PrunaAI P-Video: the still is the contract (1s, 2s, 3s, 1080p, 2cr)

PrunaAI P-Video is image-to-video. If the label is wrong, every second is waste. It is also text-to-video on the same slug. The description: fast video generation with built-in draft mode for rapid creative iteration. Text-to-video, image-to-video, and audio-to-video in a single endpoint. Generates 5s 720p video in ~10 seconds. requires_image is false. Max output 1080p. Credits 2. Durations 1s through 20s. The still is the contract when you attach one. If you skip the still, you are on the text path and the "first frame" never existed.

That dual filing is the trap. The image-to-video tool is the right door when you own a plate. The AI video generator is the text door. Same model, two contracts. Pick one per generate.

One endpoint, two jobs

Categories: text-to-video and image-to-video. Features also list audio-to-video, draft mode, prompt upsampling, lip_sync, talking_avatars, product_animation. The audio field on the snapshot is empty — do not treat this as a native-audio guarantee the way Kling V2.6 Pro is (audio: true). If you needed a baked-in mix, read the take. If you needed a mute draft, draft mode is the listed tool.

When the job is image-to-video, upload the approved still and prompt only motion. The still is frame one. A prompt that restages the product on a beach will fight the plate. When the job is text-to-video, do not upload a leftover PNG "for style." That leftover becomes a first-frame argument you did not mean to make.

Draft mode is how 2 credits stay 2 credits

2 credits is the catalog price. The description's speed claim is a 5s 720p clip in about 10 seconds. Quality settings are 720p and 1080p. Max is 1080p. Draft at 720p, 1s or 2s. Promote to 1080p and a longer take only after the motion is the motion you wanted.

Aspects: 16:9, 9:16, 4:3, 3:4, 3:2, 2:3, 1:1. Frame rates: 24fps and 48fps. Use 1s to see whether the still holds. 20s is the ceiling, not the default. A 20s generate at 2 credits is still a 20s mistake if the label was wrong.

There is no published PrunaAI brand hub. Stay on the model slug, best image-to-video, and best text-to-video.

1s is legal; 20s is the ceiling

Durations include 1s, 2s, 3s, then every second through 20s. Happy Horse starts at 3s. Wan 2.7 starts at 2s. Wan 2.6 Flash is only 5 / 10 / 15. P-Video is the row that will sell you a 1-second twitch and a 20-second hold on the same slug.

That range is why the label matters. Twenty seconds of a still you have not approved is twenty seconds of waste, even at 2 credits. Lock the plate on a stills row. Then pick 1s. Then pick 5s. Then, maybe, 20s.

If you needed native dialogue as a guaranteed field, this is not Kling and not Happy Horse. If you needed a first-frame product push at the lowest video credit in this shard, this is the 2-credit, 1080p, 1–20s row.

FAQ

Do I have to upload an image?

No. requires_image is false. The row runs text-to-video without a file. If your job is "animate this still," upload it anyway — otherwise you are not on the image-to-video contract.

What does draft mode do?

It is the listed path for rapid iteration: 5s 720p in about 10 seconds, 2 credits. Inspect motion. Then spend a longer 1080p take.

Can I make a 20-second clip?

Yes. Durations go to 20s. That is the ceiling. 1s, 2s, and 3s are the honest tests.

Is audio native on P-Video?

The audio field is unset. Features mention audio-to-video and lip_sync, but this is not a row with audio: true. Do not assume a soundtrack policy. Listen to the file or pick a native-audio slug when speech is the product.