The Paragon v1.0 model is a boon for artists seeking photorealism in their work. Developed by the collaborative efforts of s6yx, FIKOsofter, Lykon, and CornmeisterNL, this model is built on the Stable Diffusion 1.5 framework. It is renowned for its ability to generate super realistic portraits that are really close to real photographs. The model's strong contrast is a feature that has garnered much appreciation. It is versatile, capable of generating portraits in various styles, ages, and clothing. The LoRa versions of this model have been particularly well-received for their performance.
Delving into the technical architecture of Paragon v1.0, it is designed to work optimally with Euler A or DPM++ SDE Karras schedulers with 25 or more steps. The model's recommended CFG Scale ranges from 5 to 12, and it performs better with Clip Skip 2. The model also comes with a recommended negative prompt that includes terms like "deformed iris", "semi-realistic", "cgi", "3d", "render", "sketch", "cartoon", "drawing", "anime", "mutated hands and fingers", "deformed", "distorted", "disfigured", "poorly drawn", "bad anatomy", "wrong anatomy", "extra limb", "missing limb", "floating limbs", "disconnected limbs", "mutation", "mutated", "ugly", "disgusting", "amputation", and "easynegative".
The advantages of the Paragon v1.0 model are manifold. Its ability to create photorealistic portraits is unmatched, making it a valuable tool for artists who require high levels of realism in their work. The model's strong contrast adds depth and dimension to the portraits, enhancing their lifelike quality. Furthermore, the model's versatility in generating portraits of different styles, ages, and clothing expands the creative possibilities for artists.
The license for the Paragon v1.0 model, known as the "CreativeML Open RAIL-M" license, is designed to promote both open and responsible use of the model. You may add your own copyright statement to your modifications and provide additional or different license terms for your modifications. You are accountable for the output you generate using the model, and no use of the output can contravene any provision as stated in the license.