Updated: Nov 19, 2023
base model
Settings:
Sampler: DPM++ 2M Karras
CFG scale: 7
Steps: 25-40
Clip Skip: 2
Hires.fix: 4x-UltraSharp
Hires steps: 10-30
Denoising strength: 0.3-0.4
No VAE needed
Positive prompts: (cinematic lighting, volumetric), (film grain), (masterpiece, best quality, hires, high resolution:1.2), extremely detailed, realistic, intricate details, highres, - [ TBH you can just use either no image quality positive prompts and it will work fine or you can add just (best quality,masterpiece) ]
Negative prompts: bad-hands-5, bad_pictures, easynegative, ng_deepnegative_v1_75t, (worst quality, low quality:2), lowres, bad anatomy, normal quality, monochrome, grayscale, text, signature, watermark, logo, loli, teen, child, aged down, lolita, shota, minor, underage, flat chest,
AfterDetailer is a must!
Try out the SDXL Styles extension to give it a little kick, I found that it gives interesting images. I personally love using the "Fantasy Art", "Fairy Tale" and "Kirigami" styles but I do use the "Randomize Style" option when it gets repetitive.
If you are planning to use a character lora, the weights I find to work best hovers around 0.3-0.5 depending on the lora.
License & Use:
This model is openly accessible to everyone, subject to a CreativeML OpenRAIL-M license that specifies usage rights:
You are prohibited from purposefully creating or disseminating illegal or harmful content using the model.
The authors relinquish any claims to the outputs you produce. You have the freedom to utilize these outputs, but you are responsible for their usage, which must adhere to the terms outlined in the license.
You have the option to redistribute the model's weights. If you choose to do so, it is essential to include the same usage restrictions specified in the license and provide a copy of the modified CreativeML OpenRAIL-M to all your users. Please ensure you thoroughly and attentively review the license for complete details.
Use Restrictions:
You agree not to use the Model or Derivatives of the Model:
Exploiting or causing harm to minors in any manner.
Generating or distributing personally identifiable information with the potential to harm an individual.
Discriminating against individuals or groups based on legally protected characteristics or categories.
Creating or spreading knowingly false information or content with the intention of harming others.
Engaging in activities that violate any relevant national, federal, state, local, or international laws or regulations.
Offering medical advice or interpreting medical results.
Defaming, disparaging, or engaging in any form of harassment towards others.
Using the Model for fully automated decision-making that negatively impacts an individual's legal rights or establishes enforceable obligations.
Exploiting the vulnerabilities of specific groups based on their age, social, physical, or mental characteristics to distort their behaviour in a way that may lead to physical or psychological harm to them or others.
Using the Model in a manner that discriminates against or harms individuals or groups based on online or offline social behaviour or known or predicted personal characteristics.
Creating or disseminating information for use in the administration of justice, law enforcement, immigration, or asylum processes, such as predicting an individual's involvement in fraud or criminal activity through text profiling or drawing causal relationships between statements in documents in an indiscriminate or arbitrarily targeted manner.
Terms of use:
You bear complete legal responsibility for any consequences arising from the improper use of these models.
In the event you utilize any of these models for merging purposes, please specify the procedures you followed for doing so and clearly mark the locations where alterations have been implemented.