Text → PNG via OpenAI Images (gpt-image-1/dall-e-3) or OpenRouter multimodal chat (Gemini Flash Image, Flux, ...).

Narration script → N-shot storyboard: LLM plans shots, image model renders. Feeds video-slideshow.

Prepend branded intros and append outros to a video with smooth crossfades. Auto-normalises mismatched formats.

Overlay timed text bands (lower-thirds, chapter markers, call-outs) on a video. One ffmpeg pass.

Concatenate N video clips into one MP4, normalising resolution/fps, with optional crossfade.

LLM-readable video brief — metadata + silence gaps + scene changes + thumbnails. Pure ffmpeg, no cloud.

Picture-in-picture composition: overlay an inset video on a main video with configurable position, scale, border.

Images + optional audio → MP4 slideshow with Ken Burns zoom and xfade transitions. Pure ffmpeg.

Add YouTube / QuickTime chapter markers to an MP4 via ffmetadata — no re-encode, just metadata.

Burn SRT/VTT subtitles into the video stream with full styling control. ffmpeg-only.

Remove silences from a video with inaudible crossfades. ffmpeg-only, no cloud.

Mix a music bed under a video with automatic sidechain ducking. ffmpeg-only, no cloud.

Speed up or slow down a video (0.25x-4x) with pitch-preserving atempo chain for natural voice.

Text → narrated WAV + SRT. edge-tts / say / piper. Pure local, cross-platform.