Choose portrait, landscape or square before generating
Match the video frame to your reference image and intended layout. The Follow image option is a useful default for photo input; text-only clips need an explicit framing choice.
Follow the image when composition matters
For image to video, start with Follow image to keep the original orientation. A tall portrait placed into a wide frame can require substantial reframing and change how much of the person or background is visible.
The studio detects the image dimensions and warns when a manual choice differs substantially from the source. Follow image helps with composition, but does not remove model artifacts.
Think about where the video will appear
Portrait gives a vertically arranged subject more room. Landscape gives wide scenery or multiple objects more horizontal space. Square can work for a centered product or face. Review the actual output dimensions before exporting to a platform with strict requirements.
For text to video, describe the composition as well: “full body remains in frame” or “wide coastal view with the horizon visible.” No reference photo exists to supply those framing details.
Avoid using a frame change to fix a poor source
If the source cuts off hands or feet needed for the motion, choose a better photo. Changing the output orientation cannot reliably reconstruct missing anatomy.
Test five seconds, inspect the edges of the frame and simplify the movement if the subject drifts out of view. The price is based on requested duration, not on portrait versus landscape.
Begin with a five-second take.
Photo or text. Landscape or portrait. Check the current price and node availability in the studio.
Start creating ↗