Message Renderer Module

Renders LLM-generated conversations as animated iMessage-style videos with synchronized audio.

MessageRendererService

render(conversation: Conversation, post: RedditPost): Promise<Result<GeneratedVideo, MessageRendererError>>

Orchestrates the full rendering pipeline:

1. Build narration script — converts conversation messages into readable narration text
2. TTS synthesis — generates audio with word-level timestamps via TtsService
3. Frame generation — renders PNG frames for each message bubble state
4. Duration scaling — maps frame durations to match audio length
5. FFmpeg composition — muxes frames + audio using concat demuxer
6. Cleanup — removes intermediate frame images

Sub-components

FrameGenerator

Generates individual PNG frames representing the progressive state of the iMessage conversation. Each new message adds a bubble to the thread.

buildNarrationScript(conversation: Conversation): string

Converts a Conversation into a narration script that reads naturally when spoken by TTS. Includes speaker attributions and message text.

Configuration

Variable	Default	Description
MESSAGE_TARGET_COUNT	14	Target messages per conversation
MESSAGE_FRAMERATE	10	Video frame rate
MESSAGE_TYPING_DURATION_MS	800	Typing indicator display time
MESSAGE_MIN_HOLD_MS	600	Minimum hold time per message

Error Codes

Code	Description
FRAME_GENERATION_FAILED	PNG frame rendering error
FFMPEG_FAILED	Video composition error
TEMPLATE_LOAD_FAILED	iMessage template missing
NARRATION_FAILED	TTS synthesis or alignment failure