GUIDE

Master CharaMaker
in 10 minutes.

Step-by-step walkthrough to design AI characters, generate scene images, and produce viral motion-transfer videos.

01 Getting started

CharaMaker is a creative studio for designing AI characters and turning them into images and motion-transfer videos. Here's the basic flow:

  1. Sign in with Google. New accounts get 500 free credits to try the studio.
  2. Design your character. Give it a name, vibe, niche, and look — the AI fills in the rest and creates a profile image.
  3. Generate scene images. Place your character in any setting, with consistent face and identity.
  4. Generate motion videos. Upload a reference video — your character performs the same motion.
  5. Caption + share. Auto-generate platform-ready captions and post to your channels.
💡 Free credit tip
Your 500 starter credits are enough for ~1 character + 2 image generations. Make them count.

02 Designing a character

Open the studio and go to the Create Character tab. You'll see four input fields, all optional:

How to write a strong appearance

The AI honors your wording, so be specific about the things that matter to you. Vague inputs let the AI improvise.

Vague — AI improvises:
cute girl, K-pop style
Specific — AI follows closely:
early 20s Japanese woman, doll-like delicate face, slim tall figure, long straight black hair with bangs, wearing Y2K streetwear, K-pop idol aesthetic

Reference image (optional)

Upload a reference photo to anchor the visual style. The AI extracts the appearance (clothing, hair, vibe) without copying the face — useful when you have a specific aesthetic in mind.

⚠️ Real-person names
Mentions like "looks like Wonyoung" are abstracted into descriptive features (slim tall figure, doll-like face, etc.) to respect privacy and avoid copyright issues.

03 Generating images

Switch to the Generate Image tab. Pick a character profile, write a scene prompt, and choose an aspect ratio.

Aspect ratios

How to write a scene prompt

Describe the scene, pose, clothing, and mood. Don't repeat the character's appearance — the AI already preserves face and hair from the profile.

Good prompt:
sitting at a Brooklyn rooftop café at golden hour, wearing a cream silk slip dress, holding an iced matcha, candid laughter, warm soft lighting

Reference image (optional)

Upload any image — the AI analyzes it and uses its style/scene/composition for your generation. The face still belongs to your character.

💡 Identity is preserved automatically
You don't need to write "same face as before" — the system locks the character's identity. Focus your prompt on the new scene.

04 Generating videos

The Generate Video tab does motion transfer. You provide:

  1. A character profile (the face you want)
  2. A reference video (the motion you want, max 15 seconds)

The system extracts a frame, swaps in your character's face and hairstyle, then animates the entire clip with your character performing the original motion.

What works best

What to avoid

⏱️ Generation takes 5-10 minutes
You can leave the page or open new tabs — your video will keep generating in the background. Check the History tab when you return.

05 SNS captions

After any image or video generation, click "Generate captions" to get post-ready captions.

You'll get 3 styles for each of the 4 platforms:

Each platform offers witty / aesthetic / engaging tones. Pick the one that fits your brand voice.

06 Sharing to gallery

Help inspire other creators — and earn bonus credits.

  1. On any character / image / video, click "🌟 Share to gallery"
  2. Submission goes to admin review (usually approved within 24h)
  3. Approved submissions appear in the public gallery on the homepage and inside the studio
  4. You receive +20 credits for approved characters and images, or +100 credits for approved videos

You can also share content from the History tab anytime — click any item and the share button appears in the modal.

07 Prompt tips

Be specific, not poetic

The model understands concrete nouns better than abstract feelings.

Less effective:
vibing in a dreamy place
More effective:
sitting on a balcony in Lisbon, white linen dress, sunset behind, hand brushing her hair back

Lock the lighting

Lighting is the #1 differentiator between amateur and professional-looking content.

Set a camera angle

Mention the camera explicitly — close-up, medium shot, full body, low angle, over-the-shoulder, etc.

Don't redescribe the character

The system preserves your character's face and hair. Repeating "long black hair, slim figure" wastes prompt tokens and can confuse the model.

Use references for style

Style words alone ("cinematic", "editorial") are weak. Reference images speak louder than 50 adjectives.

🚀 Pro tip — iterate fast
Image generation costs 100 credits. Generate 3-5 variations of the same scene with slight prompt tweaks (different lighting, pose, angle) — pick the best one.