ChatGPT Codes and Higgsfield Video Starters
A reusable visual prompt cheat sheet for mockups, real-world placements, transformations, and feed-native motion ideas.
Tool split: ChatGPT codes produce still images from plain instructions. Higgsfield prompts are for video or motion and need the plugin connected.
Fast Visual Transformations
/billboard - place this product or design on a real street billboard
/packaging - wrap this design onto realistic product packaging
/pixel-object - rebuild the subject as a pixel-art object
/before-after - create a split image that sells the transformation
/bus-shelter - turn it into an illuminated transit poster at dusk
/shelf-test - place the product on a retail shelf beside plausible competitors
/carousel-cover - make a first slide with strong hierarchy and empty text space
/studio-clean - relight the subject as a clean studio product imageOne-Take Motion Ideas
Frozen Time
Product suspended in splash, powder, confetti, steam, shattering glass, or ingredients while only the camera moves.
Breaking the Frame
A product escapes a phone post, ad panel, printed page, sticker sheet, or gallery frame into the real scene.
Sticker and Cutout
Peel the subject off a printed image, press it onto a surface, or let paper layers separate and settle.
Surface Change
The product stays locked while the bed beneath it shifts from fruit to ice, wood to marble, or fabric to concrete.
Scale Illusion
Make the product enormous in a street, tiny on a fingertip, or architecture-scale with believable perspective.
Feed Native
Thumb stops the scroll, likes pile up physically, comments grow, a save icon files the product away, or a post becomes real.
Using the attached product image, generate a single continuous video.
Concept: [frozen splash / frame break / sticker peel / surface change / scale illusion].
Keep the product accurate and the label readable.
Use one camera move only: [slow orbit / push-in / lateral track / overhead locked shot].
Name the lighting and backdrop.
Do not add extra text, invented logos, or cuts.
Make the result feel like a premium product ad, not a random effect demo.Useful fixer: if an output looks generic, revise only camera, light, material, and label constraints. Keep the core concept unchanged so you do not lose the parts that worked.
Original Source
Adapted from Cindy Zhu's guide: 50 ChatGPT codes and 150 Higgsfield plugin prompts. The original contains the full long-form vault.