FreeVideo AIThree ways to guide H3
Text-to-video builds a scene from your description. Image-to-video starts from one reference picture. First-and-last-frame uses two images to guide an opening and ending. These are creative constraints, not a guarantee that every object, face or final frame stays pixel-perfect.
Public output settings
Choose 5, 10 or 15 seconds. Current landscape output is 896 × 512 and portrait output is 512 × 896 at 24 fps, approximately the 16:9 and 9:16 options shown in the studio. Frame-count rounding can make duration vary slightly. The public workflow produces silent MP4 footage and has no public 2K/4K upscale switch.
Direct a shot rather than a whole film
H3 is easier to evaluate when one prompt describes one continuous shot. Name the subject, a concrete action, the environment, lighting and a single camera move. Avoid combining a rapid orbit, a zoom and multiple cuts in the same short clip. Use a reference image when the initial look matters.
Plan for a queued workflow
Text tasks cost 6 credits; image-based tasks cost 8. New Google accounts receive 60 credits valid for 30 days. Only one task per account can be active. Refreshing the page can recover the current task; a network interruption is not proof that it failed, so avoid repeated submissions.
A prompt to start with
An explorer stands beneath colossal interlocking orbital towers, luminous bridges disappearing into clouds, slow forward tracking shot, cold rim light, one continuous shot.
Use this as a starting point, then adjust one action or camera instruction at a time. It is a prompt example, not a guarantee of a specific result.
Try this workflow →