How ather86/ai_projects_telegram works
This repository implements a Telegram-driven AI video-generation bot, with a separate alert webhook and a small command-line ComfyUI bridge. The main workflow accepts a script or prompt, prepares scenes and prompts, submits rendering work to ComfyUI, and processes the resulting clips into a video. The graph reflects the sampled Python excerpts and the workflow file in the tree; some Telegram handlers and processing details were not included, so uncertain connections are omitted.
Main entry points: telegram_gate.py Telegram bot and webhook · video_processor.py Scene and video processing · comfy_bridge.py Standalone bridge · video_ltx2_3_i2v.json LTX image-to-video workflow
Components 14
Bot and alerts
- Telegram bot and webhook
telegram_gate.py
Video generation
- Scene and video processing
video_processor.py
Rendering integration
- ComfyUI API and renderer
- Render queue
- LTX image-to-video workflow
workflow_api/video_ltx2_3_i2v.json
Standalone bridge
- Standalone bridge
comfy_bridge.py
Other components
- Telegram user
- Telegram platform
- Ollama LLM
- FFmpeg tools
- Alert sender
- ComfyUI input images
- Latest input image
- Generated video
Connections 16
- Telegram user → Telegram platform: sends request
- Telegram platform → Telegram bot and webhook: delivers updates
- Telegram bot and webhook → Telegram platform: replies and sends alerts
- Telegram bot and webhook → Scene and video processing: processes video request
- Scene and video processing → Ollama LLM: parses script
- Scene and video processing → FFmpeg tools: stitches clips
- Telegram bot and webhook → LTX image-to-video workflow: loads selected workflow
- Telegram bot and webhook → ComfyUI API and renderer: submits render prompt
- ComfyUI API and renderer → Render queue: enqueues render
- ComfyUI API and renderer → Telegram bot and webhook: streams render events
- Alert sender → Telegram bot and webhook: posts alert
- Standalone bridge → ComfyUI input images: searches latest image
- ComfyUI input images → Latest input image: selects image
- Standalone bridge → LTX image-to-video workflow: loads workflow
- Standalone bridge → ComfyUI API and renderer: submits render prompt
- Scene and video processing → Generated video: produces video