**
On Episode 374, Candace Owens showed her audience a surveillance still from the UVU parking garage. The episode passed a million views inside a day. She called the photos exclusive. She said they came from the 4K footage shown in the courtroom. Then she made a promise about them.
"We did not edit these in any way, okay? This is no editing. It's not compressed."
Fifty seconds earlier, she told viewers what the image proved. Without the compression and blur, she said, "you might realize that he's kind of a bit broad."
I measured the image. It is edited. And the broadness she pointed to is the edit.
Why I can speak to this
I have spent years as a documentary and street photographer. I shoot wide lenses. I correct wide lenses. I know what they do to a frame because I fight it every time I use one.
Here is the thing every photographer learns early. A super wide lens squeezes whatever stands near the edge of the frame. You have seen it yourself. Take a group photo on a GoPro or a doorbell camera and the people at the ends come out skinny. The lens did that. UVU runs super wide security cameras in that garage. You can see it in the raw footage. The ceiling line bends. The far wall curves. That is a wide lens doing what wide lenses do.
There is a right way to fix that distortion. It is called lens correction. Software that knows the lens bends the image back, and it bends it by different amounts in different parts of the frame. That is the whole point. The distortion is uneven, so the fix must be uneven.
There is also a wrong way. You grab the side handle in an editor and you drag until the picture looks right to you. Every beginner does it once. The result can look plausible. It is still a guess, and the guess decides what the subject's body looks like.
You can tell the two apart with measurement. Lens correction stretches different parts of the frame by different amounts. A dragged handle stretches everything by the same amount. That difference is the fingerprint.
The test
Her image and the widely circulated wide-angle clip come from the same camera. Same white Camry in the corner. Same Jeep Wrangler. Same black SUV. Same painted marking on the floor. Her image shows a tight section of that view.
So I ran the obvious test. I took the wide view and tried to lay it over her image.
First I enlarged it proportionally, the way any honest zoom works. It failed. Nothing lined up. The ghost of the same man appears next to himself, and the ghost is visibly narrower. That ghost is what the camera actually recorded.
Then I stretched the wide view sideways. Not a little. The math resolves to about 1.9 times more stretch sideways than vertically. With that stretch applied, her image locks onto the wide view to within about one pixel. Every car. Every floor line. Every shadow.
Her image is the wide-angle view, cropped and stretched sideways by a factor of roughly 1.9.
Then comes the fingerprint. I measured the stretch separately in the top, bottom, left, and right of the frame. It is the same amount everywhere. Between 1.88 and 1.92 in every region. Real correction for a lens this wide would vary visibly across a span this large. This does not vary at all. That is not a lens profile. That is a dragged handle.
Someone opened this image in an editor and pulled it wider until it looked right to them. Then it aired under the words "no editing."
The broadness is the edit
Now go back to what she asked her audience to see. "He is kind of a bit broad."
In her version, his shoulders measure about 28 percent of his height. That is dead center for an adult male. Normal range runs about 26 to 29 percent. He looks like a regular man because someone chose the stretch that made him look like a regular man. Pick a bigger number and he becomes a linebacker. Pick a smaller one and he thins out. His build in that image is a setting.
The raw wide view cannot rescue the claim either. In the unstretched frame his shoulders measure about 17 percent of his height. No human being is built like that. The lens squeezed him.
So neither image can tell you this man's build. One distorts him by physics. The other distorts him by hand. Anyone measuring a body off either frame is measuring the distortion, not the man.
I stretch photographs for a living. When I do it, I do not call the result unedited.
What I am not claiming
I follow the evidence and I state its limits. Here are mine.
I cannot tell you who applied the stretch. The pixels do not carry fingerprints. Her team may have done it. The person who provided the still may have handed it over already stretched. A statement of no editing fails either way, because unedited is a property of the image, not of her workflow. She vouched for the geometry of an image whose geometry had been changed, and she built a claim about a man's body on top of it.
Some will say a video export or an odd pixel format did this automatically, with no hands involved. The factor matches no standard format I have found. And it would not matter if one existed. However the stretch got there, the geometry on that screen is not the camera's geometry, and a man's build cannot be measured from it.
I cannot tell you the content was faked, and I have no evidence it was. The opposite, actually. I tested her image for real detail the circulating clip does not contain. It has it. The Jeep's grille bars resolve clearly in her version and dissolve into blur in the clip, even when the clip is enlarged by the exact same factors. Her still comes from a genuinely higher quality copy of the same camera's footage. That is consistent with her account that the source was the courtroom footage. The source is real. The geometry is not.
I also tested for signs of AI editing. The man's noise pattern matches the cars around him. His compression residuals match. His outline shows the same softness as the Jeep's hood line and the Camry's roof. And he appears in two independent sources, her still and the circulating clip, in the same clothes on the same path, and the two sources align to within a pixel. Faking that would mean faking both. I found nothing. I will also tell you the limit of that finding. Through this many generations of compression, those tests lose their teeth. So I state it as what it is. I found no evidence of AI, not proof there was none. And note what the question costs anyone who raises it. If AI enhanced this image anywhere along the way, that is one more edit under the words no editing.
Now the part her defenders will want to quote, so let me say it carefully. Yes, the raw wide view distorts him. It renders shoulders no man has. And yes, the sideways stretch pushed his proportions into a believable range. Her defenders will call that a correction. Notice what that defense costs them. Correction means editing. You cannot argue the stretch was a legitimate fix and also that there was no editing. Pick one.
And a believable range is not a measurement. The stretch factor was not calculated from the lens. It was tuned by eye until the man looked normal. So the fact that he looks normal proves nothing about him. It proves the editor has working eyes. Turn the dial further and he is a linebacker. Turn it back and he is a rail. An image whose proportions were set by hand cannot testify about proportions. That holds even if the hand guessed well.
I will also leave the socks alone. She spent a segment on his white lower legs and what they might hide. At this resolution the bright band could be legwear, blown highlights, sun on skin, or video compression smearing the tones together. The stills cannot settle it. So I will not pretend they can. That is the difference between measuring an image and narrating one.
Why this matters
This is not about one photograph. It is about a method.
The method states things as certain. No editing. Not compressed. Then it draws conclusions from the certainty. He is broad. He is left-handed. There is something under the hat. When the certainty collapses, the method does not correct. It moves to the next certainty.
In this same episode she suggested footage was intentionally degraded and blurred to hide a detail. Two minutes earlier she told you her clean stills came from 4K footage shown in a courtroom. Sit with that. The suppression she describes ran through open court and ended up on her podcast. If someone is hiding this footage, they are remarkably bad at it.
I hold myself to a different standard, and I invite you to hold me to it. Everything above is checkable. The images are public. The overlay test runs in free software. The measurements are stated with their limits. If I got a number wrong, show me and I will correct it in this article with a note saying I did.
She showed an image to her audience and told them what it proved. The image cannot prove what she said it proves. That claim I can defend all day. I just did.
