Text-to-video begins with a written scene description. Image-to-video starts from an image and uses a prompt to describe how that scene should move. Neither guarantees perfect consistency.

Text-to-video

Good for early ideas, atmospheric clips and exploring multiple compositions without a fixed image. Describe the subject, setting, action, camera framing, motion and general visual approach.

EXAMPLE PROMPT · TEXT TO VIDEO

Wide shot of a quiet modern reading room at sunrise. One person walks past large windows. Slow lateral camera tracking, realistic dust in the light, muted warm colors, single continuous shot.

Image-to-video

Better when the first frame already contains the exact design, character, composition or product you want. The prompt should emphasize movement while keeping key elements steady.

EXAMPLE PROMPT · IMAGE TO VIDEO

Animate this exact reference image with a very slow camera push-in and soft moving reflections. Preserve the room layout, subject and product branding throughout.

Practical decision rule

CHOOSE TEXT-TO-VIDEO

You need to explore scenes from scratch and have no reference frame.

CHOOSE IMAGE-TO-VIDEO

You need the first frame to follow a specific composition or design.

For longer projects, plan multiple short clips and edit them together rather than assuming one long generation will stay consistent.

Official instructions

Runway documents both modes in its Gen-4.5 guide; Midjourney documents animation from images in its Video guide.

Primary sources

Official vendor documentation was used to verify the workflow. Some options depend on plan, region, product version or account access.