If you’ve already dipped into AI video tools, you know how game-changing they are. But if you want cinematic control, the kind that gives your videos cohesion, flow, and intention, then understanding how to use first and last frames is a must. This simple technique lets you guide your AI model, protect continuity, and shape the exact transformation you want.
It’s one of the most powerful ways to steer generative video, and using it well helps you create smoother motion, sharper storytelling, and more consistent visuals. If you want a deeper dive into the technique, check out our full guide to first and last frames. It’ll change the way you create.
We tried it ourselves, and the result was a slick 15-second video that blends fashion, fantasy, and world-hopping transitions. Below, we’ll show you exactly how we made it, and how you can replicate the process for your own videos.
Watch the full edit:
Starting with a single unexpected image
We began with our first image. A simple but bold image of a woman in a white coat.
This shot wasn’t planned. It actually popped up while our Video Content Specialist, Eden Barel, was generating around 20 images for a completely different project. One of the images didn’t fit the direction he wanted, but it looked too cool to leave behind. That’s the beauty of working with AI — is a creative partner that surprises you, pushes you, and hands you visuals that spark new ideas.
Here’s the original prompt Eden used to generate this image:
Prompt:
“High-fashion editorial photo, model wearing a bold monochrome puffer jacket riding a horse of the exact same color, photographed with hard flash at night, harsh direct lighting, deep shadows, high contrast, raw candid fashion aesthetic, glossy textures, 35mm on-camera flash style, minimal background, surreal symmetry, contemporary magazine look”
Creating the final frame for sequence 1
To build a controlled transition, Eden needed a matching final frame. He switched to the Nano Banana model and generated about 10 variations until he landed on the one that clicked.
His prompt asked the model to:
- Change the background to a night cityscape
- Place the character on a rooftop
- Swap her outfit to a black evening dress
The lighting was spot-on, the roof looked real, and the mood matched the vibe we wanted.
Building the first transition in Kling 2.1
With the first and last images ready, Eden dropped them into Kling 2.1 and used a simple guiding prompt: “Gradually change the background.”
The model created a transition that blended the two environments into a smooth, organic shift.
That result came from experimentation, too. At first, Eden tried prompting a 360-degree camera move around the woman. The model moved the character instead of the camera, but even that “mistake” looked cool. So he refined the prompt to get what he wanted:
Prompt: “Make a full 360 shot that rotates around the character. The movement changes the background smoothly.
And we had a winner! The result looked like a perfect seamless set design swap. This gave us the elegant rotating effect you see in the final video.
Creating sequence 2 with tight visual continuity
For the next sequence, Eden used his second image — the woman in the black evening dress as his new starting frame.
To create the end frame, he again turned to Nano Banana. The goal was simple: change the setting and outfit without shifting the character’s pose or composition. Continuity is everything. If your character moves between frames, your transition falls apart.
So his prompt instructed the model to:
- Move the scene to Colorado
- Keep the same position and framing
- Put the character in a yellow sweater
After a few iterations, the perfect end image appeared.
He then fed both images into Kling 2.1 and generated the second transition.
Building the final transition and closing the loop
Eden repeated the same workflow for the last sequence. He used the Colorado-yellow-sweater image as the new starting frame and generated the final image he needed with Nano Banana.
Prompt: “Change her clothes to a navy blue sweater on top of a white button shirt, gray trousers, and black loafers. Put her in a tropical jungle with plants all around.”
The next step was to upload both images to Kling 2.1 to produce the final shot.
At this point, he had three sequences, each flowing into the next.
The last step was simple: connect everything in Adobe Premiere Pro. Eden lined up each clip so one shot ended exactly where the next began. That created a seamless, continuous loop and a polished 15-second visual cycle that looks intentional from start to finish.
Try it yourself in Artlist’s Image to Video
If you want this level of creative control, start building your own first-frame and last-frame sequences in Artlist’s Image to Video Generator. It’s fast, intuitive, and built to help you turn static images into cinematic motion. You’ve seen how we did it. Now try it for your own video!
Check out more of Eden’s creative work :
Artlist
Instagram
Spotify
YouTube
Did you find this article useful?
