P-Image-Try-On Brings AI Virtual Try-On to Tellers

Pruna’s P-Image-Try-On is now available on Tellers, bringing fast AI virtual try-on into the same agentic workflow creators already use for image generation, AI video creation, and AI production.

The model is designed for a simple but commercially important task: upload a photo of a person, add one or more garment reference images, and generate a new photo where the person wears those clothes.

The useful part is what stays unchanged. P-Image-Try-On keeps the same person, pose, and background while replacing the clothing. That makes it practical when you need to change the clothes of an actor, presenter, or AI character without rebuilding the whole image.

Fast Wardrobe Swaps for AI Production

In AI production, wardrobe changes often happen late.

The character is right. The pose works. The background fits the scene. Then someone asks for a different jacket, a more premium outfit, a market-specific style, or a cleaner brand color.

That is exactly where a specialized try-on model helps. Instead of sending every clothing tweak to a larger general image model and waiting roughly two minutes for a GPT Image 2 result, P-Image-Try-On gives the Tellers agent a fast wardrobe tool for the cases where the subject and scene should stay fixed.

• Around 1.9 seconds per item
• 0.45 Tellers Tokens per item
• Support for one or more garment reference images
• Person, pose, and background preservation

The point is not that P-Image-Try-On replaces every image model. Bigger models are also on Tellers and can always be called to the rescue if needed. The point is that many production edits are narrow: keep the actor, keep the frame, change the clothes, and keep moving.

What P-Image-Try-On Does

P-Image-Try-On takes two inputs:

1. A person photo: the actor, AI character, model, customer, creator, employee, or reference subject.
2. Garment reference images: one or more items to place on the person.

The output is a new image with the selected clothing applied to the person while preserving the visual context around them.

That makes P-Image-Try-On different from a generic image editing prompt. The model is specialized around garment transfer, so the key instruction is not “change this image” in the abstract. The key instruction is “put these clothes on this person while keeping everything else stable.”

Production Uses for Creators and Fashion Teams

The most obvious use case is e-commerce, but P-Image-Try-On is useful anywhere clothing variations need to be generated quickly.

AI production teams can use it to:

• Change an actor’s outfit between ad variants
• Restyle an AI character without regenerating the character
• Test wardrobe continuity before generating a full AI video
• Localize clothing choices for different markets
• Produce fast thumbnails, storyboards, and pitch frames

Fashion and retail teams can use the same workflow to:

• A/B test outfit variants before committing to a shoot
• Generate first-pass catalog visuals from existing product images
• Test creator and influencer concepts before production
• Build ad variations around the same model, setting, or pose

For teams already using Tellers, the main advantage is orchestration. The Tellers agent can use model calls as part of a broader workflow: generate or select references, produce try-on images, compare options, place selected visuals into a video, and adapt the result for different social formats.

Why This Fits the Pruna and Tellers Partnership

We integrated Pruna models because they match a product principle we care about: AI tools should be fast enough and affordable enough to stay inside the creative loop.

P-Image-Try-On follows that pattern. It is not a heavy, occasional operation reserved for a final asset. It is a model you can use while exploring, before deciding whether a shot needs a larger image model, a video model, or a manual art direction pass.

That changes the behavior of the user. Instead of asking, “Is this wardrobe tweak worth running?”, teams can ask better creative questions:

• Which garment works best on this model?
• Does this outfit read clearly in a vertical ad?
• Which market-specific styling feels most natural?
• Should we generate a full AI video from this look?
• Which variation is worth sending into production?

Fast try-on is valuable because it turns wardrobe swaps into an iterative creative surface.

P-Image-Try-On is available now on Tellers. To test fast virtual try-on with your own person photo and garment references, open Tellers and start a new generation.