Soul ID is the character-consistency layer of Higgsfield's native Soul models. Train it once on 20 or more photos of one person (up to 80), and the same face appears across every generation, regardless of style, pose, or lighting. Training takes a few minutes, after which you select the character when generating. Soul ID locks one identity from the photos you train it on; everything else in the shot follows your prompt.
What is Soul ID?
Soul ID is a trained identity layer: you give it photos of one person, it learns that identity, and applies that face to anything you generate, regardless of preset, lighting, angle, or prompt. Train once and reuse the identity.
A trained Soul ID works across the Soul family. For the Soul family overview, see How do I use Soul to generate images?
How does Soul ID compare to uploading a reference image per generation?
A reference image anchors a single generation but drifts across many separate ones. Soul ID trains a reusable identity once and holds it across every generation, style, and angle without re-uploading.
How do I train a Soul ID?
Open higgsfield.ai → Soul → Train new character.
- Prepare your photos. Upload 20 or more photos of the same person (up to 80 supported): high-quality and well-lit, with the face shown from different angles and expressions, no sunglasses or heavy shadows, no cropped faces. Include at least one full-height photo, and use recent photos for the most accurate likeness. Only upload photos of yourself, or of someone who has given you permission to use their likeness.
- Upload and start training. Training takes a few minutes.
- Name the character.
- Generate. Select the trained character in the Character tab when generating.
Which photos produce the best results?
What helps | What hurts |
|---|---|
Varied angles and expressions | Same pose repeated across all photos |
Clear, well-lit face | Heavy shadows or harsh lighting |
Recent photos (last 4 to 5 months) | Outdated photos if appearance has changed |
At least one full-height shot | Head shots only |
No obstructions | Sunglasses, hats, scarves covering the face |
Single subject per photo | Group shots |
Training quality determines output quality: a smaller set of clean, varied photos outperforms a large set of inconsistent ones.
Can I use the character outside the Soul models?
Yes, in Seedance video generation. The character carries over as an Element: your trained Soul characters appear in Elements automatically, so there's usually nothing extra to create. If you ever need to add one manually, generate an image of the character, open it in Assets, and select Create Element from the three-dot menu.
To use the character in Seedance, click @ Elements under the prompt field (or type @ in the prompt) and pick it from My Elements: the panel also lets you reference your uploads and past generations.
What are the limits of Soul ID?
Expect "clearly the same person" rather than a pixel-identical face across every generation. Weak photo sets produce weak identities, and extreme style shifts or unusual angles can introduce small drift even with a well-trained identity.
A Soul ID holds one person: for scenes with two or more consistent characters, use Elements. The trained identity lives inside the Higgsfield ecosystem: it's used within the Soul models and connected tools and is not exportable as a standalone file.
What is Soul ID best for?
- AI influencers and UGC creators — ongoing content across outfits, settings, and aesthetics with one consistent face. To invent the character first, use AI Influencer Studio: see How do I use AI Influencer?
- Fashion and clothing brands — virtual lookbooks, seasonal campaigns, and product showcases, pairing with Marketing Studio for ad production.
- Personal branding — entrepreneurs, coaches, and speakers generating images for social profiles and marketing.