Polished explainers, with a person's sign-off at every stage.
An Apple Silicon media-production studio, built first for short illustrated explainer videos, with a separate human approval at each stage. The name is provisional; a rename to FocalaneProductions is planned.
The idea
Short explainer videos are easy to generate and hard to trust. Automated pipelines tend to blur where a claim came from, whether the voice is the one that was chosen, and who actually approved the final cut. This studio is meant first to make video for Focalane's own social channels, where polish matters and every published claim needs someone who stands behind it.
As FocalaneProductions, it would make polished explainer videos, talking-head video inserts, carousels, and other formats for Focalane's Instagram and TikTok. Script and sources, voice, visual plan, review, and final export would each pass a separate human approval. The bet is that a high creative bar, set first and checked by a person at each stage, together with a few reusable illustrated mechanisms, can make quality repeatable instead of lucky.
A second, more exploratory line would make illustrated shorts and mini-documentaries for TikTok, Instagram, and YouTube, starting with licensed Pexels stock footage, possibly adding a Higgsfield MCP later, and reaching videos of eight minutes or more on a longer rollout. Today it is a version 0.8 release candidate in engineering: a first narrated film has an approved script and accepted voice quality, while the social formats and the rename do not exist yet.
How it works
Script + evidence
Voice + visual plan
Review gates
Authorized export
One production service is the only state-changing boundary; the app, MCP server, jobs, and renderers all call it. Accepted narration sets the timeline, caption timing comes from measured word boundaries or a person's waveform correction (never proportional estimates), and each gate is a fingerprinted human decision.
Where it stands today
Version 0.8 is a release candidate in engineering, and the name Production Studio is provisional while a rename is pending. Its isolated Engineering Lab runs script and evidence editing, the first approval gate, voice sampling, and narration bound to the chosen sample: a first original narrated film of about three minutes has an approved script and accepted voice quality, and its caption timing review is still open. A private four-act benchmark study, a credited close remake of a reference clip and not for publication, was accepted for creative fidelity on 24 September 2026. The three reusable illustrated mechanisms exist only as engineering drafts, the first two short explainers are not built yet, and the app has not been configured or adopted for Christian's own production work.
Working now
Local Apple Silicon runtime with Metal rendering and VideoToolbox encoding, checked against software output
Separate fingerprinted human approvals for script, voice, visual plan, video review, and final export
Narration bound to the exact voice sample the operator heard and chose
Human-reviewed caption timing against the waveform; automatic word timing only prefills candidates
Frame-by-frame capture receipts, with hardware versus software encode parity checks
A bounded MCP server on the same service: agents need a locally approved pairing, and publishing is absent
Not there yet
Caption timing acceptance for the first film, qualifying the three reusable mechanisms, the first two short explainers, and publishing remain open.
How it got here
Converted from an earlier production engine into an empty, neutral Apple Silicon app, keeping only the reusable engine, safety, agent-interface, and Apple Silicon work
Goal approved: a reusable illustrated-video foundation of three mechanisms and two unrelated short explainers
Benchmark approach changed: private, credited remakes of reference clips set the creative bar first
Four-act benchmark study accepted; the first original narrated film's script approved at the first gate
The first film's voice quality accepted; a rename to FocalaneProductions and a social-video direction decided
What comes next
Next step
Two unrelated 30–60 second explainers built from three reusable illustrated mechanisms, after the first narrated film's caption timing review.
Later ideas, not current features
Finish caption timing review for the first narrated film, then build three reusable illustrated mechanisms and two unrelated 30–60 second explainers
Choose the next real pilot once that foundation passes
Rename to FocalaneProductions and make polished explainer videos, talking-head video inserts, and carousels for Focalane's Instagram and TikTok
A separate line of illustrated shorts and mini-documentaries for TikTok, Instagram, and YouTube, starting with licensed stock footage from the Pexels API
Possibly later: AI video-generation tools, such as a Higgsfield MCP, for more precise control
Later still: videos of eight minutes or more, on a longer rollout
What I checked, and what stays bounded
Dated evidence
Clean Apple Silicon installation accepted on synthetic fixturesclean-install
Synthetic workflow and rendering acceptancefixture
Decoded-media quality and parity checksquality
Private four-act benchmark study (a credited remake of a reference clip) accepted for creative fidelitybenchmark
Four-act study captured in one session: all 1,168 frames receipted; the study's hardware and software encodes agree (VMAF 97.7, SSIM 0.999)parity
First original narrated film: script and evidence approved at the first gategate
First narrated film: voice quality accepted; caption timing review openvoice
Engineering foundation verified: 381 of 381 application tests and 4 of 4 MCP tests passtests
Boundaries it keeps
Not configured for Christian's own operations
Publishing is unavailable
The three reusable mechanisms are engineering drafts, not qualified reusable components
The four-act study is a private, credited remake of a reference clip: a benchmark, not original or publishable work
Automatic caption timing did not meet the one-frame reference, so timing needs a person's review
The name is provisional; the rename is pending
Talking-head inserts, carousels, stock footage, AI video generation, and long-form video are plans, not current capabilities
No customers, outcomes, public source release, or reusable package are claimed
My role
Set the product direction, the human-approval model, and the acceptance bar; made every gate and creative-benchmark decision; and reviewed and tested the implementation, which Codex largely wrote.