[Lao Jin Station] FLUXSigma Character Extreme Details 32K Extreme Enlargement

• Three tokenizer models need to be set up. Lao Jin uses the GGUF model, and you can choose the corresponding version of the model based on your machine's configuration.

• If using accelerated LORA, their weights vary: turbo model is 1, openflux model is 0.33.

• When using accelerated LORA, the steps are 8 and 16. If not using accelerated LORA, the steps are 50 and 60.
• CFG is 2 and 4.
• Negative prompt words also work, and you can fill them in as needed.
• If you want to draw a girl, more complex prompts are required. Using only a few words can easily result in strange images.

The current machine configuration can only run up to 8192x8192 resolution.

If you want to see the effect after enlargement, you can follow the WeChat official account: Lao Jin Station

Enter "FluxSigmaSample" in the backend to get the sample link.

WeChat Official Account: Lao Jin Station