Loading...
Loading...
Create videos from start & end frame.
This category takes two images, the frame the clip should open on and the frame it should close on, and generates the motion in between. You are specifying the destination as well as the origin, which makes it the most predictable video capability here: you already know what the last frame looks like before the job runs.
It suits transitions and controlled reveals, such as a product rotating from one angle to another, a scene shifting from day to night, a logo assembling itself, or a character moving from one pose into a second. The two frames should be plausibly connected, with the same subject and similar framing and lighting, because the model has to invent a believable path between them. Two unrelated images produce a morph rather than a movement, and matching aspect ratios keeps the result from being cropped.
Google's Veo 3.1 Fast is the model serving this category on E2X. It shares the same interface as the rest of the catalog: a model slug and an input object posted to the jobs endpoint, then a poll on the job id for the finished clip. Billing is per request against a prepaid balance, with the exact price listed on the model before you send anything.
Two, the opening frame and the closing frame. The model generates the movement that connects one to the other.
They can, but the result reads as a morph rather than as motion. Keeping the subject, framing and lighting related gives the model a path it can plausibly fill in.
When the final frame matters. Image to video only fixes the start, so the ending is whatever the model happens to produce.
One request per clip, drawn from your prepaid balance at the model's listed per-request price.