Kling 4.0 First and Last Frame Workflow

The short answer

First and last frame control lets you define both ends of a Kling 4.0 clip. You upload a starting still and an ending still, describe what happens between them, and the model generates the motion that connects the two. It is the most directed mode in the studio, and the one that benefits most from preparation.

The technique is useful when the destination matters as much as the starting image: a product rotating to its label side, a character turning from profile to camera, a door closing on a specific composition.

What the two frames actually do

The first frame fixes the opening composition, subject, wardrobe, and light. The last frame fixes where the shot lands. Everything between those two anchors is the model's job.

That has two consequences. First, your prompt should describe the transition, not the pictures. Second, the two stills need to agree with each other. If the first frame is a wide shot at noon and the last frame is a close-up at dusk, the model has to invent a zoom, a time jump, and a lighting change in five seconds. It will try, but the result is usually soft.

When to use first and last frame

This mode is a good fit when:

• The ending is the point. A logo reveal, a product label, a character arriving at a destination.
• You need a controlled transition. The subject turns, the camera moves, the light changes, but the start and end must stay exact.
• You are matching shots. The last frame of one clip can become the first frame of the next, which keeps continuity across a sequence.
• The stills already exist. Storyboard frames, product renders, or frames pulled from a previous generation.

It is a poor fit when you want the model to surprise you. If you have only a rough idea of the ending, generate with a single reference image and let the clip resolve on its own.

The workflow, step by step

1. Prepare the first frame. Choose the composition you want to open on. Keep the subject at a sensible scale for the frame.
2. Prepare the last frame. Match the lens, lighting direction, and color temperature of the first frame. Change only what should change.
3. Write the transition prompt. One sentence: what moves, how the camera behaves, and what the light does.
4. Test at 5 seconds and 720P. Five seconds is enough for a turn, a push-in, or a reveal. Ten seconds is for a two-beat transition.
5. Compare the endpoints. Watch the first and last second against your stills. If the model misses an endpoint, the stills are usually the problem, not the prompt.
6. Regenerate at 1080P. Once the transition holds, make the final version.

Writing the transition prompt

The prompt should read like a description of the gap between the two images. Useful patterns:

• Subject-led: "The character turns from profile to face the camera, shoulders following, coat settling."
• Camera-led: "The camera pushes in slowly from the wide shot to the close-up, keeping the product centered."
• Light-led: "Afternoon light fades into blue dusk as the streetlights come on, the subject staying still."
• Reveal-led: "The door swings open to reveal the lit interior, the camera holding its position."

Avoid describing details that are already fixed in the stills. Restating the wardrobe or the background spends prompt budget without changing the output.

A compatibility checklist for the two stills

Before you generate, compare the images:

• Framing: same lens feel, same subject size, same horizon.
• Light: same direction and softness; a small time-of-day change is fine, a full relight is not.
• Color: matching white balance and palette.
• Subject: same wardrobe, same identifying features, no pose that requires a new limb or prop.
• Background: compatible geometry, so the camera can travel without rebuilding the set.

If one item fails, fix the still instead of adding more instructions. A five second clip cannot resolve a mismatch that the images never agreed on.

Common mistakes

Using two unrelated images. First and last frame is not a morph tool. It works when both images describe the same scene at different moments.

Writing a full scene description. The prompt is for the transition. Long descriptions of the start and end leave the middle unspecified.

Choosing 10 seconds for a simple move. A turn or a push-in fits in five seconds. Extra time invites drift.

Ignoring the final frame during review. Pause on the last second and compare it with the still. If it does not match, the shot is not finished.

Generating at 1080P first. Test the transition at 720P. Resolution does not fix endpoint mismatches.

FAQ

Is first and last frame available for text-to-video?

It is part of the image-to-video workflow, because it needs two stills. Text-to-video starts from a prompt and has no anchor frames.

How much does a first and last frame clip cost?

The same as any other Kling 4.0 clip: 9 credits for 5 seconds and 18 credits for 10 seconds. Frame control does not add a surcharge.

Can the two stills have different aspect ratios?

Use the same aspect ratio for both. Different crops force the model to invent a reframing move, and the result is harder to control.

What if the last frame is ignored?

Check compatibility first: framing, light, color, and subject. If they match, shorten the prompt and describe one motion instead of several.

Can I chain clips for a longer sequence?

Yes. Use the last frame of one clip as the first frame of the next. Keep the prompt for each segment focused on its own transition, and expect to iterate on the joins.

Conclusion

First and last frame control turns Kling 4.0 into a shot-planning tool: you decide where the clip starts and where it lands, and the model fills the motion. The work happens before generation, in the two stills and one clear transition sentence. When the endpoints match, move to 1080P and keep the take. For prompt patterns that work in every mode, read the Kling 4.0 prompt guide; for costs, see the credit breakdown.