Novel chapter → storyboard → AI images → AI videos → narration. Powered by Agnes AI multi-modal API.
⏱ Expect long wait: Video generation is rate-limited to 1 video per minute (Agnes API). A 6-shot story takes ~30-50 minutes total. Your job persists in browser storage — you can close this tab and come back.
Storyboard
Initializing…
0%
Style:
Characters:
Done!
All shots generated. Download individual videos or use a tool like FFmpeg to merge them into one MP4.
Per shot: (a) t2i for first appearance, i2i with character ref for repeats, (b) i2v creates ~3-4s video, (c) vision QA checks 穿模/角色/风格, (d) CF Workers AI TTS for narration.
Pipeline is resumable: Job state stored in Cloudflare KV (7-day TTL). Close tab → re-open with same jobId → resume.
Why so slow?
Agnes free tier limits video generation to 1 video per minute (server-enforced). A 6-shot story is 6 minutes of queuing + 5 min/shot generation ≈ 35 minutes. The orchestrator waits 60s between video calls automatically.
What's missing vs. my full request?
❌ Character dialogue lip-sync (we use voiceover instead)
❌ Frame-by-frame consistency check across shots (only first-frame QA per shot)
✅ Video native audio (SFX/ambient/music auto-generated by Agnes)
✅ Visual QA (anatomy, prompt adherence) via vision LLM
✅ Character consistency via i2i reference image
Privacy & limits
Agnes free tier may use submitted text/images for model improvement. Don't submit sensitive material. The job state (text, plan, URLs) is stored in Cloudflare KV for 7 days. Videos are stored on Agnes's CDN with short TTL — download promptly after generation.
Comments & Ratings