You're watching faceless AI channels blow up. My Opus 5.5 already built the factory that makes them
9 rooms. 1 conveyor. every room is a free open-source repo:
SHORTS FACTORY
┣ 01 scout · yt-dlp: grabs the videos that already went viral
┃ ┗ yt-dlp/yt-dlp
┣ 02 transcribe · whisperX: turns them into word-timed scripts
┃ ┗ m-bain/whisperX
┣ 03 voice · chatterbox: narrator voice with real emotion
┃ ┗ resemble-ai/chatterbox
┣ 04 art · ComfyUI: characters + scenes, same style every time
┃ ┗ Comfy-Org/ComfyUI
┣ 05 motion · Wan2.2: stills become moving shots
┃ ┗ Wan-Video/Wan2.2
┣ 06 lip-sync · LivePortrait: the characters actually talk
┃ ┗ KlingAIResearch/LivePortrait
┣ 07 edit · remotion: the long-form cut, written in code
┃ ┗ remotion-dev/remotion
┣ 08 clips · AI-Youtube-Shorts-Generator: one long video → vertical shorts
┃ ┗ Anil-matcha/AI-Youtube-Shorts-Generator
┗ 09 post · postiz: schedules YouTube, TikTok, Reels
┗ gitroomhq/postiz-app
558K github stars. $0 in licenses. one long video in, a stack of shorts out
the screens on the roof play real renders, not mockups:
→ same little claude in 7 styles: pixel, paper, clay, toon, voxel, neon, 1930s noir
→ one style per channel. the factory doesn't care which
the whole video is one blender file claude wrote. every sound is generated in python
how to wire the 9 rooms together is in the article below ↓
try this before you scroll past another channel you could have built
12 likes · 1 repost · 697 views
Mentions
You're watching faceless AI channels blow up. My Opus 5.5 already built the factory that makes them 9 rooms. 1 conveyor. every room is a free open-source repo: SHORTS FACTORY ┣ 01 scout · yt-dlp: grabs the videos that already went viral ┃ ┗ yt-dlp/yt-dlp ┣ 02 transcribe · whisperX: turns them into word-timed scripts ┃ ┗ m-bain/whisperX ┣ 03 voice · chatterbox: narrator voice with real emotion ┃ ┗ resemble-ai/chatterbox ┣ 04 art · ComfyUI: characters + scenes, same style every time ┃ ┗ Comfy-Org/ComfyUI ┣ 05 motion · Wan2.2: stills become moving shots ┃ ┗ Wan-Video/Wan2.2 ┣ 06 lip-sync · LivePortrait: the characters actually talk ┃ ┗ KlingAIResearch/LivePortrait ┣ 07 edit · remotion: the long-form cut, written in code ┃ ┗ remotion-dev/remotion ┣ 08 clips · AI-Youtube-Shorts-Generator: one long video → vertical shorts ┃ ┗ Anil-matcha/AI-Youtube-Shorts-Generator ┗ 09 post · postiz: schedules YouTube, TikTok, Reels ┗ gitroomhq/postiz-app 558K github stars. $0 in licenses. one long video in, a stack of shorts out the screens on the roof play real renders, not mockups: → same little claude in 7 styles: pixel, paper, clay, toon, voxel, neon, 1930s noir → one style per channel. the factory doesn't care which the whole video is one blender file claude wrote. every sound is generated in python how to wire the 9 rooms together is in the article below ↓ try this before you scroll past another channel you could have built
Here you go: Anil-matcha/AI-Youtube-Shorts-Generator — thanks for commenting GEM!
Before you automate YouTube clipping, test one video end to end AI YouTube Shorts Generator documents a CLI + Python workflow for producing vertical shorts. Start with one representative video and check crop quality, caption timing and setup effort before scaling. This is a documentation review; we haven’t run that pipeline end to end. Source: Anil-matcha/AI-Youtube-Shorts-Generator
5. AI SHORTS GENERATOR 从长视频里找高光，自动切竖屏。 YouTube 长视频 → 多条竖屏片段。 Anil-matcha/AI-Youtube-Shorts-Generator