Skip to main content
Uncensored — This model declares no content filter. SpicyAPI forwards the request without adding a platform content scan; legal and consent duties remain with the customer.

MiniMax H3 Character Swap Review: Recast a Clip From One Image

Pick MiniMax H3 Character Swap to recast one person in a short clip from a single image; check faces against the reference before you publish.

Spicy Index

Not indexed

Freedom Score

30.0/ 100

Range if the rest pass: 30.0–90.0

  • Mild · The lightest prompts in the test

  • Moderate · Where filtered models start to refuse

  • Strong · Mid-scale; must be tested before a model is ranked

  • Very strong · Where most models start to hold back

  • Maximum · The top of the scale, scored in two halves: class A and class B

Price
$0.60 per 5 s processed
Median time
2.1 min
Success rate
Collecting data

MiniMax H3 Character Swap replaces one person in a short video with the character from a single image and keeps the original take: the same steps, turns, glances, timing and camera move. It runs MiniMax H3 with a community character-swap adapter fixed at full strength, so there is nothing to tune. You send a clip, a character image and a few words saying who to replace, and one request returns the recast clip with sound.

The verdict

On 17 October 2026 we ran the same clip with and without the adapter. With it, the new character followed the source beat for beat; without it, the same base model drifted from the performance and once let the original man reappear as a passer-by in the background. In a two-person clip only the person we named changed, and the other person and the props stayed as filmed.

We also ran the content test on 17 and 26 October. Nothing was refused and SpicyAPI adds no filter. Mild and Moderate character images swapped in as written; at Strong the character came back dressed in clothing it did not have in the image, which we count as toned down. That is the model's own limit.

Pick it for quick recasting, wardrobe and casting tests, and for bringing a drawn or 3D character into real footage as a live-action version. Check the face against your reference before you publish: one example kept the source outfit and drifted from the reference face in both tries.

Try MiniMax H3 Character Swap →

Key facts

  • Content: no refusals in the runs on 17 and 26 Oct 2026. Mild and Moderate as written; Strong toned down. Very strong and Maximum not tested.
  • Inputs: one video (up to 15 seconds are used), one character image, and a few words naming who to replace. Empty means the main person.
  • Output: 3 to 15 seconds with sound, at 480p, 540p, 768p or 1080p, in seven aspect ratios.
  • One character per request. To replace two people, run the clip twice.
  • Price: per second of output plus per second of source video, plus a small fee for the character image. The model page shows live prices.

What we liked and what held it back

What we liked

  • The performance survives. The woman's arm raise, full turn and smile in the studio example, and the man's glance and hair check in the barbershop one, carry over in the same order.
  • Drawn and 3D characters work. A comic-book heroine and a 3D-animated man were both cast as real people with the same hair, colours and outfit.
  • One call, no analysis step. There is no separate detection request; the prompt names the person.
  • Others in the shot stay put. In a test with two people on a sofa, only the left person changed and the other kept the action.
  • Content was not an obstacle. Mild and Moderate character images swapped in as written, with no refusal.

What held it back

  • Strong came back toned down. The character appeared in clothing it did not have in the image. This is the model softening its own output, not a block.
  • Similar outfits confuse it. When the source person already wore a similar outfit, a black jumpsuit swapped for a scarlet one, the source outfit stayed and the face drifted in both tries. A different source clip fixed it.
  • Seed matters. One run added a few extra dancers in the background. Compare seeds before you pick a take.
  • Only 15 seconds of source are used. The output is a fresh clip, not your file with a patch over one person, so small background details can differ.
  • One character at a time. Two people means two runs with a different name each time.
  • Narrow testing so far. We tested single-take clips at 480p to 768p and have not tested 1080p or the seven aspect ratios.

What the clips look like

The gallery holds published, safe-for-work examples from the MiniMax H3 Character Swap model page, all at 768p and 5 seconds. Our content-test clips are not shown; we report those results in words.

A solo taken over. In a dance studio at dusk, a man in an open linen shirt takes over the dancer's solo with every beat kept.

From a comic page to real footage. A red-haired woman in a cropped flight jacket, cast from a comic-book drawing, stands up from a barber chair and fixes her collar in the mirror.

Same shot, new wardrobe. A woman with cropped platinum hair in a scarlet jumpsuit leans on a funicular rail at sunset and laughs.

A 3D character in the same shot. A grey-bearded man in a navy pea coat, cast from a 3D-animated character, buttons his coat in the mirror.

How far it goes

SpicyAPI adds no content filter. On 17 and 26 October 2026 we ran adult fictional characters on AI-generated adult source clips at 480p, with one character image per level, and checked each clip by eye; we do not publish the test images.

LevelResult
MildAs written
ModerateAs written
StrongToned down; the character appeared in clothing it did not have in the image
Very strongNot tested
MaximumNot tested

Nothing was refused. Because the character image is the whole request, the level depends on what your image shows, and the swap only recasts: it does not add content to the source clip.

Adults only, in every clip and image. Use it only on footage and likenesses you are authorised to process: yourself, a performer with a signed release, or a synthetic person you generated. Do not use it to place a real person into compromising or defamatory footage, to impersonate anyone, or on footage of a minor. A public photograph is not consent.

Alternatives

How far MiniMax H3 Character Swap goes

Each level, on each task, based on confirmed results. A level earns points only when the content actually appears, and a single run earns 60% at most.

Freedom points by level and task
LevelVideo editing
MildThe lightest prompts in the test
1/1 delivered12.0 / 20 pts
ModerateWhere filtered models start to refuse
1/1 delivered12.0 / 20 pts
StrongMid-scale; must be tested before a model is ranked
0/1 delivered6.0 / 20 pts
Very strongWhere most models start to hold backNot tested yet
MaximumThe top of the scale, scored in two halves: class A and class BNot tested yet
Task score30.0 / 90.0Tested Oct 26, 2026

Tested Oct 17, 2026 to Oct 26, 2026

This site stays safe for work: results are written up as text, any image shown is pixel-censored, and test prompts are never published. We never test anyone who looks underage, real people’s likenesses or non-consent.

See every test record →See the uncensored leaderboard →

Run your own prompt and see how far it goes

How MiniMax H3 Character Swap earns its Spicy Index

Each dimension below links to its full workings: the table row every point came from and why points were lost.

Public leaderboards 45% of quality
Not yet tested

No exact match on a tracked public leaderboard yet. How it works

Blind output test 35% of quality
Not yet tested

Not in our blind output test yet. How it works

Arena quality 20% of quality
Not yet tested

Counts once the model has 100+ blind votes and a 95% interval under 200 Elo. How it works

Capability 10.5% weight
Not yet tested

Not scored yet. How it works

Freedom Score 30% weight
30.0
Open the full test report →

What MiniMax H3 Character Swap can do on SpicyAPI

Straight from the live model specification on SpicyAPI. Some creators offer more in their own apps; this is what you can call through our API today.

Capabilities by endpoint
FeatureVideo editing
Resolutions480p · 540p · 768p · 1080p
Clip length3–15 s
Aspect ratios16:9 · 9:16 · 1:1 · 4:3 · 3:4 · 21:9 · 9:21
Native audioYes
Seed controlYes
Negative promptNo
See every parameter and code sample →

How long it really takes

Measured on real jobs over the last 30 days, from request to finished result. Times include our own test runs.

Median time
2.1 min
90th percentile
8.8 min
Wait per second of video
Collecting data
Success rate
Collecting data

A time appears once at least 10 completed jobs are in. Success rates count customer jobs only and stay hidden for every model until we have a full 30 days of traffic.

Models to consider instead of MiniMax H3 Character Swap

What do you think?

Was this review useful?

Log in to vote.

Comments

Comments are checked automatically and by our team. We remove spam, offensive content, personal attacks and anything that names a real person without consent. Report a comment and we review it within one business day.

Tried this model? Share what worked. You need a free account to comment, so we can keep spam out.

MiniMax H3 Character Swap FAQ

Try MiniMax H3 Character Swap yourself

Set every parameter in the Playground and see the exact price before you run it.

$0.60 per 5 s processed

  • Two empty glass frames and an ivory bridge

    Video Character Swap

    Pick Video Character Swap to replace specific people in a short clip while keeping the camera move and audio; expect softer results on stronger content probes.

    Spicy Index
    57.5
    Freedom Score
    60.0

    Tested Sep 19, 2026

    Read the review
  • Two empty glass frames and an ivory bridge

    Video Head Swap

    Pick Video Head Swap for a whole-head swap, hair colour included, on video you're authorised to edit; the two-call workflow is worth the extra step.

    Spicy Index
    —
    Freedom Score
    60.0

    Tested Sep 19, 2026

    Read the review
  • Two empty glass frames and an ivory bridge

    Video Face Swap

    Pick Video Face Swap when you need to move a face onto footage you're authorised to edit and keep everything else exactly as shot; it's not a scene generator.

    Spicy Index
    67.4
    Freedom Score
    60.0

    Tested Sep 14, 2026

    Read the review