🔍 Read the full analysis: I Built An AI Video In One Evening—Here’s What I Learned on ThorstenMeyerAI.com
Get business pricing on monitors, keyboards and dev gear
- Business-only prices and quantity discounts
- Tax-exempt purchasing
- Multiple users, one account, clear invoices
TL;DR
Thorsten Meyer says he built a hand-drawn AI video engine in one evening, but the first full video it produced was slow and visually sparse. He measured the problems, changed the engine and made those measurements production checks; the account is a personal report, not an independent evaluation.
Thorsten Meyer says he built a hand-drawn video engine in one evening, then revised it after its first full video looked slow and sparse. In a personal account published November 6 on ThorstenMeyerAI.com, he describes how code review, human approval and measurable production checks shaped a later version that went live on the site.
Meyer says the initial idea came from a hand-drawn explainer he admired. To avoid copying it, he wrote down a short set of general style traits—a hand-drawn line, a continuous visual world and large numbers used as color accents—and says he used no drawings, assets, layouts or drawing code from the original. The engine used his own design system, and he says its handwriting font was openly licensed.
According to Meyer, Claude built the engine and OpenAI’s Codex reviewed it without contributing code. He says Codex identified five issues in its first review, which were fixed that evening. Each finding was paired with a test that, he says, failed against the old code before the fix was applied. Meyer also says he watched and approved a roughly one-minute test video before it was published within an article about local votes against data centers.
The first full video made the following morning did not meet his expectations. Meyer describes it as slow and boring, citing excess white space, icons that did not support the narration and drawings that stopped moving once completed. He says that video remains online unchanged as an example of the shortcomings. After revising the engine, he compared the first three scenes: frame coverage rose from 15% to 66%, still time fell from 29 seconds to zero, and camera moves increased from 3 to 27. Those figures apply to those three scenes, not necessarily to every video.
Infographic 01 · The new video look
One evening, one boring first video, one measured fix
Speed was not the point. The steps in between were: a second AI that reviews, a human who says yes, and numbers that became gates.
Share of the frame filled with content on the first three scenes after the fix. Before: 15%.
From an evening idea to a measured fix
Relative times, this week
- EveningA video I likedClean room: only a list of style traits, no drawings, assets or layouts
- Same eveningOwn engine, first reviewClaude builds; OpenAI’s Codex reviews; 5 findings fixed, each with a test that fails on the old code
- Same eveningA human says yesOne-minute test video approved and published
- Next morningFirst full video: slow and boringToo much white space, icons that did not fit, a picture that stood still
- Hours laterMeasured fix, then gatesThe first video in the improved look goes live
Source: my own build notes for the new video look, this week (engine history, review round, approval, critique, before/after measurement of the first three scenes).
Powered by Thorsten Meyer AI
How Video Metrics Became Release Checks
The account’s practical point is that generating a video quickly does not establish that it is effective or ready to publish. Meyer describes using human review to make the release decision, then turning specific criticisms into measurable checks. In his process, a scene that is too empty can be stopped before rendering, while a visual object that does not match what is being said can trigger a build warning.
That approach offers a concrete example of how a small publisher might put AI-assisted creative work into a production process: use one model to build, a separate system to review, and tests to catch known defects. These safeguards can help expose particular problems, but the account does not show that they establish overall quality or guarantee that a video is accurate, engaging or free of errors.
Meyer’s other example underscores that distinction. He says a model completed 12 of 15 answers in a legal benchmark after a weight-editing experiment, compared with four beforehand, while its legal score remained zero. A higher completion count alone did not amount to better legal performance. The result is Meyer’s description of his own experiment, and the article excerpt does not provide enough detail to independently assess the benchmark or method.
Popular right now
Apple AirPods Pro 3 Wireless Earbuds with Active Noise Cancellation
Ninja CREAMi 7-in-1 Ice Cream Maker, NC301
Dyson Ball Animal 3 Upright Vacuum – Corded, 290 AW, De‑tangling Motorbar Cleaner Head, 3 Suction Modes
As an affiliate, we earn on qualifying purchases.
From Style Reference to Original Engine
Meyer frames the video project as a personal account of taking an AI idea into production, with mistakes included. The first test video was about one minute long and, he says, went live the same evening he approved it. The first full video followed the next morning; after his critique and changes to the engine, an improved-look video about the Nasdaq’s worst day in 19 months went live a few hours later. The source does not give the date of that market event.
His article also describes a prior monitoring failure from August: a guard reportedly wrote about 1,300 warnings to a log that nobody read, while an alert email did not go out. Meyer uses the incident to argue that monitoring is useful only if it reaches a person. The warning count is a raw figure in the account; no comparison period or baseline is supplied.
The supplied article includes a roughly four-and-a-half-minute explainer about the process, with narration described as an AI voice clone of Meyer. That video and the account are produced by the same publisher. The material provided does not include outside review, source code, independent testing or evidence that the described results have been reproduced elsewhere.
“A demo that runs is not a decision.”
— Thorsten Meyer, in his November 6 article
What the Report Does Not Establish
The reported video measurements cover the first three scenes of the rebuilt video. The account does not provide a full-video measurement, a repeatability analysis, a comparison with other video tools or an independent assessment of the finished work. It is also unclear how the checks judge whether an object genuinely illustrates the narration, or how often they produce false alarms or miss problems.
The article presents Meyer’s own build notes and experiments; readers are not given the underlying code, complete test results or detailed benchmark methodology in the supplied material. The legal experiment’s increase in completed answers does not establish improved legal ability, especially since Meyer says its legal score remained zero. The source also does not specify whether the publishing checks catch factual errors, licensing problems or other issues beyond the described visual and code findings.
Further Videos Will Test the Gates
Meyer says the improved video look is being used for the explainer published with his account, and that the measurements have become checks in the build process. The next test is whether those rules continue to identify sparse scenes and mismatched imagery in later videos, rather than only improving the examples described in this one report.
The article does not announce a release schedule, a public evaluation or plans to publish the code. For now, the claims rest on Meyer’s account of his own workflow. Further videos and disclosed test results would help readers judge how consistently the process works.
Key Questions
What did Thorsten Meyer build?
He says he built an engine for hand-drawn-style explainer videos, using a continuous visual world, camera movement and design elements from his own system.
How did the first full video perform?
Meyer says it looked slow and boring, with too much white space, weak connections between icons and narration, and drawings that stopped moving when complete. He says it remains online unchanged.
What changed in the revised video?
For the first three scenes, Meyer reports that frame coverage rose from 15% to 66%, still time dropped from 29 seconds to zero, and camera moves rose from three to 27. Those figures describe those scenes only.
Did Codex build the video engine?
No. Meyer says Claude built the engine, while OpenAI’s Codex reviewed the code and reported issues. He says fixes were paired with tests designed to fail on the old code.
Are the results independently verified?
The supplied material is Meyer’s first-person account. It does not provide independent verification, the underlying code or enough benchmark detail to reproduce the reported results.
Source: ThorstenMeyerAI.com
