Midjourney -- personalize. First, collect images that users like and dislike, using a multimodal model to infer the reasons behind their preferences. Then, utilize these attributes to generate images that align with the user's preferences.

With the continuous advancement of artificial intelligence technology, NVIDIA's Lumina-T2X image generation model brings new surprises. As an open-source model, it achieves aesthetic performance and image quality comparable to the industry-leading MJ V6, a remarkable achievement in the open-source field.

Midjourney recently introduced an innovative model personalization feature that allows users to fine-tune MJ models according to their aesthetic preferences. This feature achieves personalization by analyzing the images that users have previously liked and favored, making the generated image styles closer to the users' personal tastes.