https://whop.com/videolab-073d/free-starter-pika-h3-flex-movie-scenes
FULL SYSTEM ($50): https://whop.com/videolab-073d/pika-h3-flex-movie-scenes-photos-to-video-sound-in-6-steps — use code FLEX30 for 30% off ($50 → $35.00).
## FREE vs FULL
| | FREE (this download) | FULL ($50) |
|---|---|---|
| Workflows | 1 — 1-photo reference-to-video, 5 s lite | 4 — T2V + 1-photo + 3-photo multi-shot + dialogue recipe |
| VSA speed gate | — | Yes |
| Semantic Bridge switch | — | Yes |
| 15 s multi-shot + 2x upscale | — | Yes |
| Dialogue + sound + music recipe | — | Yes |
| Prompt library | — | 10 shot prompts + dialogue + camera + sound |
| Guides | This README | Full EN quickstart + Arabic guide |
## What this is
Pika H3 Flex (Pika Labs) on MiniMax H3, in ComfyUI. 1–3 of your own photos + a shot-by-shot prompt → short movie scenes with camera moves, dialogue, sound and music together, in 6 sampling steps. Pika released the H3 Flex adapters 29 Oct 2026; the first tested local workflow landed 5 Nov 2026; ComfyUI added native nodes 10 Nov 2026. This free lite workflow gives you the core graph — one photo, 5 s at 1056x608, 6 steps, CFG 1, sampler lcm — ready to run in the newest ComfyUI.
## Files you need (not included)
From Comfy-Org/MiniMax-H3: minimax_h3_ref2va_pruned_int8_convrot.safetensors (~21 GB) → diffusion_models, qwen3vl_32b_minimax_h3_nvfp4_awq.safetensors (~16 GB) → text_encoders, minimax_h3_video_vae_fp16.safetensors (~5.2 GB) + minimax_h3_audio_vae_fp32.safetensors (~600 MB) → vae, pika_h3_flex_ref2va_v1_bf16.safetensors (~2 GB) → loras. 12 GB+ VRAM (the author measured 1m43s for a 5 s clip on a 20 GB RTX A4500; a 5 s clip fits under simulated 12 GB and 16 GB caps).
## License warning
MiniMax H3 Community License — excludes US / EU / UK / South Korea without a separate licence from MiniMax; commercial products must display "MiniMax H3" prominently (a "Powered by MiniMax H3" notice is encouraged). Read the model card before commercial use. v1 caveat: reference-audio lip sync is unsupported.
Built from the official ComfyUI docs and the Civitai resource; structure-checked, not GPU-tested in our sandbox.
Full pack on Whop: https://whop.com/videolab-073d/pika-h3-flex-movie-scenes-photos-to-video-sound-in-6-steps

