ControlNet
Compare Canny, depth, pose, line, and segmentation conditions, compatibility limits, and the narrower image tools available here.
Read the reference →Ecosystem reference
Stable Diffusion models and controls, kept separate from today’s product
These references explain model families, interfaces, and conditioning methods that remain useful across the Stable Diffusion ecosystem. They are learning pages—not a list of model selectors or controls available in this site’s current Image Generator.
Model families
Read the model page for the model—not as a promise about this generator
Workflow controls
Controls belong to the interface that actually implements them
A copied token or number does not activate a control. Check the selected model, pipeline, and interface before transferring a recipe.
Negative prompts
A separate conditioning input in pipelines that support it. Writing “negative prompt” in the main prompt box does not create that input.
Prompt weights
Some interfaces parse syntax such as (subject:1.2) or create weighted text embeddings. The syntax and effect depend on that parser, model, and pipeline; parentheses are not a universal control.
CFG and samplers
Inference settings exposed by some interfaces. Higher CFG is not automatically better, and sampler recipes are not portable across every model.
LoRA
A separately loaded compatible adapter—not an adjective or magic trigger. Naming a LoRA in a prompt does not load it here.
Stable Diffusion WebUI
A self-hosted interface family with a different operating model and control surface from this browser product.
Read the reference →Choose what you need to do now
If your goal is an image rather than ecosystem research, continue with the product workflow that matches your input and desired change.