Higgsfield Faceswap and Recast

Higgsfield face swap

Face Swap is an app inside Higgsfield (Face Swap page). You upload a target photo or video, then a clear photo of the face to insert.

The app’s own answers set the limits:

• Free accounts get “5 free image face swaps per day”. “Video face swaps are available on paid plans.”
• One face per job. The tool “does not support swapping multiple faces within a single operation”.
• A swap takes “from 30 seconds up to 2 minutes”, by plan and queue.

On credits, the pricing table puts an image swap at about 2. It shows no rate for video swaps; the cost appears on the Generate button before you confirm. Across Higgsfield, eight of 52 public sources say generations run slow or fail and still cost credits (Reddit, YouTube).

Character Swap 2.0 is the image version of a bigger change. It moves your whole character into a target scene and re-renders it to match the scene’s “style, lighting, and perspective” (Character Swap page). It costs about 2 credits an image too.

What is Higgsfield Recast?

Recast replaces the person in a video with a character from one image. Higgsfield now titles the app page Character Swap, and says its technology “aims to integrate the entire persona including body and motion consistency” (Recast page). The page says you start free with daily credits.

That is the difference that matters for ads. A face swap keeps the original actor’s body, hands and clothes. Recast swaps all of it, so the new character performs the original take.

FULL Body Character Swap on Video! (Face, Gesture & Voice) - Higgsfield AI RecastWaleed Arfeen

Higgsfield’s newer tool for the same job is Genjutsu. Its Motion Transfer “keeps the motion, camera and timing while rebuilding the scene from your references”, on a source clip of 4 to 30 seconds (Genjutsu).

More on motion tools: Higgsfield motion control.

Lip sync on Higgsfield

Lip sync lives in Lipsync Studio, under Video (lipsync article). The steps and costs are in lip sync on Higgsfield.

How do I use lipsync, voiceover, and aspect ratios?

Voice generation, voice change, and translation live in Audio, while making a character talk (lip sync) lives in Lipsync Studio, which offers several models for image-to-video and video-to-video talking clips. Some video models also generate audio together with video in a single pass. Aspect ratio is set per generation inside each tool: common options are 9:16 for vertical, 16:9 for horizontal, and 1:1 for square, and not every model supports every ratio.

You upload a face or a clip, type the line, pick a model and a scene template such as Selfie, Podcast or Car Talking.

Two kinds of model sit there. Image-to-video models make a still face talk. Video-to-video models re-sync the mouth on footage you already have, which is how you re-voice an ad that already performs.

Speak sits on Plus and Ultra only. Plus costs $49 a month and Ultra $129 a month.

Voices come from Audio. You can clone one from a recording or an MP3 or WAV file, and translate a script into 18 languages. Higgsfield’s rule: “Only clone a voice that is your own or one you have the person’s permission to use.”

Can a swapped face run in an ad?

Yes, when every person whose face or voice appears has consented and the ad does not pass the result off as real. Higgsfield’s terms make each of those your duty.

The Face Swap page says commercial use comes with paid plans (its FAQ says “all paid Pro plans”), which today means Starter, Plus and Ultra. Its FAQ also says Higgsfield has “strict policies against any form of non-consensual or harmful content generation”. The terms say what that means (Terms of Use, 26 July 2026):

• Section 5.3: uploading a face or voice warrants “all consents, releases, and permissions required under applicable law from each individual whose face or voice is included”.
• Section 4.2(b) and (c): the same for any “name, likeness, voice”, and your uploads must not infringe “copyright, trademark, publicity rights”.
• Section 5.1(ii): no output that is “deceptive” or “depicts them without permission or legal justification”.
• Section 5.2: you may not “impersonate any person” (viii) or present output as “human-generated when it is not” (xi).
• Section 5.5: disclose AI “Where required by applicable law”, and keep the provenance marks Higgsfield may embed (6.4).

Cinema Studio’s “content protection system detects real faces and protected IP” and blocks the generation (Cinema Studio article). The swap apps take a real face by design, so the consent check falls to you.

Laws add a second layer. California Civil Code 3344 bars using another’s likeness in advertising “without that person’s prior consent”, and New York’s Civil Rights Law section 50 asks for “written consent” (AI UGC legal guide).

In the EU, a deep fake is content that “resembles existing persons” and “would falsely appear” authentic. Since 2 August 2026, Article 50(4) of the AI Act has made the deployer disclose it (AI disclosure rules).

Platforms ask too. TikTok’s ad policy counts making a person “say something they didn’t actually say, using AI voice-cloning” as a significant edit that needs its AI label (TikTok AI label). YouTube asks you to disclose content that “makes a real person appear to say or do something they didn’t do” (YouTube AI disclosure). Meta’s rules: Meta AI disclosure.

What you need before a swapped face runs as an ad

1. A signed release from the person whose face you insert, naming advertising use (terms 5.3; Civil Code 3344; Civil Rights Law 50).
2. A release from anyone whose face or voice stays in the target footage (5.3, “each individual”).
3. The rights to the target footage itself, music and logos included (4.2(c)).
4. A script where nobody claims to be a customer they are not (5.1(ii), 5.2(viii) and (xi)).
5. The platform’s AI label wherever its rules call for one, and the EU deep-fake disclosure under Article 50(4) (TikTok, YouTube, AI Act).
6. The exported file untouched by any tool that strips its provenance data (5.5 and 6.4).

Training is the last clause to weigh. Under 4.4, “Your Content, Inputs, and Outputs may be used by Company to train” its models, outside an Enterprise Agreement. A model’s face uploaded for a swap falls under that clause. Three of 52 user sources object to Higgsfield’s licence and training terms (YouTube, Reddit).

For one recurring face you own outright, a trained character is the tidier route: Higgsfield Soul and Soul ID. Every clause by plan and use: Higgsfield in the rights matrix.

Questions about Higgsfield face swap

Is Higgsfield face swap free?

For images, yes: 5 free image face swaps a day. Video face swaps need a paid plan, from $15 a month.

Can Higgsfield swap faces in a video?

Yes. Face Swap takes a video as the target on paid plans, and Recast replaces the whole person in a video.

Can I swap two faces at once?

No. The Face Swap app handles one face per operation.

What is the difference between Face Swap and Recast?

Face Swap changes the face and keeps the original body. Recast swaps the whole character, body and movement included.

Does Higgsfield have lip sync?

Yes, in Lipsync Studio under Video. It holds image-to-video models such as Speak 2.0 and video-to-video models such as Kling Lipsync and Sync Lipsync 3.

Every Higgsfield guide, sorted by job, is in the Higgsfield hub →