Create the cover. Make the clip.

An image prompt is a creative brief in miniature. A video prompt adds another question: what should change over time?

Integrity brings image and video generation into the workspace. Keep the brief, references, results, and feedback near the project so that each attempt builds on an explicit direction.

Describe the job before the style

Start with the placement and purpose. A wide campaign cover needs room for a headline. A small illustration beside an article needs a recognizable silhouette. A short clip should have one movement that remains clear when viewed on a phone.

For a fictional workshop campaign, the brief might be:

A monochrome editorial illustration of two people making a paper model together. Warm, simple shapes; one small coral accent. Leave the left third quiet for a headline. No lettering inside the image.

This names the subject, composition, palette, and a practical constraint. “Make it beautiful” leaves all four undecided.

Generate, inspect, and narrow the change

Choose an available image model and generate a first attempt. Inspect the result at its intended size. Look for awkward anatomy, accidental lettering, confused objects, and a composition that competes with your text.

For the next pass, identify one concrete issue. “Keep the composition; simplify the hands and move the coral shape away from the headline area” is easier to evaluate than asking for a completely different style every time.

Use reference images when the selected model and mode support them. Keep a record of which result the team approved so that later work does not drift back to an earlier draft.

Give video one clear action

A still image establishes a composition. A clip needs a beginning, movement, and an ending. Describe those plainly.

For the same paper-model scene, you might ask for a slow camera move as one folded piece is placed on the table. Keep the movement modest enough that the model can preserve the objects and the visual identity.

Available video modes vary by model. Some support an input frame or other references; duration, sound, editing, extension, and other controls are not universal. Choose the model and mode for the actual task instead of assuming every option supports every workflow.

Review the whole clip

Watch the beginning and end, then the transition between them if you intend to loop it. Look for drifting shapes, changing faces, unexpected text, and movement that distracts from the page.

For product demonstrations, use actual interface captures. Generated video is useful for illustration and atmosphere; it is not reliable evidence of how a button, document, or database behaves. A polished invented interface can mislead even when it looks convincing.

Keep the useful record

Save the approved result with its brief and a short note explaining the choice. Generated media uses AI credits, and uploaded or generated files count toward workspace storage. Model and output settings affect the cost.

The goal is a visual that does a clear job. One coherent cover and one purposeful clip can say more than a page full of unrelated motion.