0
\$\begingroup\$

I have some images of which some look blurry, and the others look non-blurry.

I'm trying to analyze the EXIF data of such images in order to determine why some images look blurry and why others don't.

For instance, below are the tags I get when extracting EXIF data. My question is, what of those tags would tell me more about the image's quality and why it might be blurry or non-blurry?

ImageWidth
ImageLength
BitsPerSample
DateTimeOriginal
DateTimeDigitized
ExifInteroperabilityOffset
ExifVersion
ComponentsConfiguration
LightSource
Flash
FocalLength
SubjectDistanceRange
ApertureValue
Make
Model
SubsecTimeOriginal
Orientation
YCbCrPositioning
Contrast
SensingMethod
ExposureBiasValue
XResolution
YResolution
ExposureTime
FileSource
FocalLengthIn35mmFilm
ExposureProgram
ColorSpace
UserComment
ISOSpeedRatings
ResolutionUnit
WhiteBalance
MeteringMode
FNumber
Software
DateTime
ShutterSpeedValue
Saturation
SceneType
Sharpness
DigitalZoomRatio
ExifImageWidth
CustomRendered
FlashPixVersion
SceneCaptureType
ExifImageHeight
SubsecTime
ExposureMode
ExifOffset
SubsecTimeDigitized
BrightnessValue
GainControl
MakerNote

Thanks.

Improve this question
\$\endgroup\$
3
  • 1
    \$\begingroup\$ What are you trying to accomplish by automating this? Do you want to automatically weed out images that are blurry? Is this just for your personal collection of photos, or do you hope to apply it to thousands or millions of photos perhaps as part of a service? I ask because there may be some answers that achieve some of these things better than others, or at least could guide such automation a little more intelligently. \$\endgroup\$
    – user1118321
    Jul 8, 2017 at 16:11
  • \$\begingroup\$ Would ignoring the exif and build a neural network to anyalyse the image content instead be an acceptable answer? \$\endgroup\$
    – Crazy Dino
    Jul 10, 2017 at 15:49
  • \$\begingroup\$ All things being equal, a compressed (think JPG or compressed raw) 'blurry' image will have a smaller file size than a compressed sharp image. The thing that needs to be 'most' equal is ISO- a noisy, high ISO image will have a larger file size than an identical low ISO image. Granted, not in EXIF data, but sometimes useful. \$\endgroup\$
    – BobT
    Jul 10, 2017 at 17:29

2 Answers 2

Reset to default
5
\$\begingroup\$

Maybe none.
Typically, blurry images come from wrong focus distance (bad focusing), camera shake, or motion blurring.

Bad focusing: the EXIF data will tell you what the focus distance was, but it cannot tell you if this was a good choice for the shot you wanted.
You can identify this issue by checking if some pieces of the photo are in focus (pieces you don't care about).

Camera shake: the exposure time gives a hint on that. Generally, camera shake starts when the exposure time is lower than the focus length; although if you move the camera while shooting, you can make it happen at higher speeds, and if you're good, you can hold it still at a tenth of the length.
You can identify this issue by the blur being streaked irregularly, but in the same form and direction for the whole picture.

Motion blur happens when you shoot a moving object, like a fast car.
You can identify this issue by the blur being straight streaks on all moving objects.

Of course it is possible to have multiple effects combined.

The most probable cause is the first, from your description; but without an example I can't tell.

Improve this answer
\$\endgroup\$
1
\$\begingroup\$

Image quality in general is really hard to measure since it is mostly subjective. When is a photo "good"? There are a lot of great famous pictures which are not correctly focused (in a physical way, of course). When it comes to using photography as a form of art, judging the image quality seems to be impossible without maybe some kind of AI.

That "data-blindness" doesn't change much if we only want to know more about sharpness. Like Aganju mentioned in his answer, all the EXIF parameters tell you what the camera's parameters were when you took the picture. A set of parameters can lead to a super sharp image in one situation and a blurry one in another, there is no way to tell that from the data.

Improve this answer
\$\endgroup\$

Your Answer

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

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