Prompt Builder
Create a JSON prompt from scratch
Assumes you upload a reference photo to the model. The JSON updates live as you choose options.
JSON prompt
Live output updates with every filter.
{
"prompt": {
"identity": {
"reference_photo_usage": "Use the uploaded reference photo as an identity source, not as a pose lock. Preserve recognizable identity while adapting pose, gaze, lighting, outfit, and composition to the selected settings.",
"gender": "Infer this from the uploaded reference photo. Keep it consistent with the reference unless the user explicitly overrides it.",
"age_range": "Infer this from the uploaded reference photo. Keep it consistent with the reference unless the user explicitly overrides it."
},
"subject": {},
"scene": {
"location": "Infer this from the uploaded reference photo. Keep it consistent with the reference unless the user explicitly overrides it."
},
"outfit": {
"main": "Infer this from the uploaded reference photo. Keep it consistent with the reference unless the user explicitly overrides it."
},
"hair": {
"direction": "Infer this from the uploaded reference photo. Keep it consistent with the reference unless the user explicitly overrides it."
},
"expression": {
"direction": "Infer this from the uploaded reference photo. Keep it consistent with the reference unless the user explicitly overrides it."
},
"composition": {
"people_count": "Infer this from the uploaded reference photo. Keep it consistent with the reference unless the user explicitly overrides it.",
"shot_type": "portrait"
},
"camera": {
"lens": "50mm natural portrait lens"
},
"lighting": {
"setup": "soft natural window light"
},
"style": {
"visual_style": "photorealistic editorial portrait",
"color_mood": "cinematic natural color grading"
},
"output": {
"aspect_ratio": "4:5",
"target_model": "ChatGPT Image",
"quality": "high detail, realistic skin texture, natural anatomy"
},
"extra_details": [],
"negative_prompt": [
"low quality",
"distorted face",
"bad hands",
"plastic skin",
"over-smoothed skin"
]
}
}
