150 AI actors. 220+ visual styles. 50+ models. Zero prompt engineering.
Melies outputs structured JSON and ships with a SKILL.md file. Any agent that can run CLI commands can generate images and videos.

by Anthropic
Claude reads the SKILL.md and uses the CLI as a tool. Ask in plain English, it handles the rest.
# You say:
"Generate a cinematic portrait of Mei in golden hour lighting, Ghibli style"
# Claude runs:
melies image "cinematic portrait in a café" \
--actor mei --art-style ghibli \
--lighting golden --sync
# Output:
✓ Saved to portrait-cafe.webp
{ "assetId": "abc123", "url": "https://..." }

Open-source AI agent framework
OpenClaw auto-discovers Melies as a skill globally. Install once, use from any agent.
# You say:
"Create 4 YouTube thumbnails with Aria looking surprised"
# Agent runs:
melies thumbnail "shocked face reacting \
to AI news" --actor aria \
--expression surprised -n 4 --sync
# Output:
✓ thumbnail-1.webp
✓ thumbnail-2.webp
✓ thumbnail-3.webp
✓ thumbnail-4.webp
Full creative toolkit accessible via simple CLI flags. No prompt engineering.
150+ ready-to-use characters with full identity consistency across every generation. Pick an actor by name, or upload any face to create a custom AI actor.
# Generate with a built-in actor
melies image "talking to camera while \
walking in park" --actor hailey \
--aspectRatio 9:16 --lighting golden --sync
# Create a custom actor from any face
melies actor-create "jean-pierre" \
--imageUrl photo-of-jean-pierre.jpg
# Browse all actors
melies actors

--actor hailey
--actor dante
--actor aria
--actor bjorn
--actor elena
--actor dae
--actor mei
--actor axel
--actor camille

Art styles, lighting setups, camera angles, expressions, moods, weather, eras. Your agent combines flags to get the exact look. No prompt engineering.
# Visual style preset
melies image "portrait" --actor mei \
--sref lego-set --sync
# Combine cinematic flags
melies image --actor mei \
--camera arri-alexa --expression grieving \
--shot wide --composition asymmetric \
--camera-angle high --sync
One command generates an image and animates it into a video. Your agent gets a complete visual asset without chaining steps.
# One command: image + video
melies pipeline "neon Tokyo street" \
--actor hailey --camera arri-alexa \
--lighting neon --movement tracking --sync
# Preview cost before generating
melies pipeline "neon Tokyo street" \
--actor hailey --best --dry-run
# Output:
Image: nano-banana-2 (15 credits)
Video: veo-3.1 (400 credits)
Total: 415 credits
Hailey
ARRI Alexa
Neon


Movie posters with title and credits. YouTube thumbnails in batch. Image upscaling and background removal. Everything your agent needs for content production.
# Movie poster with two actors
melies poster "Midnight Protocol" \
--actor dante --actor elena \
--style anime --lighting neon --sync
# Batch thumbnails
melies thumbnail "AI news" --actor aria -n 4
# Upscale + remove background
melies upscale --imageUrl photo.webp --sync
melies remove-bg --imageUrl photo.webp
Every command returns structured JSON. Use --dry-run to preview cost before generating.
melies imageGenerate images from text prompts. 30+ models, actors, art styles, lighting, camera angles, and more.
Full documentation--actorAI actor for consistent characters--art-styleArt style (anime, ghibli, noir, oil...)--lightingLighting setup (golden, neon, noir...)--aspectRatioOutput ratio (1:1, 16:9, 3:4...)--qualityUse a higher quality model--syncWait and return result URL Use --fast, --quality or --best and Melies will automatically pick the best model for your task. Or specify any model directly with --model.
--model flux-kontext--model flux-dev--model flux-pro-ultra--model nano-banana-2--model seedream-4.5--model grok-imagine--model veo-3.1--model kling-o3--model kling-v3--model seedance-v1--model hailuo-02--model wan-v2.2--model ltx-2-pro--model runway-gen4--model luma-ray-2--model elevenlabsInstall the CLI and start generating images and videos from your terminal in under a minute.
npm install -g melies