Wan 3.0 Prime First to Last Frame Video from Two Images
Last verified: November 1, 2026
Wan 3.0 Prime locks two images in as the opening and closing frames of a clip up to 30 seconds long, then generates the motion that carries the first image into the second. To use it, put the image where the shot starts in First Frame and the one where it has to land in Last Frame. Then describe the journey in the Prompt and set Video Duration so the change has the time it needs. That makes Wan 3.0 Prime first-to-last-frame generation a good fit for shots whose ending is already decided: a product reveal that has to finish on the packshot, a character who must end in a set pose, or a scene that needs to hand off cleanly to the next one.
Alibaba defines this mode strictly. The two images are assigned to the first and last frames, and its documentation calls this pixel-level reproduction of each, so neither end of the shot is left to interpretation. That changes how you plan. You can design the clip to cut against the shot before it and the shot after it, and approve both compositions before you generate anything. Each run then becomes a test of the motion alone. Prime is Alibaba's faster tier of Wan 3.0, which helps when you're tweaking the prompt and re-running between two approved frames.
Explore more first to last frame
Wan 3.0 Prime Settings at a Glance
The inputs, clip lengths and output options this generator gives you.
Input
First Frame (JPG, JPEG, PNG or WEBP, up to 10 MB) + Last Frame (JPG, JPEG, PNG or WEBP, up to 10 MB) + a text prompt of up to 5,000 characters
Video Duration
3 to 30 seconds
Resolution
480P, 720P or 1080P
Aspect Ratio
Auto, 1:1, 3:4, 4:3, 9:16, 16:9
Audio
Switch on or off, on by default
Output
1 video per run
How to Write Prompts That Move Cleanly Between Two Frames
Describe the Motion Between the Frames, Not the Frames Themselves
Write the Soundtrack Into the Prompt
Finish the Prompt with a Negative List
How to Turn Two Images into One Video
Six steps, from uploading your two frames to downloading the finished clip.
Upload the opening image in "First Frame"
Click "Upload first frame" and choose the image the clip should open on. Crop it to the framing you want before you upload. The model reproduces this picture as the very first frame rather than using it as loose inspiration, so the background, horizon line and subject position are all exactly where the shot begins.
Upload the closing image in "Last Frame"
Click "Upload last frame" and add the image the clip must end on. Both images are required. Give it the same shape as the first image, and keep the subject, lighting and setting close enough that one believable movement can connect them. The further apart the two pictures are, the more your prompt has to explain.
Describe the transition in the "Prompt" box
Write what happens between the two frames: the action, the camera path, how light or weather changes, and any sound you want. The box holds up to 5,000 characters, which is enough for a timed shot list. If you only have a rough idea, the "Enhance Prompt" button under the box suggests fuller versions. Your text only changes when you pick one.
Set "Video Duration", "Resolution" and "Aspect Ratio"
"Video Duration" runs from 3 to 30 seconds in one-second steps and starts at 3. That's tight for anything beyond a simple move, so give a big change like a room makeover ten seconds or more. "Resolution" offers 480P, 720P and 1080P, with 480P selected by default: stay there while you test the motion, then switch to 1080P for the version you keep. "Aspect Ratio" offers Auto, 1:1, 3:4, 4:3, 9:16 and 16:9 and starts on 1:1, so change it to match the shape of your two images.
Check audio and seed, then click "Generate"
Before you click "Generate", check "Generate Audio", which is on by default. Leave it on if you want generated sound, or switch it off if you'll add music in editing. Enter a number in "Seed" when you want to compare runs where only one setting changes. Each run returns one video.
Review and download the result
Watch the whole clip before you download it, and pay most attention to the middle. Your uploads fix the opening and closing frames, so the motion between them is the part they don't control. If the movement works, download it. If not, adjust the prompt or the duration and generate again.
Two Frames or a Different Wan 3.0 Prime Workflow?
When pinning both ends of the shot is the right call, and when another way of generating with the same model fits better.
| Criterion | Wan 3.0 Prime | Alternatives | Best For |
|---|---|---|---|
| Control over where the shot ends | Both uploaded images are reproduced as the clip's opening and closing frames, so the ending is settled before you generate. | Image To Video pins a single image as the opening frame, and your prompt decides where the shot goes from there. | Choose two frames when the final shot has to match a packshot, a storyboard panel or the next clip; choose Image To Video when you're happy to let the model find the ending. |
| Carrying a subject into new scenes | Consistency comes from the two frames themselves, which open and close the clip exactly as uploaded. | Reference To Video builds a scene you describe from reference images, video clips and audio that set a subject's look, motion and voice. | Pick two frames when you already have the exact start and end compositions; pick Reference To Video when the subject needs to appear in scenes you haven't created yet. |
| What you need before you start | You need two finished images plus a prompt that describes the change between them. | Text To Video starts from a written prompt alone. | Start with Text To Video when you're still exploring the look, and move to two frames once you've settled on the opening and closing compositions. |
Common Problems with Two-Frame Clips and How to Fix Them
What usually goes wrong between the first and last frame, and what to change before you run it again.
My widescreen images came out as a square video
The video cuts to a different shot halfway instead of one smooth move
Text, logos or signs look garbled in the middle of the clip
My image won't upload
The change between my two images looks rushed
Frequently Asked Questions
Do I need both a first and a last frame?
Yes. First Frame and Last Frame are both required, because the clip is built to travel from one image to the other. If you only have a starting image and want the model to decide where the shot goes, use the Image To Video page, which is built around a single frame.
Can I add reference images or my own audio track along with the two frames?
No. In Wan 3.0's first-and-last-frame mode, the only inputs are the two frame images. Reference images, reference video and audio can't be combined with them in the same request. For audio-driven generation, Alibaba points to the reference mode, which is the Reference To Video workflow for Wan 3.0 Prime on Vidofy. That's also the place to go when a subject's look needs to carry into new scenes.
Does Wan 3.0 Prime add sound to the clip?
It can. Generate Audio is switched on by default, so a run includes generated sound unless you turn it off. Alibaba lists audio texture as one of the areas still being refined. If the sound carries the piece, plan to replace or remix it in your editor and use the generated audio as a guide.
Which image files can I upload for the first and last frame?
First Frame and Last Frame each take one image in JPG, JPEG, PNG or WEBP format, up to 10 MB and at least 300×300 pixels. Both are required for every run. Using two images with the same shape and similar resolution keeps the start and end of the clip consistent with each other.
What makes one generation cost more than another?
The credit cost of a run is calculated from the options you choose and shown above the Generate button once you're signed in. Video Duration and Resolution are the settings that change it, so check the figure again after adjusting either one.