Sorry, I really don't have enough computing power to train. I can only merge numerous checkpoints based on my own aesthetic sense. This is the satisfactory result I finally got: real light and shadow, skin texture, material details and spatial perspective, presenting a 2.5D aesthetic model that lies between illustration and reality and is highly impactful.

• Native support for high-quality ???? content generation, without the need for additional unlocking or suppression of negative prompts. The degree of freedom is extremely high.

• Excellent restoration of hand, foot, facial features and hair details

• Excellent lighting and material presentation: metallic, leather, silk, water surface, skin highlights, etc., all with rich textures

Recommended Settings

• Samplers: Euler A | Euler

• Steps: 25+

• Resolution: 1024+

• CFG: 3-6

欢迎各位大佬来交流群一起讨论：1075448864