A structured JSON prompt for an image-to-image transformation, instructing the model to preserve the original image's background and colors while adding tiny human workers performing specific actions like sweeping pigment or rolling paint, maintaining a slow, natural movement style.
{
"base_image": "use_uploaded_image_exactly_as_is",
"instructions": {
"preserve": {
"background": true,
"colors": true,
"shadows": true,
"water": true,
"sand": true,
"palette": true
},
"subjects": [
{
"type": "tiny_human_worker",
"actions": [
{"action": "sweep", "target": "pigment"},
{"action": "roll_paint", "target": "color_blocks"},
{"action": "climb", "target": "ladder", "careful": true},
{"action": "carry", "object": "small_bucket", "place_down": true},
{"action": "fix_edge", "tool": "brush", "target": "color_block"}
],
"movement_style": "small_natural_slow"
}
],
"effects": {
"special_effects": false,
"light_effects": false,
"glow": false,
"smoke": false
},
"camera": {
"movement": false
},
"objects": {
"add": false
},
"text": {
"add": false
},
"sound": false
}
}

