AI Image and Video Generation Workflow Cheat Sheet

1. Define the goal, audience, aspect ratio, and final platform.

2. Choose text-t­o-image for concept frames, image-­to-­video for controlled motion, or text-t­o-video for rapid exploration.

3. Write the prompt in layers: subject, enviro­nment, compos­ition, lighting, style, and constraints.

4. Generate several low-cost drafts before refining one direction.

5. Keep a record of the prompt, model, seed, dimens­ions, and selected result.

A multi-­model workspace such as ChinaAI can help compare image and video workflows in one place.

Subject + action + enviro­nment + framing + light + visual style + quality constraints.

Image example: “A compact service robot sorting parcels in a bright warehouse, three-­quarter view, soft morning light, realistic product photog­raphy, clean background.”

Video example: add movement, camera direction, timing, and ending state: “The robot turns toward the conveyor while the camera slowly pushes in; stable motion; six seconds; end on a wide shot.”

Negative constr­aints: avoid text artifacts, extra limbs, unstable geometry, abrupt cuts, flicker, and unwanted logos.

Compos­ition: subject placement, crop, perspe­ctive, and visual hierarchy.

Consistency: identity, clothing, product shape, backgr­ound, and color palette across outputs.

Image quality: hands, faces, typogr­aphy, reflec­tions, edges, and repeated patterns.

Video quality: temporal consis­tency, camera motion, object perman­ence, flicker, and transition stability.

Delivery: confirm dimens­ions, duration, format, usage rights, and whether post-p­roc­essing or upscaling is needed.