Download
1 variant available
fp16 SafeTensor
LTgirl20231103-000002.safetensors
Half precision, best balance • 5.19 GB
Verified: 3 years ago
9
63
Using more than 15,000 high-definition images with a resolution of 1024*1536, we trained the model for several days and nights. After the model was generated, the generated image effect was quite good, and only a small amount of prompt words were needed to produce satisfactory images. The compatibility of the model is also quite good! We welcome everyone to test and upload feedback, and provide more suggestions!
The suggestions are as follows:
Base model: chilloutmix_Ni.safetensors
VAE: chilloutmix_NiPrunedFp32Fix.vae.ckpt
Open ADETAILER, especially for full-body pictures. It's okay not to open it for close-up half-body pictures!
Suggestion: After 512768, use the method of doubling the size to speed up the image output, and you can also directly output images of 7681152 or 1024*1536.
These are already high-definition images, so don't overly emphasize the high-definition aspect, otherwise it will be overblown!