4 (new) trained models were mixed in this merge + SPO LoRa as well as overall model + CLIP temperature was raised to 1.25 and perturbed attention was added at default value (3.0). That altogether makis it a great tool for your SDXL image generation for photo and artistic prompts!

1. https://civitai.com/models/133005?modelVersionId=782002

2. https://civitai.com/models/617652/natvis-natural-vision?modelVersionId=690475

3. https://civitai.com/models/312530/cyberrealistic-xl?modelVersionId=709456

4. https://civitai.com/models/299933/halcyon-sdxl-photorealism?modelVersionId=709468