// Track

AI Image Generation

From Prompt to Pixel

From prompt to pixel — master GPT Image 2 as your primary production model, build fallback chains across Gemini, Leonardo, and Stable Diffusion, and learn the craft of visual prompting that separates operators from amateurs.

8 lessons~80 min total

Lessons are shown in recommended order. Complete them in sequence for the best experience — or jump to any lesson.

Lesson 96·9 min read

The AI Image Generation Landscape

Six platforms. One production chain. The operators who ship at scale have made their provider choices deliberately — not because one model is 'best,' but because they understand what each is optimized for and when to fall through to the next one.

Lesson 97·10 min read

The Art of Visual Prompting

Text prompts and image prompts are different languages. The operators who get cinematic output from AI image models have internalized five descriptors — Subject, Style, Lighting, Composition, Mood — and they stack them in order, every time, without exception.

Lesson 98·10 min read

Midjourney — The Creative Professional's Tool

Midjourney v7 sets the quality benchmark every other image model is measured against. Understanding its parameter syntax, Vary (Region) in-painting, and style reference system makes you dangerous for creative direction — even without an API.

Lesson 99·11 min read

GPT Image 2 — OpenAI's Visual Layer

GPT Image 2 is OpenAI's current production image model — gpt-image-1 is deprecated (shutdown October 23, 2026) and DALL-E 3 is already gone. Multi-image reference input, robust identity preservation, reliable text rendering, and multi-turn editing are why this track teaches it first.

Lesson 100·10 min read

Stable Diffusion and Open Source Image Gen

Open source image generation changes the cost structure completely. No per-image fees, no content filters you didn't add yourself, no rate limits you didn't impose. The tradeoff is infrastructure. This lesson is about when that tradeoff is worth making.

Lesson 101·9 min read

Leonardo AI — Production API and Cinematic Quality

Leonardo AI is where quality meets programmability. Phoenix 1.0 with Alchemy mode produces cinematic output through a clean REST API — no Discord, no self-managed infrastructure, no prompt rewriting. This is the fallback #1 in every serious production image pipeline.

Lesson 102·9 min read

Gemini Imagen — Google's Visual Intelligence

Google's Gemini API gives you two image paths: Gemini 2.5 Flash Image, with a real free tier that covers content pipeline volumes, and Imagen 4, the paid flagship for text rendering and photorealism. Both are synchronous, clean APIs — and they integrate directly into Claude Code sessions via the mcp-image MCP server. This is your primary slot.

Lesson 103·12 min read

Building a Production Image Pipeline

The complete fallback chain: Gemini → Leonardo → GPT Image 2. Quality gates, async generation, CDN delivery, cost tracking per image. This is the production architecture pattern behind every content pipeline that needs images on autopilot.