coreml-RealBiter / README.md
Zabriskije's picture
Metadata update: we changed the `not-for-all-eyes` format (#1)
0f0803d
metadata
license: creativeml-openrail-m
tags:
  - coreml
  - stable-diffusion
  - text-to-image
  - not-for-all-eyes

Core ML Converted Model:

  • This model was converted to Core ML for use on Apple Silicon devices. Conversion instructions can be found here.
  • Provide the model to an app such as Mochi Diffusion Github - Discord to generate images.
  • split_einsum version is compatible with all compute unit options including Neural Engine.
  • original version is only compatible with CPU & GPU option.
  • Custom resolution versions are tagged accordingly.
  • vae tagged files have a vae embedded into the model.
  • Descriptions are posted as-is from original model source. Not all features and/or results may be available in CoreML format.
  • This model was converted with vae-encoder for i2i.
  • Models that are 32 bit will have "fp32" in the filename.

Note: Some models do not have the unet split into chunks.

RealBiter:

Source(s): CivitAI

This blend model aims to achieve versatility in generating images that are almost realistic but with a touch of fantasy and a polished aesthetic. It performs well with both illustrations and simulated photographs, but it particularly excels with the latter.

It is recommended to use an appropriate VAE to avoid color gradients, and the vae-ft-mse-840000-ema-pruned model works especially well.

The blend includes the Offset Noise model, which enhances contrasts and black processing. Incorporating an Offset Noise LORA in image generation can further enhance the results.

The model is not specifically designed for NSFW content, but it easily generates this type of images.

The model has been reviewed with CLIP tensors checker and it shows no deviation.