Nikon Puts Its Winning Cilia Video Back Under Review

Nikon is re-evaluating its 2026 Small World in Motion winner after scientists who diagnose a rare lung disease said the clip does not match real airway cells. The $3,000 (£2,250) first prize went to Dr. Ning Xu of Tsinghua University in Beijing for a brightly colored sequence of cilia, the hair-like parts that sweep mucus through a child’s airways, in primary ciliary dyskinesia, or PCD.

Xu says an unsupervised neural network only helped color real microscope measurements. Nikon first called that legal, then changed the caption and opened the file again.

A First-Place Clip of Diseased Airway Cilia

On September 15, 2026, Nikon Instruments 16th annual video competition winners and put Xu at the top of a field of more than 340 videos from 40 countries. The still-image contest behind Small World dates to 1975. The motion contest, judged on originality, information, technique, and visual impact, is in its 16th year.

Xu is listed as an optical engineer in Tsinghua’s Department of Precision Instrument. He later moved to the National University of Singapore as a research fellow. The winning clip, recorded in December 2025, uses diffractive super-resolution microscopy at 100X on a patient-derived sample. Xu said a still frame is not enough. “To understand this disease, you need to observe around 10 seconds of motion, not just a still image.”

Eric Flem, senior manager for communications and CRM at Nikon Instruments, said the clip is why the company added video 16 years ago: motion can change what a viewer understands. Nikon posted the sequence on its Small World account two days after the unveil.

🥇 Congratulations to Dr. Ning Xu for winning this year's Nikon Small World in Motion competition!

Read more about Dr. Ning Xu and his winning video in his Masters of Microscopy article: https://t.co/5fAIQZg8bY pic.twitter.com/JuU4sZs2fa

— Nikon Small World (@NikonSmallWorld) September 17, 2026

THE 2026 CASH LADDER

Honorable mentions pay $200 each. The cash is not what most of these labs are chasing. Hickey, who finished fifth with living leaf cells, said the recognition matters more than the top purse, and that it would be unfair if a winner had been built with generative AI.

Xu Ran an Unsupervised Network on Grayscale Frames

The fight is not about whether a computer touched the file. Xu has said that it did. In a Masters of Microscopy piece Nikon updated on September 22, 2026, he described a custom microscope that uses a digital micromirror and diffractive optics, a grayscale reconstruction, then a neural network used to distinguish and visualize features in grayscale data.

He and his collaborators spent years on the hardware. Work on the system began in 2021 with partners including Professor Sarah Bohndiek at the University of Cambridge and Dr. Calum Williams, now at the University of Exeter. The prize clip came from the fifth version of that rig. Xu’s line on live samples is blunt: strong light damages cells and changes how they behave, so a pretty video of dying tissue is the wrong goal.

False color is ordinary in this contest. Judges allow stains, dyes, and mapped hues on a sample that started gray. Xu’s network sits in that gap. It does not, on his account, invent the beat. “AI was not used to generate the experimental movie, the cilia, or their motion,” he wrote. He also told readers there is a difference between making something that was never in the experiment and processing data that was.

The important question is not simply if AI was used; we should ask how was it used.

Dr. Ning Xu, optical engineer, in comments after the award

On Nikon’s own winner profile he went further, advising other microscopists to “embrace AI to reveal the hidden beauty in your data.” That sentence is now the problem, because the people who look at PCD for a living say the beauty is in the wrong place.

Why the Cilia Look Wrong to PCD Labs

Robert Hirst, a lead scientist at Leicester University whose lab uses microscopy used to diagnose PCD, said the clip set off alarms as soon as the scales and details came into view. Features in the movie, he said, “just are not consistent with nature.”

Healthy airway cells carry a dense brush of cilia, not a single hair. Hirst says the prize video trains the eye on one cilium per cell, when a real cell has about 200. He also says the layer under the hairs is wrong in scale and wrong as biology, and that it “just looks completely fabricated.”

WHAT SPECIALISTS FLAG IN THE CLIP

• Cell count: Hirst says the movie reads as one cilium per cell, against about 200 on a real airway cell.
• The layer beneath: Christophe Leterrier, a neuroscientist at CNRS and Aix-Marseille University, says the band under the hairs has been read as tiny nuclei, mitochondria, and membrane, which would not sit there like that.
• Motion that starts from nowhere: Edward Phelps, an associate professor at Florida University, said structures “pop in and out of existence” and that cilia “appear from nowhere.”
• Size: Other scientists said the hairs look too big for the cells they sit on.
• A suspected watermark: Ian Donovan, a PhD student at UT Southwestern Medical Centre, said a source image appeared to carry a mark associated with AI output.

Leterrier is a former Small World jury member and a past winner. He does not argue that software is forbidden in a modern microscope lab. He argues this processing went too far. “You have something very nice and very engaging visually but then it’s absolutely not representing a biological reality,” he said.

That gap, visual punch against cell structure, is also what people watching the official post kept returning to. The loud complaint is not the neon palette. It is that hairs seem to generate on the fly, that the red, blue, and magenta channels are never mapped to a stain a biologist would recognize, and that the unprocessed grayscale sequence has not been shown next to the prize file. Two former contest judges also said they had doubts about whether the image was real.

When you start looking at the scales of the movie and some of the details of it, it was quite clear to the community…that some of these features just are not consistent with nature.

Robert Hirst, lead scientist, Leicester University

Melanie White, a cell biologist at the University of Queensland who works with quantitative imaging, said the clip is beautiful and still hard to trust as a biological observation. “It’s very difficult to understand if what is being shown actually faithfully represents the biology.”

Nikon’s Rule Banned Generation but Not Processing

The written rule is short. “AI generated videos are not permitted and Sponsor holds the right to request the original video for verification.” Entries “generated by AI, script, macro or other automated means are void.” Videos must be taken by the entrant on a light microscope, cannot run past 60 seconds, and cannot add sound or graphics.

Generated is the word Nikon has to defend. Xu’s network, in his telling, is a visualizer of a reconstruction, not a camera that dreamed the beat. A Nikon Instruments spokesman said the company had reviewed Xu’s technical notes on the microscope, the capture, and the processing, and found those methods “technically legitimate and consistent with established microscopy practices.”

“Based on the information currently available, Nikon can confirm that an AI-assisted enhancement tool was used by the entrant as part of post-processing,” the spokesman said. Nikon treated that use, enhancement of microscope source data, not an image made from scratch, as inside the rules.

White’s split is the one the rule never wrote down. She accepts tools that lift information already in the capture. She rejects tools that invent information that was not there. She also said it is difficult to understand where Nikon is drawing the line.

Hickey put the same point in contest language. One rule, he said, was that you could not use generative AI to produce things. The other was that the footage had to come from a microscope. If the network painted in a cell body that the optics never resolved, those two sentences collide.

Other Finalists Shot Worms, Jellyfish and Leaf Cells

The rest of the top five is ordinary in the way Small World is ordinary: living samples, named optics, no claim that a network had to finish the picture. Nam Nhat Nguyen’s second-place DIC clip follows a tiny roundworm and a single-celled Dileptus. Benedikt Pleyer’s third-place darkfield clip holds jellyfish larvae in water droplets. Andrew Moore, a cell biologist at HHMI’s Janelia Research Campus, took fourth with neighboring cells dividing in time.

Hickey’s fifth-place confocal movie watches mitochondria in green and chloroplasts in red inside Turtle Vine leaf cells. That is the kind of clip the contest has sold for years: a dark room, a living specimen, a public who will never sit at that eyepiece. “If you spend a lot of your time in a dark room looking down a microscope, it’s really nice to be able to show people and to show the public,” he said.

He also named the bargain. There is, he said, a fusion of science and art, and many of the images scientists make already look abstract. The bargain only holds if the abstract picture is still a picture of the thing on the slide.

The Caption Changed From Capture to Rendition

Nikon did not pull the prize in September. It did move. After the first wave of comments, the company updated the winning entry page to say AI was used in post-processing. Then it changed the movie caption and called the clip a rendition.

Asked whether its stance had shifted, a spokesman said the company is reviewing the AI policy. “At this time, we don’t see that any rules were violated, but we are reinvestigating and re-evaluating the entry.” Xu, Nikon said later, was “respectfully complying” and had sent more technical documents.

HOW THE FILE MOVED IN PUBLIC

1. December 2025: Xu records the sequence on the fifth version of the custom microscope.
2. September 15, 2026: Nikon unveils him as the Small World in Motion winner.
3. September 17, 2026: The official account posts the clip and points to his Masters of Microscopy profile.
4. September 22, 2026: That profile describes the unsupervised network and tells others to embrace AI on their data.
5. Late September 2026: Nikon confirms AI enhancement, retitles the clip a rendition, and reopens the entry while saying it still sees no rule break.

The 2026 judging panel was Joe Dragavon of the University of Colorado Boulder, Quinten Geldhof of the Museum of Science in Boston, science writer Corey S. Powell, and Meredith Sagolla of Genentech. None of them has issued a public dissent of their own. The pressure is coming from people who were not in the room and who look at these cells for a living.

PCD Patients Do Not Have Cells Like These

Hirst’s sharpest worry is not the leaderboard. PCD is diagnosed in part by watching cilia beat, because the disease shows up in motion. Families already live with a rare, poorly explained condition. A prize clip that looks like a cartoon of their cells, he says, lands in the wrong place.

“What worries me most of all is that PCD patients will see this video and think, ‘My cilia look like this, and my cells look like this,’ and they just don’t, unfortunately.” Leterrier says the field’s fight is not with AI as a tool. It is with how far it was used, and how little of that use was on the label when the clip was crowned.

Xu has called the criticism understandable. AI is moving fast, he said, and people want to know what is real. The still-image Small World winners are due on October 21, 2026, under the same family of rules. Until Nikon finishes with Xu’s file, the motion contest’s claim that these movies are a shared language for living cells sits on a clip it now files as a rendition.