# Day 7 — AI Image & Video Generation

- Treat the first generation as a draft. Iterate on the same result instead of starting from a blank prompt.
- For on-brand images always supply the asset's section or job, the style guide, and a palette screenshot.
- ChatGPT is a general image and enhancement tool; Ideogram is strong for legible text in graphics; OpenArt supports images and AI avatars.
- Finish assets with upscaling and background removal so a subject can be reused across layouts and variants.
- Canva's free tier is useful for templates, assisted creation, a small Brand Kit, and licensed stock media.
- Text-to-video tools based on models such as Veo create short B-roll from a prompt or still image, useful when real footage is unavailable.
- AI avatars and voice cloning can create presenter-led content and narration from a reference, while Higgsfield focuses on cinematic motion and HeyGen on avatar presentation.
- Editing turns raw generation into publishable work. OpusClip finds and reframes moments, Claude Design can make coded branded animations, and CapCut combines AI editing with manual control.
- Homework: create two or three on-brand ad images, generate B-roll, edit it into a finished piece with captions and music, and submit prompts.

## Relevance to the final project

- Build five visually distinct but coherent ad concepts. Keep the palette, type, tone, and visual identity consistent across ads, pages, emails, and the lead magnet.