "look better most of the time" at recorded images. This is one of the reasons many people give for using RAW - most effects will be consciously applied by the photographer.

  • 3
    \$\begingroup\$ A DSLR can even do similar when saving to RAW, if you don't switch it off - see photo.stackexchange.com/q/93500/57929 \$\endgroup\$
    – Tetsujin
    Sep 26, 2022 at 9:19
  • \$\begingroup\$ @Tetsujin - fair point, though I don't think we're seeing ADL here so I'll just tweak it to be a bit less specific. \$\endgroup\$ Sep 26, 2022 at 9:26
  • \$\begingroup\$ Agreed, it looks like sharpening artifacts to me... \$\endgroup\$ Sep 26, 2022 at 13:23
  • 1
    \$\begingroup\$ This is the correct answer. It's a garbage "enhancement" effect most camera software does to make the images look "sharper", that may look good to lay users unfamiliar with it, but which looks glaringly hideous and wrong to anyone who recognizes it. \$\endgroup\$ Sep 28, 2022 at 0:47

Coming from an image compression background, I would refer to this as one kind of "ringing artefact".

The culprit is generally that the image has been processed in the frequency domain. Discontinuities, like sharp edges, can't be represented perfectly in a finite frequency transform. This results in spurious peaks and throughs appearing near the edge, ringing.

Many different things can lead to ringing. Some lossy image compression formats, like JPG, store the image in discrete cosine transform form. This is a frequency domain transform, which inherently has ringing. So would any bandpass filter, or an unsharp mask, as mentioned in another answer.

  • 6
    \$\begingroup\$ The signal is discrete in the space domain to begin with, so “discontinuities” can be perfectly represented in the finite frequency domain: the discrete Fourier transform is perfectly reversible. In this case, the culprit is clearly a sharpening filter. Whether it is applied in the frequency domain or in the spatial domain (through convolution) is an implementation detail that has no effect on the outcome. \$\endgroup\$ Sep 27, 2022 at 9:58
  • \$\begingroup\$ @EdgarBonet Do you have a Layman's version of that comment? \$\endgroup\$ Sep 27, 2022 at 11:35
  • \$\begingroup\$ @spikey_richie: No, I don't. The argument put forward by this answer is both very technical (it is the so called Gibbs phenomenon) and completely bogus. I don't see how I could debunk it without going into the technical details. The takeaway is that the artifacts seen by the OP have nothing to do with the frequency domain of the signal being finite. \$\endgroup\$ Sep 27, 2022 at 12:05
  • \$\begingroup\$ "spurious peaks and [troughs]" ? Or is a throughs a thing? Which wouldn't surprise me; none of this makes any sense, +1. \$\endgroup\$
    – Mazura
    Sep 28, 2022 at 1:09

This sort of effect can also come from how digital camera sensors work.

For most sensors used in phones and digital cameras, each pixel can only sense brightness. These individual pixel sensors have a filter in front of them so that each pixel sensor is actually only measuring the brightness of a single color (usually red, green or blue).

The processor takes the raw data from the pixel sensors and builds the individual pixels of the final image. On sharp edges, mistakes can be made, especially with the very small sensors used in phones, etc.

You can find more about the filters used here: https://en.wikipedia.org/wiki/Bayer_filter

  • 1
    \$\begingroup\$ Interesting! Though this time it's probably caused by processing, because I've found two phones with the same Bayer sensor where only one produces these artifacts. \$\endgroup\$
    – gronostaj
    Sep 27, 2022 at 6:20
  • \$\begingroup\$ The "RGB" color filters in front of sensors are nowhere near strong enough to only allow a single color of light through. Some blue and red light gets through the "green" filter. A LOT of green light gets through the "red" filter (that's actually somewhere between yellow and orange), and a lot of green and even some red gets through the "blue" filter (that's actually a blue-violet color). That's why all three RGB values have to be interpolated for every single pixel from the monochromatic brightness values captured by the sensor. \$\endgroup\$
    – Michael C
    Sep 27, 2022 at 12:33
  • \$\begingroup\$ Without this overlap, which mimics the overlapping sensitivity of three types of cones in the human retina, there would be no way to create color images that can stimulate our retinal cones to produce the same color responses in the human brain. There are no implicit colors in various wavelengths of electromagnetic radiation, there are only colors in the perception of portions of the EMR spectrum. We call it visible light because it's the portion that our retinas chemically react to, not because of the fundamental nature of those wavelengths. \$\endgroup\$
    – Michael C
    Sep 27, 2022 at 12:43

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.