Animation scenes generated by natural language can be used
*Important! The best output resolution is 768x1024, 1024x1024, 1024x768 (1408x768, 768x1024 large format is also acceptable). In short, use multiples of 64, and then control the image area to about 1024 squares. Adjust the proportions on this basis. Too small or too large an area is not conducive to the output effect
VAE: Default Sampling method: DPM++ 2M Karras Sampling steps: 30 CLIP: 2
Prompt word guidance coefficient CFG: 7 (you can adjust it yourself)