Pixverse 5.6 Image to Video: the still is the contract (5s, 8s, 10s, 1080p, 75cr)
Pixverse 5.6 Image to Video is image-to-video. If the label is wrong, every second is waste.
Pixverse 5.6 Image to Video is image-to-video. If the label is wrong, every second is waste.
The catalog is blunt: latest Pixverse model for converting images to high-quality animated videos. It requires an image. Durations are 5s, 8s, and 10s. Max output is 1080p. Qualities: 360p, 540p, 720p, 1080p. Credits: 75 on the 1080p step. Catalog audio is true. There is no aspect-ratio field — the still's crop is the frame.
A text-to-video prompt on this row is a mislabel. You will pay 75 credits to animate a picture you never locked, or the call will fail for missing image_url. Either way the seconds are gone.
Crop the plate before you buy motion
Pixverse 5.6 Image to Video infers aspect from the source image. 9:16 in the prompt does nothing. Crop the still to the delivery ratio, approve the face and the pack, then describe only what moves: head turn, steam, a slow push, hair in wind.
The still is the contract because every frame is a child of that file. A muddy logo in the jpeg is a muddy logo for 5, 8, or 10 seconds. A wrong crop is a wrong video. Fix the plate on a stills row. Then animate.
If you do not have a plate, you are on Pixverse 5.6 Text to Video — a different card, still 5s / 8s / 10s and 1080p, but it invents the frame. Do not pretend this I2V row will invent one.
The launcher for the locked-still path is the AI image to video tool. The image-to-video glossary is the family name. PixVerse's roster is video-only; it will not save you from a bad still.
5s, 8s, 10s — one action each
The duration enum is three values. 1080p is the expensive step (75 credits; 720p is listed at 45, 360p and 540p at 35). Pick the second count that matches the gesture, then pick the quality the delivery actually needs. A 10-second 1080p pass of "she lives her life, cinematic, epic" is how 75 credits buy mush.
Prompt the delta:
- What in the photo moves.
- How far, how fast.
- What must not change (identity, wardrobe, product label).
Negative prompt is a real field here. Put "extra people, morphing face, new background" there, not in a paragraph that restates the photograph.
Audio is on this I2V row
Catalog audio is true, with a generate-audio switch. A kitchen still plus a mute prompt can still come back with sizzle and chatter. Write the sound that belongs to the plate — or write that you want quiet — the same way you would on a native-audio text-to-video model. You are paying 75 credits at 1080p either way.
Do not use this row as a talking-head inventor. If the job is a script on a face, that is lipsync. Pixverse 5.6 I2V animates a still you already believe.
FAQ
What happens if I skip the image?
You are not on this model. The row requires an image. Every second you spend prompting a scene from text belongs on the text-to-video sibling, or on a stills model first.
Why is 75 credits the number in the title?
That is the catalog 1080p step on Pixverse 5.6 Image to Video. Lower qualities are cheaper on the same card. The duration is still 5s, 8s, or 10s.
Can I set 16:9 in the prompt?
Set it in the still. This I2V path has no aspect-ratio field; it reads the image. Recrop, then generate.
Does the clip include a soundtrack?
Catalog audio is true. Treat sound as part of the brief. A silent prompt is how you get a track you did not choose on top of a plate you did.