All content on AIBooru is generated with an AI image model, which generates latent noise and simultaneously converts it into an image visible to the human eye. Uploading an image to AIBooru will show all data extracted from the PNG metadata from that image. This can range from CivitAI friendly prompts to full on ComfyUI workflows.

Tagging Models

• Only fill the model name out if the model name or model hash is explicitly given (ie: given by CivitAI or by inspecting the image in ComfyUI or Automatic1111).
• Do not tag the base model if the model used was a finetune of that base model.
• If multiple models were used (as an otherwise unnamed merged model or in any other way), you can tag all the relevant models.

Do not guess what the model could be. Instead, tag the image with model_request. If the image's PNG info is wiped or otherwise returns nothing, tag the image with metadata_request as well. If a model is named, but cannot be verified, tag the image with the undefined tag.

If any LoRAs are listed in the AI metadata, or explicitly mentioned in the artist commentary, the lora tag should be used. The same can be done with ControlNets, regional prompting, latent couples, etc as well

Please see list_of_models for a list of commonly used AI models.