<aside> 💡

Q: How is the model uploaded? What is the specific format? A: Typically it can be a Stable Diffusion model, which can generally be found on other model websites. The format is either .ckpt or .safetensors. The uploaded model is not a 3D model but an AI model used for generating images. For specific instructions, please click here for details.

</aside>

<aside> 💡

Q: Can self-trained models be downloaded? A: Self-trained models can definitely be downloaded. However, it is also possible to allow others to download them by adjusting the permissions.

</aside>

<aside> 💡

Q: What does the trigger word mean during model training? A: The trigger word can be understood as a passphrase. It is how the model knows that you want to activate a particular style. The default trigger word can be assumed as “LookX,” and for some models that do not have a trigger word, you can enter “none.”

</aside>

<aside> 💡

Q: What does a higher LoRA’s intensity (strength) represent? A: The LoRA model can be understood as a stylistic filter. The higher the strength, the more the generated results will lean towards the style of the LoRA model. The default strength is 0.5, and it is generally recommended to adjust it between 0.1 and 0.7. Different models or different images may require different strength setting.

</aside>

<aside> 💡

Q: What is the significance of large models? A: Large models are typically in .ckpt format with file sizes of 2 to 5 GB. They are pre-trained models designed for various scenarios and serve as a foundation. The LoRA models trained on the platform can be understood as stylistic filters.

</aside>