AsiafeedTech — April Update: AI Image Gallery & Video Engine Optimization
A quick update on what shipped this month in the AsiafeedTech pipeline.
AI Visual Gallery
Every article now gets a unique cover image generated by Imagen 4.0. The prompt is built from the article title + category context + a fixed cinematic style directive. Images are saved to persistent storage and served via a proxy endpoint in the main backend — the browser never talks to the generator service directly.
You can browse the gallery at asiafeedtech.com/gallery.
Video Engine Optimization
The Shorts pipeline went through several rounds of improvement:
-
aformatnormalization per stream for consistent audio sync in complex FFmpeg filter graphs - Dynamic font sizing for title cards based on character count
- Whisper caption overflow fix via
WrapStylein the ASS template - Improved Imagen prompt structure — category-specific visual concepts per news category
TikTok Integration
The TikTok Content Posting API app is currently under review. Cross-posting will be automatic once approved — same flow as YouTube upload.
Stack
Angular SSR · Spring Boot · PostgreSQL · Flyway · Gemini · Imagen 4.0 · FFmpeg · Whisper · Docker · Linode
Top comments (0)