The distinction from a starting image is worth being pedantic about, because they are different slots on the form. A starting image is the first frame — it appears, exactly as supplied. A reference never appears; it is a description made of pixels, and the model is free to show that subject from another angle, at another size, in different light.
Models publish how many they will hold at once, and that ceiling is real. Beyond it, extra images are dropped rather than averaged in. Within it, variety helps: three angles of one product teach the model more than three near-identical crops.
Everything in the frame is a reference, not just the part you meant. A subject photographed against a red sofa often brings the red sofa along, which is why cutouts and clean plates give more predictable results than snapshots.
In practice
- Different angles beat duplicates — you are describing a thing in the round.
- Crop out what you do not want to inherit; the model cannot tell which part you meant.
- Reference slots are separate from the prompt: the prompt still decides what happens in the shot.
Models that take reference images
Models with a reference slot, and how many images each will hold at once. 38 of the 296 models in the Versely catalog qualify.
| Model | Provider | Type |
|---|---|---|
| Nano Banana 2 | Image | |
| HunyuanImage 3.0 Instruct Edit | Hunyuan | Image |
| Kling Image O1 | Kling | Image |
| Flux 2 Klein 9B Base | Flux | Image |
| Qwen Image Edit 2511 | Qwen | Image |
| Flux 2 Klein 4B Base Edit | Flux | Image |
| Flux Kontext | Flux | Image |
| Qwen Image | Qwen | Image |
Browse all 13 spec pages for full settings, resolutions and credit costs.
The mistake to avoid
Using a heavily filtered or stylised photo as a product reference. The model learns the filter along with the product and applies both.
Where you will run into it
- Hivewell Raw Honey — AI-animated product ad — a Pixar-style Hivewell honey jar character tells raw honey's story across 10 cinematic scenes: ancient Egyptian and Greek healers, Indian Ayurveda, a corn-syrup factory villain turn, then the Hivewell hero return and CTA. 40s vertical reel.
- 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
Reference-to-video
Reference-to-video builds a clip around subjects supplied as separate reference images, rather than starting from one fixed opening frame.
Character consistency
Character consistency is whether the same person, mascot or product still looks like itself across separate generations.
Image-to-video
Image-to-video animates a still you supply: the picture becomes the opening frame, and the prompt describes only what happens next.
First-last frame
First-last frame generation takes two stills — where the clip starts and where it ends — and generates the motion that gets from one to the other.
Seed
A seed is the number that decides the random starting noise for a generation, so the same seed with the same settings reproduces the same output.
The all-in-one AI studio for creators. 60+ models for video, image, voice, music and lipsync in a single app.