Image Generation Settings
⚖️ Conditioning Scales
At 0 the prompted visual will remain intact, regardless of the control nets.
At 2 the control nets will be applied tightly to the prompted visual, possibly overriding the prompt.
Image generation engines can often generate disproportionate body parts, extra limbs or fingers, strange textures etc. Use negative prompting to avoid disfiguration or for creative outputs like avoiding certain colour schemes, elements or styles.
By default, each run produces one output per Model.
An increase in output quality is comparable to a gradual progression in any drawing process that begins with a draft version and ends with a finished product.
At lower values the image will effectively be random. The standard value is between 6-8. Values that are too high can also distort the image.