Model Basics

The foundational vocabulary of generative media models. Read top-to-bottom for a first pass, or dip in for a specific term.

Planned topics

• Diffusion intuition — how generation actually works, no math.
• Seeds — the reproducibility lever. (page)
• Samplers & schedulers — what they do, which to pick when.
• Steps & CFG — how much to think and how hard to listen to the prompt.
• LoRAs — lightweight style/subject adapters.
• Quantization & file types — why the same model ships in many sizes.
• Resolution & aspect ratios — native sizes and why going off-spec breaks things.
• Conditioning — ControlNet, IPAdapter, depth/pose/edge. The “steering wheel.”
• Embeddings & textual inversion — teaching a model a new word.
• Checkpoints vs LoRAs vs embeddings — disambiguation.
• Fine-tuning vs LoRAs vs prompting — when to reach for which.