Loading...
Loading...
Animate still images into dynamic videos.
Image to video animates a still picture. You give the model one frame plus a description of what should happen, and it returns a short clip that begins from your image. Because the opening frame is fixed, this is the most controllable route into video: whatever you settled on in a still, whether a product shot, a character or a piece of artwork, stays recognisable once it starts moving.
The input frame does most of the work, so it pays to get the still right first, often with a text to image model, before spending a video request on it. Keep the motion request modest as well: a slow push in, a turn of the head, drifting smoke or cloth read far better than a demand for several simultaneous actions. Sharp, well-lit sources animate more cleanly than noisy or heavily compressed ones, and the aspect ratio of the input frame carries into the clip.
Google's Veo 3.1 Fast serves this category on E2X today. It is called through the same interface as the rest of the catalog, a model slug plus an input object posted to the jobs endpoint followed by a poll on the returned job id. You pay per request from a prepaid balance, so a clip that does not land costs one more request rather than a plan upgrade.
A source image and a prompt describing the motion. The image becomes the opening frame of the clip that comes back.
Not from this category. If you need a defined final frame, use start and end frame to video, which takes both ends and fills in the motion between them.
Considerably. A clean, sharp, well-lit frame animates far better than a small or heavily compressed one, and the input aspect ratio carries through to the output.
Per request, from your prepaid balance, at the price listed on the model's catalog entry before you submit the job.