RefVid

Hailuo Subject Reference - RefVid

Turn your own photos into videos in minutes · Your first video is free, no watermark

  • Free first video
  • No watermark
  • Commercial use
  • Travel Vlog video of Mia, a woman with a copper-red bob
  • Product 360 video of AERO 1, a grey and orange running shoe
  • Example video of the Wallet Open and Close template
  • Floating Product video of AERO 1, a grey and orange running shoe
  • Bullet Time video of LUMEN, a serum bottle with a dropper
  • Water Splash video of LUMEN, a serum bottle with a dropper
  • Unboxing video of LUMEN, a serum bottle with a dropper
  • UGC Ad video of Mia and LUMEN

First video free, no cardGenerating confirms everyone pictured agreed. Terms

Hailuo subject reference keeps one character the same in a new clip, from photos of that person. MiniMax, the company behind Hailuo AI, first offered it as the S2V-01 model, which took one face photo; its newest video model, MiniMax H3, does it in reference mode with several images.

The box above is set to MiniMax H3: add up to 5 photos of your character or product, write the scene, and get a 5- to 15-second clip with sound at 768p or 2K.

Mia: 4 photos

Mia, reference photo 1: three-quarter view by a window
Mia, reference photo 2: front view
Mia, reference photo 3: profile
Mia, reference photo 4: full body
Video of Mia, a woman with a copper-red bob
MiniMax H3, 5 s, 768p
Mia is an invented person. MiniMax H3 in RefVid from her four photos, our test run on November 5, 2026: 5 seconds at 768p, back in 33 seconds, with street sound and her line "Hi, I'm Mia." Not edited.

What subject reference is

  • Subject reference is MiniMax's name for generating a video around a person you show it, instead of a person the model invents from words. MiniMax's S2V-01 API takes one reference image of a character's face in a subject_reference field, plus a prompt of up to 2,000 characters, according to MiniMax's API documentation.
  • MiniMax H3, MiniMax's newest video model and one of the models in the Hailuo AI app, does the same job in reference mode: up to 9 images, 3 video clips and 3 audio clips, at most 12 files, and a clip of 4 to 15 seconds at 768p or 2K with sound, according to MiniMax's video generation guide.

Hailuo reference character: S2V-01 vs MiniMax H3

  • S2V-01 reads one face photo, so the model has to guess every angle that photo does not show. H3 reads several photos of the same subject, plus video for motion and audio for a voice on MiniMax's API, which gives it more of the character to keep.
  • RefVid runs MiniMax H3, not S2V-01. It sends your saved photos as Image 1, Image 2 and so on, up to 5, with the sentence that tells H3 they show one subject, so you write only the scene.
S2V-01 subject referenceMiniMax H3 reference mode
Reference images1 face photoUp to 9 (5 in RefVid)
Video and audio referencesNoneUp to 3 of each, 15 s in total
SubjectsA person (character)People, products, places, style
OutputNot stated on the API page4–15 s, 768p or 2K, with sound
Sources: MiniMax's S2V-01 API reference and its MiniMax H3 video generation guide, read November 7, 2026.

Run it online in 3 steps

  • 1. Add up to 5 photos of your character in the box at the top of this page: front, three-quarter and profile, in even light. It is already set to MiniMax H3 at 768p.
  • 2. Write what happens in one or two sentences, and put any spoken line in quotes: H3 says it, with the sound of the scene. Settings changes the length (5 to 15 seconds), the shape (16:9, 9:16 or 1:1) and 2K.
  • 3. Press Generate. If you are not signed in, you sign in first and the clip starts right after. Our test clip came back in 33 seconds; a busy queue takes longer, and we email you when it is ready.

What it costs

  • Every clip is priced in credits per second, and the studio shows the total before anything runs. New accounts get free credits for a first clip, with no watermark. After that, single videos start at $2.99, one-time credit packs at $9.90 and monthly plans at $10; credits never expire, and a failed clip refunds its credits automatically.

More on AI video models

Reference to video →

Frequently asked questions

Is Hailuo the same as MiniMax?

Hailuo AI is MiniMax's video app, and MiniMax is the company. The app's models were Hailuo 01, Hailuo 02 and Hailuo 2.3, and MiniMax H3 is its newest; some sites call it Hailuo 3.

How many photos does Hailuo subject reference take?

S2V-01 takes one face photo. MiniMax H3 takes up to 9 images on MiniMax's API; RefVid sends up to 5, because fal.ai charges for every image after the fifth.

Can I use Hailuo subject reference for a product?

Not on S2V-01, which reads only a character's face. MiniMax H3 takes products, places and styles as references too, so the box above works for a product as well as a person.

Does the clip have sound?

Yes. MiniMax H3 makes every clip with sound, and speaks any line you write in quotes.

Is Hailuo subject reference free on RefVid?

New accounts get 180 free credits for a first clip, with no card and no watermark. After that, a single video costs from $2.99, a credit pack from $9.90 and a plan from $10 a month.

By All guides