When can a photo be trusted?

Fourandsix on TwitterFollow @Fourand6
« Photographic Memories | Main | Photo Forensics from Flash Photography (part 3/3) »

The Case of the Pole in the Middle of the Road 

I recently stumbled upon a bizarre and seemingly improbable image showing a utility pole in the middle of Highway 251 east of Montreal Canada. I don’t mean a downed utility pole; I mean, an upright and functioning utility pole placed directly in the middle of a highway. At first glance, it seemed likely that this was a photo hoax, particularly since the pole’s shadow appears to either be missing or impossibly small. 

Shown below is the image in question and a magnified view of the base of the pole. From the shading on the pole (brighter on the left and darker on the right) it is clear that the sun is to the pole’s left. This suggests that the pole’s shadow should be cast to the right. The shadow on the roadway, however, seems oddly small given the size of the pole. 

[Photo credit: Jocelyn Riendeau / Canadian Press / Sherbrooke La Tribune]

At first glance I thought that it was possible that the pole was digitally inserted, that the forger forgot to add its shadow, and that the thin black shadow on the road was cast from an overhead cable. Given the simplicity of this scene — a single distant light source, a cylindrical pole, and a planar roadway — it is fairly straightforward to determine if this scene is physically plausible. That is, is there a reasonable 3-D configuration of these scene components that gives rise to the shading and shadows in this image?

Shown below is a 3-D rendering of the basic components of the above scene. In order to generate this image, I created a virtual camera with a 50mm focal length. The camera was placed 60 feet away from the pole and 6 feet above the ground plane. I then adjusted the direction of the light until the orientation of the shadow matched. In order to match the width of the shadow, I rotated the ground plane 4 degrees away from the camera. In the second image below I’ve colored the pole and shadow red and superimposed it atop the original image. In this image you can see that the shadow is a perfect match.

This 3-D reconstruction is not unique — there are any number of 3-D scene and camera configurations that would have led to the same scene. The reason is, in part, that there is an inherent ambiguity between the distance to objects in the scene and the camera focal length. The point, however, is that a reasonable scene and camera configuration can give rise to the shading and shadows in this image. So while the configuration that I chose may not exactly match the real world scene, it does prove that the scene is physically possible.

It is clear from this 3-D reconstruction that the highly compressed shadow is simply a result of perspective foreshortening. That is, the road is slanted slightly away from the camera and this causes a compression in the cast shadow. We saw this same effect in The Case of the Missing RPG Shadow.

In general, I have found that this sort of 3-D rendering can be extremely helpful in the analysis of images (as it was, for example, in my analysis of the Lee Harvey Oswald photo). This is particularly true when making judgments of 3-D geometry that can otherwise be difficult to make from only a single image.

By the way, here is an article that describes the story behind this pole, and here is an even more bizarre story about a house in the middle of a highway.

PrintView Printer Friendly Version

Reader Comments (1)

The only interesting thing I see are the timestamps in the metadata... The EXIF says it was created on 2012-11-18 14:46:49 and the IPTC also says Date Created: 2012-11-18. However, the IPTC also gives a description that says it was taken on 2012-11-19 and other metadata says it was processed on a Mac on 2012-11-19.


There is another timestamp that says it was processed on 2012-11-19T19:22:23Z. This is also a problem, since the news article is timestamped "Nov 19, 2012 3:46 PM ET". The timezone different is -0500. This is a problem since the news report was allegedly published with the picture ONE HOUR BEFORE the picture was edited on the Mac.

The image artifacts are consistent with Adobe Photoshop CS2 -- identified in the metadata. This doesn't mean "modified", it just means someone saved the image using CS2, and probably used the same program to add the IPTC metadata.

This really just means someone likely listed the "processed" date as the photo's date, the timestamps are inaccurate/wrong, and things may have been updated. However, any inconsistency in a media photo means we cannot fully trust the media photo. And since the photo's timestamp is inconsistent with the article, we cannot trust the information in the article. And if we cannot trust the article, then we certainly cannot assume that the picture is real.

[As with most media photos, this photo was edited and, at a minimum, cropped from its original. With respect to the time stamps, you say that the article was initially published with the photo. I'm not sure how you know this. Notice that the article was posted on Nov 19, 2012 3:46 PM ET and updated the next day on Nov 20, 2012 10:12 AM ET, after the processed date in the image metadata. This may or may not explain the discrepancy in dates, but I'd say that one needs much stronger evidence before concluding that the photo is not real. -Hany]

December 15, 2012 | Unregistered CommenterDr. Neal Krawetz

PostPost a New Comment

Enter your information below to add a new comment.
Author Email (optional):
Author URL (optional):
Some HTML allowed: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <code> <em> <i> <strike> <strong>