P-Video-Animate Brings AI Motion Transfer to Tellers

One week after we announced our partnership with Pruna, they have already cooked a new model: P-Video-Animate is now available for AI video creation on Tellers.

P-Video-Animate does something very practical. Give it a reference image and a reference video, and it generates a new video that keeps the motion, timing, camera movement, and scene structure of the video while using the style and subject identity of the image.

Creators are increasingly asking for more control over AI video — not just more generation, but exact motion, timing, and visual identity. P-Video-Animate is built for that demand.

What P-Video-Animate Adds to AI Video Generation

Most AI video generation models start from text, an image, or both. That is powerful, but it can be hard to control exact acting, pacing, and camera movement from a prompt alone.

P-Video-Animate changes the control surface:

• Reference image: the subject, character, product, or style you want to animate
• Reference video: the motion, timing, acting, camera move, and scene structure to preserve
• Generated output: a new clip that follows the reference motion with the new visual identity

For creators, this is useful because motion is often the hard part. A winning ad gesture, meme format, camera push, or short performance can become a reusable motion template instead of a one-off clip.

Where P-Video-Animate Beats Other Model Types

P-Video-Animate is not replacing every video model. It is a specialised video generation model for when acting it out is easier than prompting.

• Versus text-to-video: better when exact timing and movement matter more than open-ended generation.
• Versus standard image-to-video: better when you want to preserve a specific performance, camera move, or scene rhythm.
• Versus heavier video-to-video models: more cost-efficient for fast variations where the source motion is already good.

That makes it especially strong for:

• UGC ad variations with different creators, customers, or personas
• Viral meme remixes using the same movement template
• Character or avatar animation from a still image
• Game cinematic and concept variations
• Product shots where the camera motion is already proven

Launch Discount on Tellers Tokens

Buying Tellers Tokens in bulk works out to $0.05 per token, which makes it easy to read P-Video-Animate pricing in dollars too.

P-Video-Animate is already one of the most attractive models at its regular price:

• 720p regular price: 0.9 Tellers Tokens per second (~$0.05/s)
• 1080p regular price: 1.8 Tellers Tokens per second (~$0.09/s)

Through May 30, 2026, Tellers is running a 70% launch discount — starting at about $0.01 per second at 720p:

• 720p launch price: 0.27 Tellers Tokens per second (~$0.01/s)
• 1080p launch price: 0.54 Tellers Tokens per second (~$0.03/s)

So a 5-second 720p animation costs 1.35 Tellers Tokens ($0.07) during the launch window instead of 4.5 Tellers Tokens ($0.23). A 5-second 1080p animation costs 2.7 Tellers Tokens ($0.14) instead of 9 Tellers Tokens ($0.45).

For a model built around rapid variation, that discount matters. You can test more characters, more product angles, more meme formats, and more ad concepts before deciding what deserves a heavier generation or final edit.

Why This Fits Tellers

Tellers works best when models are treated as creative building blocks, not isolated tools. The agent can combine generation, editing, timeline assembly, and model selection inside one workflow.

P-Video-Animate gives the agent a new kind of building block: reusable motion. Instead of asking a model to invent every shot from scratch, you can anchor the motion in footage that already works and use AI to explore new visual directions.

P-Video-Animate is available now on Tellers. If you want to test motion-transfer generation while the 70% launch discount is active, open Tellers and start with a short reference clip.