Lesson Content

Image as reference vs. animation.

Using an image in Pika has two distinct modes: (1) animation – animating the exact image as a starting visual; (2) reference – providing an image as a visual style or subject reference while the text prompt drives what actually happens in the generated video. Understanding which mode is in use (and which is intended) determines the expected output behavior.

Text prompt in combined generation.

When an image is used as reference, the text prompt describes what should happen in the generated video – the image influences the visual style and appearance while the text drives the content and motion. This can produce more visually anchored text-to-video results than text alone.

When to use combined generation.

Combined image+text generation is particularly useful when: you have a visual reference that defines the style or aesthetic you want (from a prior generation, a found reference image, or an illustration) but want to generate new content consistent with it, rather than animating the exact image. The image anchors the visual feel while the text defines the new content.

Evaluating combined results.

Evaluate whether the generated clip is visually consistent with the reference image's style (if used as reference) or whether the animation accurately reflects the motion description (if used as animation). These are distinct success criteria depending on the intended use of the image input.