Captions that actually post, and sharper video direction
Every finished episode now gets a caption and hashtags, and the instructions behind each clip were rebuilt from scratch.
Captions and hashtags now save every time
Finished episodes were quietly ending up with no caption and no hashtags. The caption was being written correctly and then thrown away over a formatting detail, so nothing ever reached the preview page. Fixed, and every existing episode was backfilled.
Captions sound like your show again
Captions were being written without the show's tone or name attached, which made them read generically no matter which series they came from. They now carry the show's actual voice, and can use up to three emoji when they fit the moment.
Rebuilt the instructions behind every clip
The brief sent to the video model now follows one fixed structure — who is on screen, what must not change, what happens, how the camera behaves, who speaks, and what the audio carries. Tighter and more specific, which means fewer wrong-speaker and drifting-character clips.
Shot framing accounts for your cast size
A close-up was being chosen for scenes that had four people in them. Shot framing is now reconciled with how many characters actually have to fit in the frame.
Also fixed
- A silent scene no longer tells the video model to animate someone speaking.
- Scene descriptions can no longer be cut short by a long show style or character description.
- Hashtags are de-duplicated and capped at eight.