Kling O3 Pro Image to Video: the still is the contract (3s, 4s, 5s, 70cr)

Kling O3 Pro Image to Video is image-to-video. If the label is wrong, every second is waste.

The catalog: Kling O3 Pro image-to-video with advanced reasoning-enhanced generation and camera controls. Requires an image. Durations 3s through 15s. Credits 70 for the first 5 seconds, 14 for each extra second, ceiling 210. Catalog audio is true, with an audio-generation feature. There is no max-output-resolution chip on this card. The start frame has a floor: the schema wants an image at least 300×300, under 10MB, aspect between 0.4 and 2.5.

Reasoning does not invent a pack you refused to lock. It reasons over the plate you uploaded.

Start frame, then 3s, 4s, or 5s

Kling O3 Pro Image to Video takes image_url as the start. Crop that still to the delivery ratio before you upload. 3s, 4s, and 5s are the default-shaped windows (5s is the billed base). 6s–15s exist; they add 14 credits a second. Do not buy 15 seconds to discover the logo is illegal.

Prompt motion and camera, not a new scene. Kling's tip wants named camera movement, subject action, environment, motion direction and pacing. That language earns its keep here: "slow push-in, she turns to camera, steam drifts, no new wardrobe." Restating the photograph fights the start frame.

An optional end frame exists. If you supply one, you are describing the path between two approved stills, not "what happens next" from a single hope. Both stills are contracts.

The AI image to video tool is the launcher. The stills row that should have produced the plate is Kling Image 3.0 at 3 credits. If you do not have a still, you are on Kling O3 Pro Text to Video — also 70 credits, also 3s–15s, also audio, and a different identity lottery.

Seventy credits is the first five seconds

The matrix is base_plus_per_second: 70 for the first 5s, 14 each extra second. A 3s or 4s clip still sits on that 70-credit base. Treat 70 as the cost of trying the plate in motion, not as a tip. If the still is wrong, you threw 70 credits at a label that never passed still review.

Camera control is listed in the features. Use it as a control and as prompt language, not as a way to recompose a bad crop. A start frame that is already too tight will not become a wide by asking for a crane.

Kling's roster is large. This page is only O3 Pro I2V. Standard O3 I2V is a different published card at 56 credits. Do not mix the bills in your head.

Audio is available; it is not the still

Catalog audio is true. The O3 generate-audio default is off — unlike some Kling V2.6/V3 rows that default on. If you want a soundtrack, turn it on and write the sound. If you leave it off, do not be surprised by a mute plate. If you turn it on with a silent prompt, you can still get a bed you did not choose. Either way the still remains the contract: a talking mouth on an unapproved face is two problems.

Do not use this row to invent a presenter. That is lipsync or a talking model. O3 Pro I2V animates a plate.

FAQ

What if I only have a text prompt?

Then this is the wrong row. O3 Pro Image to Video requires a start frame. Use O3 Pro Text to Video, or lock a still on Image 3.0 first. Prompting a scene into an I2V endpoint is how 70 credits buy nothing on-brief.

Why mention 3s, 4s, and 5s when the enum goes to 15s?

Because 70 credits already cover the first 5 seconds, and most product turns fail or succeed in that window. Extra seconds are 14 credits each. Lengthen only the keeper.

Does reasoning-enhanced mean I can skip a clean still?

No. Reasoning-enhanced generation still starts from image_url. A soft, cropped-wrong, or unlicensed still is the whole video, just with nicer camera grammar.

Will I get native audio automatically?

Not by default on O3. Audio is in the catalog and the feature list; generate-audio defaults false. Decide the switch. Write the sound if you turn it on.