Image Generation

Use this skill to generate campaign images via NanoBanana (Gemini image generation).

Process

1. •

   Infer Visual Style from Research

   Based on brand intake and competitor research, determine:

   • •Color palette: Brand colors or complementary
   • •Photography style: Lifestyle, product-focused, abstract, UGC
   • •Aesthetic: Minimalist, bold, playful, premium, editorial
   • •Text overlay: Required or clean images

   Example inference:

   "Based on your brand assets and competitor analysis, I recommend a lifestyle photography style with warm, earthy tones. Your competitors use polished studio shots, but user-generated-style content outperforms in your niche."

2. •

   Confirm with User

   Present style recommendation and ask:

   "Does this visual direction feel right for your brand? Or would you prefer a different style? Options include:

   1. •Hyper-realistic product photography
   2. •Minimalist with bold typography
   3. •Lifestyle/UGC aesthetic
   4. •Illustrated/graphic style
   5. •Something else (describe)"

   Wait for confirmation before generating.

3. •

   Generate Images

   Use NanoBanana with appropriate aspect ratios:

   • •Instagram Feed: 1:1 (square) or 4:5 (portrait)
   • •Instagram Stories: 9:16 (vertical)
   • •Carousel slides: 1:1

   For each image, create a detailed prompt:

   code

   [Scene description]. [Style: lifestyle/minimalist/etc]. 
   [Color palette: brand colors]. [Mood: energetic/calm/bold]. 
   [Composition: centered product/lifestyle scene/text overlay space].
   High quality, professional marketing photo.
   

4. •

   Prompt Engineering Best Practices

   Include in prompts:

   • •Scene and subject description
   • •Style keywords (e.g., "editorial photography", "flat lay")
   • •Lighting (natural, studio, golden hour)
   • •Color guidance (specific colors or mood)
   • •Composition notes
   • •Quality markers ("professional", "high resolution", "marketing quality")

   Avoid:

   • •Vague descriptions
   • •Too many conflicting elements
   • •Text in images (Gemini struggles with text)
5. •

   Save Images

   Use the save_campaign_asset tool:

   python

   save_campaign_asset(
       content=image_bytes,
       brand_name="brand-name",
       asset_type="images",
       filename="instagram-post-1.png"
   )
   

Example Prompts

Lifestyle product shot:

A woman in her 30s using [product] in a bright, modern home office. 
Natural window lighting, warm tones. 
The product is visible but the focus is on the calm, productive moment.
Lifestyle photography, editorial quality, soft shadows.
Colors: [brand primary], neutral whites and beiges.

Bold graphic for carousel:

Minimalist background in [brand color]. 
Clean composition with ample negative space on the right for text overlay.
Subtle gradient from [color 1] to [color 2].
Modern, premium aesthetic. Sharp edges, clean design.

UGC-style product:

Authentic-looking photo of [product] on a messy desk with coffee and notebook.
Shot from above, iPhone-quality aesthetic.
Natural daylight, slight shadows. Lived-in, real feeling.
Not overly polished - casual lifestyle shot.

Critical Rules

1. •Infer first, confirm second - Show reasoning then ask
2. •Appropriate aspect ratios - Match platform specs
3. •Detailed prompts - Specific descriptions generate better images
4. •No text in images - Create text overlays separately if needed
5. •Save to deck folder - Use the custom save tool

Output

Save generated images to campaign-deck/[brand]/images/