What is a first-frame / last-frame pipeline in AI video?
By FXI Studio
Published September 16, 2026 · Updated November 8, 2026
The short answer
A first-frame / last-frame pipeline gives the video model two stills: the frame a clip starts on and the frame it ends on. The model generates only the motion between them. Chaining then reuses the real last frame of each clip as the first frame of the next, so a sequence of shots plays with no cut or crossfade.
TL;DR
The two frames are the controls: the first frame fixes where the clip starts, the last frame fixes where it lands, and the prompt describes only the move between them. For worked shot-size moves, see the shot-size tutorial; for chaining a full storyboard, see the storyboard tutorial.
Start free with 65 credits, no card required.
Try this prompt
“One continuous move. The camera glides from a sunlit kitchen window, across a table set for breakfast, and out the open back door into a blooming garden. Smooth steady dolly, soft morning light, no cuts.”
Opens the studio with this prompt filled in · 65 free credits
What is a first-frame / last-frame pipeline in AI video?
A first-frame / last-frame pipeline is frames-first video generation: the stills that open and close a shot are designed before any video is rendered, and the video model fills in the motion between them. Frame chaining extends the same idea across shots. Shot A is generated, its final frame is extracted, and that frame becomes the first frame of shot B.
Because the two shots literally share a frame, there is no cut and no crossfade. The camera appears to keep moving through the scene.
Why is frame chaining better than a crossfade?
Frame chaining beats a crossfade because a crossfade blends two unrelated images and reads as an effect. Frame chaining produces real motion the model authored, so a zoom, a travel, or a scene change looks like one continuous take.
In FXI Studio this is a built-in capability of the node pipelines, not something you assemble by hand.
What does the last frame control in AI video generation?
In AI video generation the last frame is a target, not an afterthought. When you supply one, the model has to land the shot on that exact image, which fixes where the camera ends, where the subject stands and what the light looks like at the cut. Without it the model picks its own ending, and the next shot has nothing to continue from.
That is why a last frame is worth designing as carefully as a first frame. In FXI Studio you approve both as stills before the video renders, so the expensive step only runs once the shot already starts and ends where you want. The free last-frame extractor pulls that frame from any clip you already have.
Step by step
How to frame-chain AI video
- 1Generate the establishing shot
Create shot A, the wide or opening frame of the sequence.
- 2Extract the last frame
Take the final frame of shot A to use as the starting point for the next shot.
Free last-frame extractor - 3Generate the next shot from that frame
Feed that frame in as the first frame of shot B so the model generates continuous motion from A into B.
- 4Repeat down the sequence
Use each shot final frame as the next shot first frame. Chain A to B to C for a seamless run with no jumps.
- 5Speed-ramp in editing
Optionally ramp travel portions faster and hold on detail frames slower to give the sequence a cut rhythm.
Frequently asked questions
What is a frames-first pipeline in AI video?
It is a workflow where the first and last frames of a shot are designed as stills before the video is generated, and the model authors the motion between them. Chaining reuses one shot’s last frame as the next shot’s first frame, so the shots share a frame and there is no cut or crossfade.
Is frame chaining better than a crossfade?
For continuity, yes. A crossfade blends two unrelated images and reads as an effect. Frame chaining produces real, model-authored motion, so the sequence looks like one continuous take.
How long does a chained transition take to generate?
A single chained transition typically generates in a few seconds. In FXI Studio, frame chaining is built into the node pipelines rather than assembled by hand.
Can a first-frame to last-frame shot change the shot size?
Yes. Give the model a wide shot as the first frame and a close-up of the same subject as the last frame, and it generates the camera move between them, usually a push-in. Spell the ending framing out in the prompt as well, so the model has words and an image to land on.
How do I plan a wide-to-close-up transition with AI?
Design both frames as stills from the same character reference and lighting, keep the subject near the same point in the frame, then generate the clip between them. For an extreme jump, such as a landscape to a macro detail, add a medium frame in between and chain two clips.
What does the last frame do in AI video generation?
It tells the model exactly where the shot must end: camera position, subject placement and light at the cut. Supplying a last frame makes the ending predictable and gives the next shot a frame to continue from. Leave it out and the model chooses its own ending, which is where most AI video sequences start to jump.
Chain your first shot free
Start free with 65 credits, no card required.
Related guides
Hands-on tutorials
- TutorialHow to turn a storyboard into one continuous shot
- TutorialHow to prompt P-Video for seamless transitions
- TutorialHow to Make an AI Drama With Gemini Omni 1.1
Reference documentation
Looking for a specific tool? Browse the documentation, the guides, or all tutorials.