A ten-second animation sounds simple. In practice, a short video can contain several independent failure modes: a promising prompt that does not match the output, an incorrect aspect ratio, invented subtitles, persistent generator branding, and a technically valid file that still fails the visual joke.
This production note describes a single experimental Office Cats short from SolarClabs, titled The Sticky Note Disaster. It is a fictional cartoon, not documentary footage or a product demonstration. I describe what was measured, what was visibly checked, what was changed, and what remains unknown. I do not claim performance, audience retention, or independently verified music licensing. The final edited short has now been published on seven video destinations, and the provider publication receipts are linked below.
Start from the actual animation, not the prompt
The story uses two recurring characters: an orange tabby in a white office shirt and blue tie, and a smaller white kitten wearing pink. Their mishap revolves around one pink sticky note. The initial image already shows the orange cat with a pink note on his nose and a transparent container falling out of position. Around the last second, the white kitten manages to remove the note; it winds up on her own nose, and the orange cat laughs.
That is the intended premise. The editable source is a real uploaded MP4 named video (23).mp4. The post-production process was based on decoded frames from that file, not on a substitute slideshow or a rendered mockup of a prompt.
Confirm the media facts first
A media probe reported 720×1280 pixels, 24 frames per second, 240 video frames, and a 10.000-second H.264 video stream with stereo AAC audio. The container duration was 10.005 seconds. The source SHA-256 is e978cab85f38485c47cbafcc639f85a3237679017ff2ba72bd56e31411f0f435.
These details are useful because a source image may have been 1080×1920 while the generated animation itself was not. Advertising the final file as native 1080p would be inaccurate. A separately encoded derivative can be upscaled, but that does not create new source detail.
The production used a 1080×1920 output profile, 24 frames per second, H.264 and retained AAC audio. It preserved the ten-second story without padding. The final video stream was separately decoded from beginning to end after export. Technical success is not the same as creative success; it is one necessary layer of evidence.
Judge the first second without depending on captions
A generated title card can make a film look polished while hiding a weak visual premise. In this short, the opening frame already shows the accident, so editorial text is secondary. An overlay saying “ONE STICKY NOTE...” appears briefly in the upper part of the screen after motion has already begun. The wording is a creative hook, not dialogue and not a claim about something the cats said.
The opening was sampled at 0.00, 0.15, 0.30, 0.60, 0.90, 1.50, 2.00, 2.50 and 3.00 seconds. The visible box and the characters' reaction change, so the action does not depend solely on a camera zoom. The rest of the clip continues the same physical gag and ends on the white kitten's nose-note reaction.
Separate scene truth, subtitles and creative text
The plan requested no spoken words. But the plan cannot serve as a transcript for an independently generated video. The presence of an AAC stream proves that audio exists, not what the voices may have said. Without a verified listening or transcription pass, the correct choice is not to fabricate words or add a timed fake subtitle track.
Therefore, this package includes an editable SRT containing the editorial hook only; it is explicitly not a dialogue transcript. The export preserves the source audio. A separate full audio listening and rights audit has not been independently documented, so we do not claim verified dialogue transcription, audio licensing for every destination, or accessibility-caption completeness.
Handle a visible mark with localized care
Source-frame review found a small white star-shaped generated mark in the lower-right area throughout the short. The edit targeted that constant small region using local reconstruction, leaving both cats, most props, and the joke untouched. Comparison crops from the beginning, middle and end were used to assess visible residues. A regional correction is preferable to covering a whole corner with an oversized black box or shifting the shot enough to cut off important movement.
One independent drawback remains in the source: some incidental sticky-note writing looks like unstable AI-generated lettering. Unlike the persistent corner mark, that text changes with moving characters and props. Automatically erasing it risks introducing much more conspicuous smearing or warping. This limitation is reported instead of misrepresenting it as corrected.
Export and validate the derivative
The reviewed derivative contains a 1080×1920 H.264 video stream, 24 frames per second, exactly 240 video frames and a ten-second running time. The original AAC audio content was carried through rather than replaced by guessed dialogue. The output was probed and decoded again, then a contact sheet sampled the first second, intermediate beats, final reveal, and last visible frame.
The cover is a real frame extracted near 9.69 seconds, when the orange cat is laughing and the pink note is on the white kitten's nose. It uses a separate short headline and is saved as a 1080×1920 PNG. The source resolution is disclosed as 720×1280, upscaled for delivery.
Distribution is a separate decision
The editorial package prepares different native drafts for YouTube Shorts, TikTok, Instagram, Facebook Reels, Threads, Bluesky, X, Reddit, Pinterest, LinkedIn personal and company Page, plus DEV.to as a development article—not another video endpoint.
Those drafts are not identical cross-posted text. For example, the public entertainment destinations emphasize the visual gag, while a professional platform can discuss the technique and the evidence limits of the production itself. Platform tags, native topics, board IDs, subreddit flair and captions are not interchangeable fields.
On 14 October 2026, we used a guarded Postiz release via a self-hosted GitHub Actions runner. The release verified the uploaded MP4's SHA-256 before publishing, checked provider integrations and existing episode records to prevent duplicates, then published to one network at a time. Success required the exact Postiz record state PUBLISHED and an HTTPS release URL; a green workflow alone was not counted.
The GitHub Actions publication receipt shows seven confirmed destinations: YouTube Shorts, Instagram Reels, Facebook Reels, Threads, Bluesky, and both a LinkedIn personal post and Page post. This is 7/7 enabled video destinations, not 12/12 total planned channels; X, Reddit, Pinterest and TikTok remain disabled or require further configuration and provider approval. DEV.to is an article destination, not a video host.
The linked provider states are evidence of publication as reported by Postiz. They are not an independent playback audit, copyright clearance, nor proof that the audiences watched or enjoyed the video.
What this experiment can and cannot prove
We can prove that a particular ten-second source was decoded and converted to a specific technical profile, that a brief editorial hook was added, and that a localized corner mark was addressed. We can visually inspect the intended first-second accident and the final nose-note payoff. We cannot prove that the video will go viral, that real viewers will remain to the end, or that all cross-platform licensing and accessibility questions have passed without further review.
The seven provider public URLs and their Postiz IDs are now verified. The next meaningful evidence is 24-hour and 72-hour performance at comparable content age, including first-second abandonment, completion, replays, shares and comments, each interpreted in its own platform context. Those metrics have not yet been measured for this episode and no viral-performance claim is justified.
Reproducibility checklist:
- Read the original video stream metadata and checksum; never assume the prompt describes the actual output.
- Sample the first 0–3 seconds muted and the true last frame, then inspect the full source with sound.
- Distinguish editor-written hooks from transcribed dialogue; only caption audible words that have been verified.
- Preserve rights and provenance information, and report unresolved source artifacts rather than claiming perfect clean-up.
- Re-encode and decode-check the exact output, then review its cover and key frames at phone size.
- Reconcile old queue records and exact provider integration identities before writing to the social API.
- Require published state and a provider URL for every successful destination; retain a receipt for audits.
The broader lesson is simple: a valid ten-second MP4 is not proof of a successful ten-second story, and a submitted social post is not proof of a published one.
Top comments (0)