A workflow stores the cast as records, not as adjectives in a prompt. You upload reference images once, they land under a key, and later scenes point at that key. The upload is not a generation; the key is what the director reads when it writes the next shot. Products, props and locations use the same object with a different type.
Character-consistency is the quality you hope for — does the face hold. A reference-image slot on a single generate is a per-job input that dies with that job. Brand-kit stores colours, fonts and a logo, which is a look, not a person. None of those are a named character you can call in scene four without re-attaching files.
The agent opens an Asset Builder card, or several at once for a full cast. Hosted URLs can skip the card. Assets belong to that workflow, not a global pool, which is why cloning a public recipe and swapping the asset map is how you recast it as yours.
In practice
- Register front, three-quarter and profile under one key rather than four nearly-identical characters.
- Crop out furniture and filters you do not want inherited; the model cannot tell which part of the still was the person.
- Point later scenes at the key instead of re-pasting the image into each prompt.
The mistake to avoid
Pasting the same reference into every scene prompt and never saving the asset. The later scene has nothing to look up, so identity is hoped for rather than addressed.
Go deeper
request_asset_upload registers a cast; it does not generate a scene
Reusable characters and products are one named setup job. Prompting a face every scene, or stuffing a logo into memory, spends generate credits on identity you already had.
Where you will run into it
- Plush Lipstick Influencer Review — Sara the influencer reviews the Plush Baby Pink Lipstick in 3 scenes: greeting, product review with features, and promo code CTA. Consistent voiceover narration throughout.
Related terms
Character consistency
Character consistency is whether the same person, mascot or product still looks like itself across separate generations.
Reference image
A reference image is a picture supplied alongside the prompt so the model can copy an identity, product or style from it, without that picture becoming a frame of the output.
Brand kit
A brand kit is the saved set of colours, fonts, logo, product images, tone of voice and default frame shape that generations are expected to obey without being reminded.
Saved workflow
A saved workflow is a reusable video recipe — scenes, assets, style and model stored so you can run it again with a fresh plot.
Text-to-video
Text-to-video is generation from a written prompt alone — you describe a shot, the model invents every frame of it, and no image or footage goes in.
The all-in-one AI studio for creators. 60+ models for video, image, voice, music and lipsync in a single app.