Skip to main content

​
Overview

Instead of manually constructing PhotoVariation JSON, you can describe what you want in plain English. The text_to_variation() function uses Gemini 2.5 Flash to convert your description into a valid config.

​
Usage

import asyncio
from penquify.generators.config import text_to_variation

config = asyncio.run(text_to_variation(
    "blurry photo with a coffee stain on the upper right, taken from a strong angle "
    "with an old Samsung phone, warehouse lighting, worker wearing gloves"
))
print(config)
Output:
{
  "name": "warehouse_blurry_stained",
  "camera": "Samsung Galaxy A5 2017",
  "year_device_style": "2017 Android mid-range",
  "angle": "45 degree oblique",
  "skew": "strong",
  "motion_blur": true,
  "blur_direction": "horizontal",
  "glare": "mild",
  "uneven_lighting": true,
  "hand_visible": true,
  "glove": "warehouse glove",
  "stain": {
    "type": "coffee",
    "location": "upper_right",
    "opacity": "semi-transparent",
    "text_obstruction": "partial"
  }
}

​
Example Descriptions

DescriptionKey Config Generated
”perfect scan-like photo”angle: "straight", motion_blur: false, glare: "none"
”quick photo taken in a hurry by a warehouse worker”motion_blur: true, skew: "moderate", hand_visible: true, glove: "warehouse glove"
”document with grease stains, folded, old phone”stain.type: "grease", folds: "multiple", camera: "budget Android"
”cropped photo missing the header”cropped_header: true, missing_area: "top 10-15%"
”stapled multi-page document on a desk”stapled: true, stacked_sheets_behind: 2, hand_visible: false
The generated config follows the exact PhotoVariation schema. You can use it directly or modify specific fields before generating photos.