Your character, any scene
Add a photo of a person, a character or a product, describe the scene, and get a clip with sound that keeps them recognisable.
Pay per render · exact credit total shown first · credits never expire
Input
A reference photo and a prompt
Length
5–15s
Quality
720p · 1080p
Audio
Generated with video
Made with P-Video 2 Pro Reference
The clip is 5 seconds at 720p from P-Video 2 Pro Reference, with the sound the model generated. Tap to hear it. The prompt under each clip is the literal prompt that made it.
Prompt: The woman from the reference walks through a sunlit flower market holding a coffee, smiles at a vendor and turns toward the camera, smooth tracking shot, warm morning light. Sound: market chatter, birds, light acoustic music.
Make one with P-Video 2 Pro ReferenceP-Video 2 Pro Reference is the reference-to-video version of P-Video 2 Pro. Instead of animating your photo as the first frame, it builds a new scene around what is in it: the same face, outfit, mascot or product in a different place, pose and camera move. Add a character reference (a photo or one of your saved AI influencers) or a first frame, write the scene, and pick 5 to 15 seconds. It costs 2 credits per second at 720p and 3 at 1080p.
Keep the same person, mascot or AI influencer across many clips without re-shooting anything.
Put the product in your photo on a kitchen counter, a beach towel or a studio set, with motion and sound.
Unlike a first frame, the reference does not have to be the opening shot: the scene, pose and camera are yours to describe.
Model comparison
All 14 options are visible in one TalkPix studio, so you can compare capability and price before generating.
| Model | Use cases | Length | Quality | From | Open |
|---|---|---|---|---|---|
| P-Video 2 Pro | Sharp, natural motion with generated sound, ready in about half a minute. | 5–15s | 480p, 720p | 1 cr/s | P-Video 2 Pro guide |
| P-Video 2 Pro ReferenceCurrent | P-Video 2 Pro built around your reference: the person, character or product in your photo carried into a new scene with sound. | 5–15s | 720p, 1080p | 2 cr/s | P-Video 2 Pro Reference guide |
| Seedance 1.0 Pro Fast | ByteDance's fast Seedance 1.0 Pro: the lowest price for a 720p or 1080p clip. Picture only, no sound. | 3–12s | 720p, 1080p | 1 cr/s | Seedance 1.0 Pro Fast guide |
| Seedance 1.5 Pro | Fast, affordable clips with generated sound. | 4–12s | 480p, 720p, 1080p | 1 cr/s | Seedance 1.5 Pro guide |
| Seedance 2.0 Mini | Affordable native-audio motion for drafts and frequent posting. | 4–15s | 480p, 720p | 2 cr/s | Seedance 2.0 Mini guide |
| Seedance 2.0 | Advanced motion, prompt accuracy, native sound, and output up to 4K. | 4–15s | 480p, 720p, 1080p, 4K | 4 cr/s | Seedance 2.0 guide |
| Seedance 2.5 | Best consistency for complex actions, identity, and longer final clips. | 4–30s | 480p, 720p | 5 cr/s | Seedance 2.5 guide |
| Hailuo 2.3 | MiniMax Hailuo 2.3 animates your photo with lifelike motion and expressions, in 6 or 10 second clips at a budget price. | 6–10s | 720p | 2 cr/s | Hailuo 2.3 guide |
| MiniMax H3 | Reference-driven video up to 2K at the lowest 2K credit rate. Includes generated sound. | 4–15s | 720p, 2K | 4 cr/s | MiniMax H3 guide |
| Veo 3.1 Lite | Google's Veo 3.1 Lite: Veo's dialogue and sound at the lowest price, at 720p or 1080p for the same rate. | 4–8s | 720p, 1080p | 2 cr/s | Veo 3.1 Lite guide |
| Veo 3.1 Fast | Google's Veo 3.1 Fast: spoken lines, sound effects and ambience generated with the picture, at 720p or 1080p for the same rate. | 4–8s | 720p, 1080p | 4 cr/s | Veo 3.1 Fast guide |
| Veo 3.1 | Google's full-quality Veo 3.1: its most realistic picture, lip-synced dialogue and rich sound in one take, at 720p or 1080p for the same rate. | 4–8s | 720p, 1080p | 13 cr/s | Veo 3.1 guide |
| Kling 3.0 | Premium motion, consistency, multi-shot quality, and native sound. | 3–15s | 720p, 1080p, 4K | 6 cr/s | Kling 3.0 guide |
| Kling 4.0 | Kling's newest model: cinematic motion and native sound in one take of up to 30 seconds, from text or a first and last frame. | 3–30s | 720p, 1080p | 6 cr/s | Kling 4.0 guide |
How it works
The model reads the reference before it writes the scene, so faces, clothing and product details carry over.
Ambience, effects and music are generated with the video.
2 credits per second at 720p, 3 at 1080p: 10 credits for a 5-second clip at 720p, shown before the render starts.
Prompt example
The woman from the reference walks through a sunlit flower market holding a coffee, smiles at a vendor and turns toward the camera, smooth tracking shot, warm morning light. Sound: market chatter, birds, light acoustic music.
Try this modelTransparent credits
720p
2 credits per second
5 seconds = 10 credits
1080p
3 credits per second
5 seconds = 15 credits
Model line-up and credit rates last updated October 25, 2026.
FAQ
It is the reference-to-video model of the P-Video 2 Pro family: you give it a picture of the subject and a prompt, and it renders a new 5 to 15 second scene with sound that keeps the subject recognisable.
P-Video 2 Pro starts from text alone. P-Video 2 Pro Reference needs a picture (a character reference or a first frame) and builds the clip around it, which is what you want when the same person or product must appear.
2 credits per second at 720p and 3 at 1080p. A 5-second 720p clip costs 10 credits, a 15-second one 30. The exact total is shown before anything renders.
A clear photo of a person, pet, character or product, or one of your saved AI influencers. You can also give a first frame, and optionally a last frame, instead of a character reference.
Yes. Describe the ambience, effects or music in the prompt and they are generated with the picture.
Open the studio with this model selected. Review the exact credit total before any render starts.
Open P-Video 2 Pro Reference