1.5- Simple three-view

It's harder for Lora to generate complex character images. Large model selection: You can try it out Number of sampling steps: 25-30 CFG: 7-8 lora control weight at: 0.65-0.8 Sampler: DPM++SDE A series Size: 768, 512 Positive Reminder: A simple character description is fine Just use the opposite: low quality, poor quality, extra fingers, long texture, deformed, You need to turn on HD Repair, and you can also draw a card if you don't open it. The return map below has parameters and can be directly generated. You may need to pull out the card and return the picture and like it by the way. Thank you here

Files with SHA256: 61a3bdf8eda4513b728e64378406b33ec85c67efd4453e9996de179537c93f73