Loading
First subscription · First month 35% off / first year 25% off
Best starter model for faces and people
Official resolutions, durations, inputs and audio support in one table.
| Brand | MiniMax |
| Type | Video generation (Image / Text -> Video) |
| Max resolution | 512p · 768p · 1080p |
| Max duration | 6s · 10s (cap 10s) |
| Native audio | Not supported |
| Inputs supported | image · text |
| Editorial rating | 4.0 / 5.0 |
| Credit range | 12 – 57 credits / job |
Hailuo 02 from MiniMax is a portrait-first video model. It handles facial expressions, head turns, and upper-body gestures better than other starter options. Upload a headshot or half-body photo, describe a simple action like "person nods and smiles," and the output keeps the face consistent across frames. Standard version outputs at 768p; the Pro upgrade pushes to 1080p at higher credit cost. Not the right pick for landscapes or product shots — this model is built for people.
Face identity stays consistent across frames — other starter models often distort faces mid-clip
Smiling, nodding, blinking — subtle facial movements render naturally without uncanny-valley artifacts
Gesturing, waving, turning — works if the action stays above the waist
Low-res family photos still produce usable results — the model fills in facial detail surprisingly well
Copy these prompts to use directly, or tweak them to fit your needs.
person slowly turns head to face camera, slight smile, soft natural light
Simple head turn with smile
person nods gently while maintaining eye contact, warm indoor lighting
Gentle nod with steady gaze
person takes a deep breath, shoulders rise and fall, relaxed expression
Breathing motion with shoulder movement
Headshots and half-body shots work best. Make sure the face is well-lit and not occluded. PNG or JPG, under 4MB.
Keep it to one clear action: "turns head and smiles" or "nods slowly." Combining multiple actions reduces face consistency.
Standard: free, 768p, 12 credits for 6s. Pro: 1080p, 57 credits for 6s. Try Standard first to validate the motion, then upgrade to Pro for the final render.
One person doing one thing produces the cleanest output. Adding a second person or second action drops quality noticeably.
person slowly nods and smiles warmly
This model reads facial descriptions well. "Eyes narrow with amusement" gives better results than just "smiles."
eyes light up, corners of mouth lift gently
"Slowly," "gently," "softly" produce smoother motion than fast or aggressive verbs.
Standard: 12 credits for 6s at 512p (daily free credits won't fully cover it — top up or save two days). Pro: 57 credits for 6s at 1080p. Test with Standard, render with Pro.
Grok is faster and better for illustrated content. Hailuo 02 wins on face consistency and portrait realism.
Seedance handles full-body dance moves better. Hailuo 02 handles faces and upper-body emotion better. Pick based on whether the focus is body or face.
Hailuo 02 is best for Headshot and portrait animation. Its strongest advantage is Best face consistency among starter models — identity holds across all frames, while the key limitation is Standard version caps at 768p — upgrade to Pro for sharper output. For lower iteration cost, validate prompts on free models first, then switch to this model for final renders.
Grok Imagine vs Hailuo 02
Grok Imagine is faster and ships with audio for ideation; Hailuo 02 is more stable on portraits and natural human motion when you need character fidelity.
Hailuo 02 vs Hailuo 2.3
Upgrade to Hailuo 2.3 when you need premium 1080p output and native audio. Stay on Hailuo 02 when iteration cost matters more than peak fidelity.
HappyHorse 1.0 vs Hailuo 02
HappyHorse offers native audio plus 1080p at the free tier; Hailuo 02 is better for tight portrait motion when audio is not needed yet.
Jump straight into the generator workspace, upload one photo, and test your first AI video now.