P-Video 2: Text & Image to Video Online

Every value here is what VdoBloom's endpoint and Runware integration enforce for prunaai:p-video@2, not a vendor headline. The two settings people most often misread: Draft is a separate cheaper tier at either resolution (not a third resolution), and on Image to Video there is no aspect-ratio control because the output follows the shape of your photo.

The formula is rate x (seconds + 1/24), rounded up to a whole credit, because Runware bills the real clip length and every P-Video 2 clip is one frame longer than requested. VdoBloom shows the exact total before you generate and refunds it automatically if the job fails. The dollar figures use the Lite plan rate ($15 for 300 credits, $0.05 a credit); a one-time 300-credit pack is $19.50 and its credits never expire, while subscription credits reset each billing cycle.

All three sit in VdoBloom's video pickers and bill from the same credit balance. P-Video 2 against the original P-Video: twice the maximum length, native audio, a last-frame slot and a Draft tier, for a few credits more per clip (11 against 8 for 5 seconds at 720p). Against Seedance 2 Fast, ByteDance's quick Seedance 2 tier: P-Video 2 reaches 1080p and 20 seconds where Seedance 2 Fast stops at 720p and 15 seconds, and costs less at every matched setting in the table; Seedance 2 Fast adds the 21:9 frame, a 480p tier and the Seedance family's reference inputs.

P-Video 2 lives in the P-Video family of the model picker on two tabs. The price and the 1-20 second range are the same on both; what changes is the input and the aspect-ratio control.

On text to video, VdoBloom sends P-Video 2 the exact frame size from Pruna's P-Video 2 table for the ratio and resolution you choose. The sizes differ slightly from the original P-Video (16:9 at 720p is 1280x704, not 1280x720), so check them if the clip must match an edit timeline exactly. On image to video only a 720p or 1080p preset is sent and the frame follows your photo.

Because audio is always on, the model fills the soundtrack even when you do not ask, so describe the sound you want. Put a spoken line in quotes and say who says it: 'A barista slides a cup forward and says, "Oat flat white, extra hot."' Match the words to the length; a 5-second clip fits one short sentence. Describe the voice (calm, whispering, out of breath), tie sound effects to actions (boots on gravel, the kettle clicking off), name the room tone, and say what you do not want: 'no music, natural sound only' is often the difference between a usable clip and a random soundtrack.

Work in two passes. Iterate at 720p Draft on a 3-5 second length (4-7 credits a try), change one thing at a time, then switch Draft off for the final at 720p or 1080p and the full length. On image to video, spend the prompt on motion, camera and sound rather than re-describing the photo; two very different first and last frames need 6-10 seconds to look like motion instead of a blend.

It is not a fast model: each job queues at the provider and renders video and audio together, so expect minutes rather than seconds; the clip saves to My Creations and you can leave the page. Draft lowers the cost of testing more than it lowers the wait. There is no audio input on VdoBloom's P-Video 2: the soundtrack comes from your prompt only, so a clip driven by your own voice track needs a different model. 20 seconds is the ceiling per clip, there is no 4K, no 21:9 frame and no 48 fps option (only 24 fps is exposed). Image to video cannot be forced into a different shape than the photo. Prompts are screened before a generation starts, and photos of real people should be your own or used with consent. If a generation fails, the credits return to your balance automatically.

VdoBloom does not score models on subjective quality; the figures above are spec and price. A 7-credit Draft answers whether the look suits your shot.