Creating models brings me pleasure and the most successful of them I share with you.
v4.2
- If you asked me to cut it, I did.
- added details and more realism
v4.1:
working on lighting: contrast balance.
removal of artifacts, more realism
more girls in the dataset
v3:
Made a merger of a large number of models and personal developments.
Looking for art style? You will find good detail, great backgrounds and interesting characters.
The model is sensitive to artist names, try adding them to the description and note their influence.
Downloading V3 will take you to the world of 2.5D art - concepts, portraits, the style of the classic artstation from 2014. And you will be fully satisfied with the implementation, but having stepped on the path of using V4, you will jump into the harsh world of reality, where important things are the detailing of hair and skin, where the idea and design of the surrounding space is sometimes lost, narrowing your view as in reality to two points eyes opposite. But if you tear yourself away from your eyes, look around, you will also notice the details. The choice is yours, whether you need a drawing or a hyper-realistic drawing.
Recommended parameters:
Sampler: Euler a, DPM++ SDE Karras
CFG Scale: 6-11
Resolutions: 512x832, 832Х512
Sampling steps: 30+
Clip Skip: 1.
Negatives: use any standard negatives: (worst quality, low quality:1.4), logo, watermark, drawing,
I appreciate your feedback and examples of work, it motivates me to continue!
Description
kirkwood-2:
Architecture
SD-v1
UNET-v1
UNET-v1-SD
CLIP-v1
CLIP-v1-SD
VAE-v1
VAE-v1-SD
Rejected
UNET-v1-Inpainting: Missing required keys (1 of 686)
model.diffusion_model.input_blocks.0.0.weight (320, 9, 3, 3)
UNET-v1-Pix2Pix: Missing required keys (1 of 686)
model.diffusion_model.input_blocks.0.0.weight (320, 8, 3, 3)
UNET-v2-SD: Missing required keys (64 of 686)
model.diffusion_model.input_blocks.4.1.transformer_blocks.0.attn2.to_v.weight (640, 1024)
model.diffusion_model.input_blocks.8.1.transformer_blocks.0.attn2.to_v.weight (1280, 1024)
model.diffusion_model.input_blocks.1.1.transformer_blocks.0.attn2.to_v.weight (320, 1024)
model.diffusion_model.input_blocks.2.1.proj_out.weight (320, 320)
model.diffusion_model.input_blocks.4.1.proj_in.weight (640, 640)
…
UNET-v2-Inpainting: Missing required keys (65 of 686)
model.diffusion_model.input_blocks.4.1.transformer_blocks.0.attn2.to_v.weight (640, 1024)
model.diffusion_model.input_blocks.8.1.transformer_blocks.0.attn2.to_v.weight (1280, 1024)
model.diffusion_model.input_blocks.1.1.transformer_blocks.0.attn2.to_v.weight (320, 1024)
model.diffusion_model.input_blocks.2.1.proj_out.weight (320, 320)
model.diffusion_model.input_blocks.4.1.proj_in.weight (640, 640)
…
UNET-v2-Depth: Missing required keys (65 of 686)
model.diffusion_model.input_blocks.4.1.transformer_blocks.0.attn2.to_v.weight (640, 1024)
model.diffusion_model.input_blocks.8.1.transformer_blocks.0.attn2.to_v.weight (1280, 1024)
model.diffusion_model.input_blocks.1.1.transformer_blocks.0.attn2.to_v.weight (320, 1024)
model.diffusion_model.input_blocks.2.1.proj_out.weight (320, 320)
model.diffusion_model.input_blocks.4.1.proj_in.weight (640, 640)
…
SD-v1-Pix2Pix: Missing required classes
UNET-v1-Pix2Pix
SD-v1-ControlNet: Missing required classes
ControlNet-v1
SD-v2: Missing required classes
UNET-v2
CLIP-v2
SD-v2-Depth: Missing required classes
Depth-v2
UNET-v2-Depth
CLIP-v2