AI Image-Generator Prompts Brief
Unfovi 7-Pack Makeup Sponges (Soft Puppy E1075) — 7 paired prompts + post-prod specs
1 · Overview
This brief is the bridge from the image audit + A+ Content brief into actual rendered images. Each entry below has:
- Reference image (the current state, embedded from the product folder)
- Nano Banana prompt — natural-language descriptive prompt for Gemini 3.1 Flash Image (AI Studio, "Nano Banana" model). Best for photography + lifestyle + product-in-context.
- GPT-Image prompt — structured prompt for OpenAI's image generator. Best for graphic-design compositions (callouts, comparison charts, infographics).
- Post-production overlay — headline, badge text, alt text. Added after the AI render in Canva or Photoshop. Two-part principle: AI for the photo, humans add headlines & badges in post.
- Filename — drop the final composite into the product folder using this name.
Run on: Nano Banana 2 (gemini-3.1-flash-image-preview) at 2K square aspect. GPT-Image at 1024×1024. Auto-driveable via Claude in Chrome MCP if requested.
What this brief does NOT cover
- Slot 1 hero refresh (diagonal isometric). Designer task — repositioning 7 existing sponges on a diagonal axis + micro-shadow grounding is a Photoshop composite, not an AI render.
- Slot 7 comparison chart. Native designer task — table layout + checkmarks + brand mark. AI image generators can do passable comparison charts but the typography hygiene is fragile.
- Slot 3 + Slot 4. Keep as-is per the audit.
- A+ Module 4 (comparison) + Module 6 (Q&A). Both are native Amazon module types — no external image needed.
2 · Image prompts (paired Nano Banana + GPT-Image)
1. Slot 2 keystone — Latex-Free objection pre-empt (rebuild)
Current slot 2 scores 37/80 (Generic tier). Mechanism #1 (category shortcut) is replaced with mechanism #3 (objection pre-empt).
| Render size | 2000×2000 px square (Amazon zoom-eligible) |
Current state (reference)
Nano Banana 2 prompt (Gemini 3.1 Flash Image)
GPT-Image prompt (OpenAI)
Post-production overlay (Canva / Photoshop)
- TOP HEADLINE (post-production, 80px+ Serif): "Latex-Free • Won't Drink Your Foundation"
- SUBLINE (40px Sans, below headline): "100% non-latex polyurethane foam. Hypoallergenic. Micro-pore structure for less absorption."
- BOTTOM-RIGHT BADGE TRIO (pill-shape, white background): • Latex-Free • Hypoallergenic • 7-Month Supply
- MACRO INSET LABEL: arrow pointing into the macro circle with text "Micro-fine pores — holds less foundation"
- FILENAME after composite: Images/slot2_final.jpg
- ALT TEXT (≤100 chars): "Latex-free Unfovi makeup sponge with micro-pore structure - hypoallergenic foundation blender for sensitive skin"
2. Slot 5 — Feature/benefit swap (replace duplicate)
Current slot 5 duplicates slot 3's anatomy callouts. Keep the model + sponge-to-cheek scene; swap the callouts to the three material-and-care benefits.
| Render size | 2000×2000 px square |
Current state (reference)
Nano Banana 2 prompt (Gemini 3.1 Flash Image)
GPT-Image prompt (OpenAI)
Post-production overlay (Canva / Photoshop)
- TOP HEADLINE (post-production, 60px Serif): "Engineered for Your Skin"
- LEFT-SIDE CALLOUTS (3 stacked, with arrows pointing to model/sponge): "Non-latex polyurethane — sensitive-skin safe" · "Micro-pore structure — absorbs less foundation" · "Damp-and-go — 30-second prep"
- FILENAME: Images/slot5_final.jpg
- ALT TEXT: "Latex-free hypoallergenic makeup sponge with micro-pore structure for sensitive skin and quick prep"
3. Slot 6 — Proof-of-outcome reframe (move macro inset to slot 2; reframe this slot)
Macro pore visual moves to slot 2 (where it does more work). Slot 6 reframes around the aspirational endpoint.
| Render size | 2000×2000 px square |
Current state (reference)
Nano Banana 2 prompt (Gemini 3.1 Flash Image)
GPT-Image prompt (OpenAI)
Post-production overlay (Canva / Photoshop)
- TOP HEADLINE (post-production, 60px Serif): "Bare to Blended. Effortless."
- SUBLINE (32px Sans): "Streak-free finish for liquid foundation, BB cream, and concealer."
- OPTIONAL LABEL on the left half: "Bare" · on the right half: "Sponge-applied"
- FILENAME: Images/slot6_final.jpg
- ALT TEXT: "Bare skin versus sponge-applied foundation showing streak-free blending finish with Unfovi makeup sponge"
4. A+ Module 1 — Premium Hero (3000×600 wide canvas)
A+ Content is currently parked. This is the new module 1 hero — wide canvas Premium Hero, the brand-promise frame.
| Render size | 3000×600 px (Premium Hero canvas) |
Current state (reference)
Nano Banana 2 prompt (Gemini 3.1 Flash Image)
GPT-Image prompt (OpenAI)
Post-production overlay (Canva / Photoshop)
- LEFT-ALIGNED HEADLINE (60-80px Serif): "Latex-Free Makeup Sponges. Drugstore Price."
- SUBHEAD (32px Sans, below): "Seven hypoallergenic blending sponges. For everyday foundation, concealer, and powder."
- RIGHT-ALIGNED Unfovi logotype + small tagline: "Everyday beauty, no compromise."
- FILENAME: Images/aplus-module1-hero.jpg
- ALT TEXT: "Unfovi 7-pack latex-free makeup sponges hero banner with hand holding lavender sponge - drugstore-priced beauty blender alternative"
5. A+ Module 2 — Material / safety deep-dive (970×600)
Latex-free risk-kill module with macro pore visual.
| Render size | 970×600 px |
Current state (reference)
Nano Banana 2 prompt (Gemini 3.1 Flash Image)
GPT-Image prompt (OpenAI)
Post-production overlay (Canva / Photoshop)
- TOP HEADLINE (40px Serif): "Made for Sensitive Skin"
- BODY TEXT (16px Sans, right panel): "100% non-latex polyurethane foam. Hypoallergenic. Dermatologically friendly."
- MACRO ARROW LABELS (left panel): "Micro-fine pores" + "Less absorption"
- FILENAME: Images/aplus-module2-material.jpg
- ALT TEXT: "Macro view of micro-pore latex-free polyurethane foam interior of Unfovi makeup sponge for sensitive skin"
6. A+ Module 3 — Anatomy + use-case (970×600)
Three application surfaces × three face zones. Collapses what was slot 3 + slot 5.
| Render size | 970×600 px |
Current state (reference)
Nano Banana 2 prompt (Gemini 3.1 Flash Image)
GPT-Image prompt (OpenAI)
Post-production overlay (Canva / Photoshop)
- TOP HEADLINE (40px Serif): "Three Edges. One Sponge. Every Zone."
- RIGHT-SIDE LABELS (3 stacked, 18px Sans): "1. Pointed Tip — under-eyes & nose" · "2. Flat Edge — contour & jawline" · "3. Rounded Base — full-face coverage"
- FILENAME: Images/aplus-module3-anatomy.jpg
- ALT TEXT: "Three application surfaces of teardrop makeup sponge - pointed tip flat edge rounded base for every face zone"
7. A+ Module 5 — Care & replacement cadence (970×600, 4-panel)
Four-panel care + value math module.
| Render size | 970×600 px (split into 4 equal panels of 242×600) |
Current state (reference)
Nano Banana 2 prompt (Gemini 3.1 Flash Image)
GPT-Image prompt (OpenAI)
Post-production overlay (Canva / Photoshop)
- PANEL TITLES (post-production, 24px Sans bold below each panel): "1. Rinse" · "2. Dry" · "3. Replace monthly" · "4. 7-month supply"
- PANEL CAPTIONS (14px Sans, one line under each title): "...with mild soap." · "Air-dries in 1 hour." · "For skin health & color performance." · "Always a clean sponge in your routine."
- FILENAME: Images/aplus-module5-care.jpg
- ALT TEXT: "Four-step care routine for Unfovi makeup sponges - rinse, dry, replace monthly, 7-month supply"
3 · QC checklist
QC checklist (every final composite)
- 2000×2000 px (gallery slots) or 970×600 / 3000×600 px (A+ modules)
- sRGB color profile · JPEG · ≤10 MB
- Headline reads at 100×100 thumbnail (slot 2 priority)
- All claims substantiated in bullets / description / A+ — no orphan claims
- No competitor brand names, no banned superlatives ("best", "#1", "top-rated")
- Filename matches the spec above (lowercase, hyphen-separated, no brand prefix)
- Alt text added in Seller Central upload step
- Mobile + desktop preview verified for A+ modules
4 · Auto-generation (optional)
If requested, this brief can drive Gemini / AI Studio in Chrome to auto-generate all seven renders. The flow:
- Navigate to
aistudio.google.comin the user's main Chrome (via Claude in Chrome MCP). - For each entry above, open a new chat, paste the Nano Banana prompt verbatim, set the model to
gemini-3.1-flash-image-preview, resolution 2K, aspect 1:1 (or 5:1 for module 1, 16:10 for modules 2/3/5). - Download each render to
Images/ai-image-.-nano.jpg - Human-in-the-loop QC: review each render, request re-roll if pass-quality is below 7/10.
- Pass the raw renders to Canva/Photoshop for the post-production overlay (headlines, badges, labels).
- Save composited finals as
Images/slot2_final.jpg,Images/slot5_final.jpg, etc.
Auto-generation is opt-in — requires Claude in Chrome MCP connected + user signed in to Google AI Studio. The default delivery is the brief above, run manually.