How do you check whether a widely shared photo is real before you pass it on?

Add one step before all of those, because it resolves a surprising number of cases without any technical work: check what the picture is being claimed to show.

Very often the image is entirely genuine and the caption is false. A real photograph of a real road somewhere, presented as being from a different country, a different decade, or a different event. Nothing in the pixels is wrong; the claim attached to them is.

That failure mode is far more common than fabrication, and it is invisible to every forensic tool, because there is nothing to detect. Asking "what exactly is this being said to prove" separates the two questions and usually tells you which one you are dealing with.

It also means a picture can pass every authenticity check and still be misinformation, which is worth holding onto.

The order matters, because the cheap steps resolve most cases and the expensive ones are rarely needed.

1. Reverse image search, on more than one engine. This resolves the majority of shared images in under a minute. You are looking for the earliest appearance and for versions without the interesting element. Different engines index differently, so a miss on one is not an answer.

2. Look for the unedited original. Composites are usually built on a stock photo, and stock libraries are indexed. Finding the same road without the tree is conclusive and takes no expertise.

3. Read the image's own internals. Sometimes there is metadata giving a camera, a date, a location. Absence proves nothing — social platforms strip it — but presence occasionally hands you the answer.

4. Check the scene against maps. If the claim names a place, satellite and street imagery either shows the feature or does not.

5. Only then look at pixels. Error level analysis and similar tools are the last resort, they require interpretation, and they produce confident-looking output that is frequently misread.

On your instinct about the lighting: trust it enough to investigate, and not enough to conclude.

Human intuition about composites is genuinely quite good on the things we evolved to notice — shadows pointing in inconsistent directions, edges that are too clean, scale relationships that do not work, reflections that do not match. Those are worth flagging.

It is also unreliable in the other direction. Real photographs frequently look wrong because of unusual lens compression, overcast light with soft shadows, heavy processing, or simply an unfamiliar scene. "This looks fake" has a high false positive rate.

So use the instinct as a trigger for step 1, never as the verdict.

Practical note for the specific genre this belongs to: images offered as proof that some infrastructure choice is possible are worth checking especially carefully, because the demand for them is high.

There are real examples of roads and buildings routed around trees, and there are also a lot of illustrations and renders circulating as photographs. The existence of genuine cases is exactly why the fabricated ones spread — the claim is plausible, so nobody feels a need to check it.