: If the image has a very specific character or art style, the metadata will list "LoRAs" (small sub-models) that were active. 4. Reverse Image Search
: The prompt likely contains keywords for lighting (e.g., "volumetric lighting"), style (e.g., "digital painting"), and subject matter.
: Look for whether it was made with SDXL , SD 1.5 , or a custom model like Pony Diffusion or Realistic Vision . 00016-3878770009.png
: Upload the file to the PNG Info tab in Stable Diffusion (Automatic1111) or a site like Civitai .
Since I cannot see the specific image file you are referencing, 1. Extracting the Metadata (The "Recipe") : If the image has a very specific
: This will reveal the Prompt , Negative Prompt , Sampler , CFG Scale , and the Model (Checkpoint) used. 2. Understanding the Filename Structure
Most AI-generated PNGs contain "invisible" text data (EXIF/metadata) that holds the instructions used to create them. To see the guide for your specific image: : Look for whether it was made with SDXL , SD 1
: This is the Seed . If you use this exact number with the same settings and model, you will get the exact same image again. 3. Recreating the Image