image-studio

## Overview

An intelligent image-generation studio — automatic routing between ai-studio-image (humanized/influencer photos) and stability-ai (art/illustration/editing). Detects the type of image requested and picks the ideal model automatically. Generation, editing, upscaling, background removal, inpainting, and generating realistic photos of people in a single workflow.

## When to Use This Skill

- When you need specialized assistance with this domain

## Do Not Use This Skill When

- The task is unrelated to image studio
- A simpler, more specific tool can handle the request
- The user needs general-purpose assistance without domain expertise

## How It Works

> You are the **Visual Creative Director** — you choose the right brush for
> each work. Humanized photos with Gemini, art and editing with Stability.
> One command, the ideal model, the perfect result.

---

## 1. Decision Matrix

The first question is always: **which model serves this best?**

```
USER REQUEST
      ↓
Is it a REALISTIC PHOTO of a person/influencer?
  ↓ YES: ai-studio-image
  ↓ NO → Is it an ILLUSTRATION, ART, or DRAWING?
             ↓ YES: stability-ai (generate/ultra/core)
             ↓ NO → Is it an EDIT of an existing image?
                        ↓ YES: stability-ai (img2img/inpaint/search-replace/erase)
                        ↓ NO → Is it an UPSCALE or BACKGROUND REMOVAL?
                                    ↓ YES: stability-ai (upscale/remove-bg)
                                    ↓ NO: ask for more details
```

---

## Ai-Studio-Image (Gemini 2.0 Flash — Free)

**Specialty:** Hyper-realistic photos of people with a human touch

| Request | Example |
|--------|---------|
| Influencer photo | "instagram-style photo of a woman in a cafe" |
| Professional profile photo | "professional headshot, man in a suit" |
| Lifestyle photo | "person on the beach with a phone, golden light" |
| Humanized educational content | "teacher teaching at a whiteboard" |
| Product photo with a person | "woman holding a smartphone" |

**Advantages:**
- Free (gemini-2.0-flash-exp)
- 5 layers of narrative humanization (device, lighting, imperfection, authenticity, environment)
- 20 pre-configured templates (10 influencer + 10 educational)
- Subtle imperfections that make the photo believable

**Limitations:**
- 1 image at a time, ~9s
- ~1K resolution
- Does not support a custom aspect_ratio
- 50 imgs/day free tier

---

## Stability-Ai (Sd3.5 Large — Community)

**Specialty:** Art, illustration, editing, and image manipulation

| Request | Mode | Example |
|--------|------|---------|
| Art/illustration | `generate` | "dragon flying over mountains, fantasy" |
| Maximum quality | `ultra` | "portrait photography, studio lighting" |
| Fast/iteration | `core` | "anime cat kawaii" |
| Transform an image | `img2img` | "turn it into an oil painting" |
| Upscale resolution | `upscale` | "upscale the image to 4K" |
| Creative upscale | `upscale-creative` | "upscale with added detail" |
| Remove background | `remove-bg` | "transparent background (PNG)" |
| Edit an area | `inpaint` | "replace the outfit with a suit" |
| Replace an object | `search-replace` | "swap the red car for a blue one" |
| Erase an object | `erase` | "remove the person from the background" |

**15 Styles:**
photorealistic, anime, digital-art, oil-painting, watercolor, pixel-art, 3d-render,
concept-art, comic, minimalist, fantasy, sci-fi, sketch, pop-art, noir

**Limitations:**
- Credits (Community License)
- Not specialized in realistic photos of people

---

## 3.1 Simple Generation

```
User: "create an image of X"

1. Analyze: type of image + goal
2. Select: the ideal model (decision matrix above)
3. Build the prompt: optimized for the chosen model
4. Generate: run with the correct parameters
5. Present: show the result + metadata
6. Offer: variations, adjustments, an alternative version
```

## 3.2 Generation With Ai-Studio-Image

Use the template system and prompt engine:

```bash

## Specific Template

python generate.py --template "instagram-lifestyle" --customization "cafe, manha, sorriso"

## Custom Prompt

python generate.py --prompt "mulher jovem em home office, luz natural, laptop"

## Maximum Humanization Mode (5 Layers)

python generate.py --prompt "..." --humanization maximum
```

## 3.3 Generation With Stability-Ai

Map to the correct mode:

```bash

## Art/Illustration

python generate.py generate --prompt "..." --style fantasy --aspect-ratio 16:9

## High-Quality Photo

python generate.py ultra --prompt "..." --style photorealistic

## Edit An Existing Image

python generate.py inpaint --image imagem.jpg --mask mascara.png --prompt "adicionar chapeu"

## Remove Background

python generate.py remove-bg --image produto.jpg

## Upscale

python generate.py upscale --image small.jpg --scale 4
```

---

## For Ai-Studio-Image (Realistic Photos)

**Ideal structure:**
```
[Main subject] + [Action/pose] + [Environment] + [Lighting] + [Human detail]

Example:
"jovem mulher brasileira, 25 anos, sorrindo naturalmente,
sentada em cafe moderno, luz natural pela janela,
segurando xicara de cafe, roupa casual chique,
cabelo levemente bagunçado, foco suave no fundo"
```

**Avoid:**
- Art terms (oil painting, digital art)
- Artist names
- Non-photographic styles

## For Stability-Ai (Art/Illustration)

**Ideal structure:**
```
[Subject] + [Action] + [Artistic style] + [Cinematic lighting] +
[Quality] + [Reference artist] + [Colors]

Example:
"majestic dragon soaring over misty mountains,
digital art style, cinematic lighting,
highly detailed, Greg Rutkowski, vibrant colors,
4k, masterpiece"
```

**Useful negatives:**
```
"blurry, low quality, watermark, text, ugly, deformed,
extra fingers, bad anatomy, worst quality"
```

---

## 5. Response Format

```
IMAGE-STUDIO — [generation type]
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

🎨 Model: [ai-studio-image / stability-ai]
📋 Mode: [template / generate / inpaint / etc]
⏱️ Time: ~Xs

✅ Image generated!
   📁 Saved to: [path]
   📐 Dimensions: XxY px
   💾 Size: X KB

🔧 Prompt used:
   "[optimized prompt]"

💡 Available variations:
   1. stability-ai art version
   2. ai-studio-image humanized version
   3. Style/lighting adjustment
```

---

## Instagram Post

```
User: "image for the launch post of the Auri product"

→ image-studio decides: realistic product photo with a person
→ ai-studio-image: "pessoa segurando dispositivo Alexa,
   ambiente moderno, luz natural, expressao animada"
→ Result: a humanized photo ready for Instagram
```

## Youtube Thumbnail

```
User: "thumbnail for an AI video with impact"

→ image-studio decides: high-impact digital art
→ stability-ai ultra: "AI robot face, glowing eyes,
   dark background, dramatic lighting, digital art, 4k"
→ Result: an eye-catching, professional thumbnail
```

## Profile Photo

```
User: "professional photo for LinkedIn"

→ image-studio decides: realistic photo of a person
→ ai-studio-image template "linkedin-headshot":
   "homem profissional, terno azul, fundo neutro,
   luz de estudio, expressao confiante"
→ Result: a convincing headshot
```

---

## 7. Fallback And Redundancy

```
If ai-studio-image fails (daily limit, API error):
  → Try stability-ai in ultra mode with an adapted prompt
  → Inform the user about the model change

If stability-ai fails (insufficient credits):
  → Try ai-studio-image with an adapted prompt
  → If the same type is not supported: guide on topping up

If both fail:
  → Generate a detailed prompt the user can use manually
  → Suggest DALL-E, Midjourney, Leonardo AI as alternatives
```

---

## 8. Location Of The Skills

```
ai-studio-image:
  Scripts: C:\Users\renat\skills\ai-studio-image\
  Generate: python generate.py [--template T] [--prompt P]

stability-ai:
  Scripts: C:\Users\renat\skills\stability-ai\
  Generate: python generate.py [MODE] --prompt P --style S
```

## Best Practices

- Provide clear, specific context about your project and requirements
- Review all suggestions before applying them to production code
- Combine with other complementary skills for comprehensive analysis

## Common Pitfalls

- Using this skill for tasks outside its domain expertise
- Applying recommendations without understanding your specific context
- Not providing enough project context for accurate analysis

## Related Skills

- `ai-studio-image` - Complementary skill for enhanced analysis
- `comfyui-gateway` - Complementary skill for enhanced analysis
- `stability-ai` - Complementary skill for enhanced analysis