A beginner's guide to the Turbo-Enigma model by Shefa on Replicate

This is a simplified guide to an AI model called Turbo-Enigma maintained by Shefa. If you like these kinds of analysis, you should join AImodels.fyi or follow us on Twitter.

Model overview

turbo-enigma is a text-to-image model developed by shefa that applies Distribution Matching Distillation to a SDXL base. It supports zero-shot identity generation, producing high-quality images in 2-5 seconds. This model can be compared to similar fast text-to-image models like sdxl-lightning-4step and uform-gen.

Model inputs and outputs

turbo-enigma takes in a text prompt and various optional parameters to control the generation process. The output is a generated image.

Inputs

• Prompt: The text prompt to generate the image from
• Seed: A random seed value to control the image generation (leave blank to randomize)
• Image: An input image to guide the generation
• Width: The desired width of the output image
• Height: The desired height of the output image
• Guidance Scale: The scale for classifier-free guidance
• Num Refine Steps: The number of refinement steps to apply
• Num Inference Steps: The number of denoising steps to apply
• Faceswap Fast: Whether to use ONNXRUNTIME-GPU for fast faceswapping
• Faceswap Slow: Whether to use CPU-only ONNXRUNTIME and GFPGAN for slower but higher-quality faceswapping
• Save Embeddings: Whether to save the optimization experiment embeddings
• Disable Safety Checker: Whether to disable the safety checker for the generated images

Outputs

• Generated Image: The output image generated based on the provided inputs

Capabilities

turbo-enigma is capable of producing...

Click here to read the full guide to Turbo-Enigma