Introduction to Image Generation

Lesson 1 • What Is AI Image Generation

Defines AI image generation and distinguishes it from traditional digital art. Establishes vocabulary used throughout the course.

Lesson 2 • Core Concepts in Machine Learning

Introduces the machine learning principles underlying image generators. Connects abstract math concepts to practical generation behavior.

Lesson 3 • History of Generative Visual AI

Traces the evolution from early neural art to modern diffusion models. Provides context for understanding current capabilities and limitations.

Lesson 4 • Overview of Major Model Architectures

Surveys GANs, VAEs, and diffusion models at a conceptual level. Prepares students to choose the right tool for a given task.

Lesson 1 • Text-to-Image Conditioning

Explains how text prompts guide the denoising process via conditioning. Links language understanding to visual output quality.

Lesson 2 • Model Checkpoints and Variants

Distinguishes base models from fine-tuned checkpoints and merged variants. Prepares students to select appropriate models for specific visual styles.

Lesson 3 • Sampling Methods and Schedulers

Introduces common samplers and their effect on speed and image quality. Enables students to make informed sampler choices in practice.

Lesson 4 • Latent Diffusion Models

Covers how operating in latent space reduces computation while preserving quality. Explains the encoder-decoder pipeline used in modern tools.

Lesson 1 • Style and Artist References

Explores how referencing artistic styles and movements shapes visual output. Addresses ethical considerations when invoking living artists.

Lesson 2 • Token Weighting and Emphasis

Covers syntax for increasing or decreasing token influence in supported tools. Enables fine-grained control over compositional and stylistic elements.

Lesson 3 • Positive and Negative Prompting

Explains how negative prompts exclude unwanted elements from outputs. Demonstrates the interplay between positive and negative prompt weight.

Lesson 4 • Iterative Prompt Refinement

Introduces a systematic process for diagnosing and improving prompt failures. Builds the habit of evidence-based iteration over random trial and error.

Lesson 1 • CFG Scale and Prompt Adherence

Defines classifier-free guidance scale and its effect on creativity vs. accuracy. Guides students to calibrate CFG for different prompt types.

Lesson 2 • Step Count and Quality Trade-offs

Analyzes how inference steps affect detail, coherence, and generation time. Establishes practical step ranges for prototyping vs. final output.

Lesson 3 • Batch Size and Variation Strategies

Teaches generating multiple candidates simultaneously to accelerate selection. Connects batch workflows to efficient creative decision-making.

Lesson 4 • Seed Values and Reproducibility

Explains how seeds initialize noise and enable exact result reproduction. Demonstrates seed variation for controlled creative exploration.

Lesson 1 • Image-to-Image Fundamentals

Introduces img2img as guided generation seeded from an existing image. Explains denoising strength as the primary control parameter.

Lesson 2 • Outpainting and Canvas Extension

Teaches extending image boundaries beyond the original frame coherently. Demonstrates use cases in scene expansion and format conversion.

Lesson 3 • Upscaling and Detail Enhancement

Covers AI upscaling methods and high-resolution fix workflows. Produces print-ready or large-format images from lower-resolution generations.

Lesson 4 • Inpainting Workflow Essentials

Covers mask creation and inpainting to replace or repair image regions. Connects masking precision to output coherence.

Lesson 1 • Depth and Normal Map Controls

Uses depth and normal maps to maintain three-dimensional spatial relationships. Supports realistic lighting and perspective in generated scenes.

Lesson 2 • Pose and Skeleton Controls

Applies OpenPose and DWPose to lock human body positions in generated images. Enables consistent character posing across multiple outputs.

Lesson 3 • Combining Multiple ControlNets

Demonstrates stacking ControlNet units to enforce simultaneous constraints. Teaches weight balancing to prevent conflicting guidance artifacts.

Lesson 4 • Introduction to ControlNet

Explains ControlNet as a conditioning layer that adds spatial control to diffusion. Positions it relative to prompt-only and img2img approaches.

Lesson 1 • Textual Inversion and Embeddings

Teaches training new tokens to represent concepts without altering model weights. Produces lightweight, portable concept embeddings.

Lesson 2 • LoRA Training Fundamentals

Introduces low-rank adaptation as an efficient fine-tuning method. Produces small, stackable model adapters for style or subject training.

Lesson 3 • Evaluating and Deploying Custom Models

Establishes metrics and visual tests for assessing fine-tune quality. Covers packaging and sharing custom models responsibly.

Lesson 4 • Dataset Preparation and Curation

Covers image selection, cleaning, and captioning for training datasets. Quality of dataset directly determines fine-tune output quality.

Lesson 1 • Post-Processing and Compositing

Covers essential post-processing steps to finalize AI-generated images. Integrates generated assets into broader design and compositing pipelines.

Lesson 2 • Batch Production at Scale

Teaches automation techniques for generating large asset volumes efficiently. Covers scripting, queuing, and output organization strategies.

Lesson 3 • Quality Assurance and Review

Defines quality criteria for anatomy, coherence, and brand alignment. Establishes a structured review process before client delivery.

Lesson 4 • Designing a Repeatable Workflow

Structures generation tasks into defined stages from brief to final delivery. Reduces rework by standardizing decision points and checkpoints.