Collective Action —Solo ↔ Group Participation

Collective Action — Solo ↔ Group Participation

Collective Action invites other people into whatever your subject is doing.

Prompt a subject performing an action normally, then increase the LoRA weight. As the weight rises, additional people tend to appear and plausibly join, help with, or participate in the same action. A person cooking may gain kitchen helpers; someone moving furniture may acquire extra hands; a lone musician may become an ensemble; an individual protest can become a movement.

The useful range is generally around +1.5 to +4. Lower weights tend to introduce the first additional participants, while higher weights produce increasingly large groups and more aggressive reinterpretations of the scene.

The important distinction is that this isn't intended simply to add random people to an image. The added people usually try to participate in the verb of the prompt. The exact solution is still determined by the checkpoint, prompt and seed, so sometimes the model finds surprisingly creative ways for everyone to help out.

Invite people responsibly. At high weights, Collective Action can become very collective: crowds multiply, compositions become dense, anatomy gets increasingly ambitious, and eventually ADetailer begins reconsidering its career choices.

No trigger phrase is required. Describe the subject and what they're doing, then adjust the LoRA weight to control how enthusiastically everyone else decides to get involved.

Suggested range: +1.5 to +4
Experimental/extreme: +5 and beyond
Trigger: None

The noun (subject) gets joined. The verb (action) gets shared.