With Korea Alive, I wanted someone new to Korean history to be able to follow a complete story in a short video. An object could be explained through its structure; a historical figure could be followed through a consequential choice. English narration and captions carried the explanation. By the YouTube Studio check on 5 September 2026, four Shorts were public: Jagyeongnu, Yi Sun-sin, the Baekje gilt-bronze incense burner and Myeongnyang.
The Baekje incense-burner film was published on 4 September and the Myeongnyang film on 5 September. The work had moved from preparation and scheduling to actual releases, but a repeatable way of making the next films was still being worked out.
A name to pronounce and a name to display
For the Baekje film, I adjusted how the English voice read the Korean name. The speech input used beck-jeh, while the public caption retained Baekje. Replacing the name on screen with a phonetic spelling would make it harder for a viewer to look up. The production therefore kept separate spoken and displayed strings, even when they belonged to the same sentence.
The actual configuration preserves that distinction. It uses Typecast’s Rusty voice and the final speech timing to align captions. The local Baekje master is a 38.5-second, 1080×1920 portrait video, with the background sound reduced under narration. Duration and caption counts were only part of the problem: the name had to be intelligible, and its caption needed to appear at the right point in the speech.
AI assisted with script preparation, generated video and speech, and editing automation. I chose the subjects and format and reviewed what needed changing. Generated scenes are not footage of the historical event, so I also made the distinction between documented facts and visual reconstruction part of the production criteria.
A release did not end the editing questions
Watching the Baekje film after its 4 September release, I noticed a jarring transition. A usable individual shot could still feel wrong beside the next one when its scale, brightness or movement changed abruptly. I set a requirement to adjust the cut points, use a short overlap where appropriate, and watch the complete film again after editing. A new rule did not retroactively fix the released film.
The next crown episode exposed a different failure. The 17-second preview I saw on 6 September felt like a sequence of still photographs. I rejected it as a finished film and revisited the production approach. Passing caption, loudness and file checks did not make it the moving film I had asked for. That preview was not published. On 6 September, I stopped work to reconsider the production approach.
Rebuilding the production process in October
Production resumed in October with three completed films on Angbuilgu, Jeong Yak-jeon and Changdae, and Busan Cinema Center. Each was saved as a scheduled post on YouTube, Instagram and TikTok. On 19 October I confirmed that the Angbuilgu film, Why Put Animals on a Public Clock?, was public on YouTube. I keep saved schedules separate from confirmed releases.
In the revised workflow, I check what actually moves within a scene rather than relying on effects that move a still photograph. After reviewing the individual scenes, I bring them together with narration and English captions, then watch both the assembled edit and the final 1080×1920 file from beginning to end at normal speed. File-format checks alone no longer decide whether a film is finished.
For the hanok folding-door film, I used an explanatory 3D model to show a folded pair of door leaves rising around its upper horizontal axis. The model makes the pivot visible; it is not a measured reconstruction of a particular house. A historical scene and a mechanism explanation need not use the same production method.
From a release to a repeatable process
The Baekje film had 1,172 YouTube views in the 5 September check. That is a dated view count, not a comparison of videos observed at equal ages. It does not establish sustained repeat viewing, attributable subscriptions or revenue.
By 19 October, I had also completed six follow-up films as 1080×1920 masters and reviewed each from beginning to end at normal speed. Those six are finished production files, not yet scheduled posts or confirmed releases. In September, the process had stalled over how to make the next film; it now carries scene production, narration, caption editing and final review across several films.