Why do two JPEGs from the same photo have different EXIF data and dimensions?
Asked 9/20/2015
3 views
2 answers
0
I found two folders of photos that I expected to contain identical files, but some JPEGs differ when compared byte-for-byte. ExifTool shows different filesystem timestamps, one file has an Orientation tag of "Rotate 90 CW," and the other file reports swapped image dimensions (4608×3456 vs 3456×4608) along with some EXIF fields moved or rewritten. What likely caused these differences, and which version is more likely to be the original?
Originally by Photography Stack Exchange contributor. Source · Licensed CC BY-SA 4.0
Photography Stack Exchange contributor
10y ago
2 Answers
3
The image has been rotated. Image in directory 1a is 4608 px wide, 3456 px high and has an Orientation tag of Rotate 90 CW. So it is a portrait picture that is in landscape orientation. If you look at it with a image viewer that has been told to ignore orientation, it will look like it is turned on it's side. The image in 1b is 3456 px wide and 4608 px high, so it has been rotated.
I would assume that the images in dir 1a would be the originals, since when a camera saves the image, it is in landscape mode and it sets the Orientation tag based on the orientation of the camera, but I would double check to be sure.
It looks like the rest of the data has just been reorganized. Items like Thumbnail Offset and Preview Image Start have been properly changed to reflect the reorganization of the exif data. Thumbnail Length has changed, which seems to me that the Thumbnail has also been rotated or possibly rewritten, though I could be wrong on that. The rest of the Exif data has changed position within the Exif as well, but still fine.
Both ACDSee and Windows Photo Viewer appear to be able to losslessly rotate jpg images, though ACDSee appears to need have a 'force lossless rotation' option set.
Originally by user37960. Source · Licensed CC BY-SA 4.0
user37960
10y ago
0
Generated from our catalog & community — verify before relying on it.
The most likely cause is that one copy of the JPEG was physically rotated and resaved.
In the version from folder 1a, the image data is 4608×3456 and the EXIF Orientation tag says "Rotate 90 CW." That means the camera likely stored the JPEG in its normal landscape pixel layout and used the Orientation tag to tell software to display it as portrait.
In the version from folder 1b, the dimensions are swapped to 3456×4608, which strongly suggests the pixels themselves were rotated. When that happens, EXIF/JFIF metadata may be rewritten or reordered, and offsets for embedded thumbnail/preview data will also change. That explains why the files are not binary-identical even if the visible image looks the same.
The differing Windows date fields are filesystem timestamps, not proof that the photo content changed in-camera.
Based on the metadata shown, folder 1a is more likely to contain the original camera-generated file, and folder 1b is a rotated/resaved copy. To confirm, open both in software that can ignore EXIF orientation: the original should appear sideways unless the viewer honors the Orientation tag.
Recommended products
UniqueBot
AI10y ago
Your Answer
Related Questions
Why do XnView and ExifTool report different EXIF orientation descriptions?
Why do portrait photos look squashed on my digital photo frame after rotation?
Why are some Lightroom exports rotated incorrectly after adding a watermark?
Why do my Canon 5D Mark II RAW files appear to have different pixel dimensions in portrait vs. landscape?
Why do NEF files differ between two SD cards in Nikon backup mode?