Calico Cat Tower is RouWei based checkpoint.

VAE is baked in.

RouWei has its own specific features and prompting, so please refer to the RouWei page for instructions on how to use it.

You need to use the webui that support v-prediction if you use v-prediction version.

• ComfyUI

• reForge

• Forge

• AUTOMATIC1111 (place the config file in the same folder as the checkpoint)

License

This model merges the NoobAI-Xl model, so the license follows NoobAI-XL License (Modified Fair AI Public License 1.0-SD) and is available for non-commercial use only. This license prohibits any form of commercialization, including but not limited to monetization or commercial use of the model, derivative models, or model-generated products.

Recommended Setting for V-prediction version.

• Steps: 20-28

• CFG scale: 3-5

• Sampler: Euler a

• Use ADetailer as needed

Recommended Setting for Epsilon-prediction version.

• Steps: 20-30

• CFG scale: 3-7

• Sampler: Euler a, DPM++ 2M Karras

• Use ADetailer as needed

Positive Prompt

masterpiece, best quality

Negative Prompt

worst quality, bad quality, low quality, lowres, scan artifacts, jpeg artifacts, sketch, light particles, watermark,

Merge recipes (e-pred v2.0)

1. Merging the difference between rouwei_080_epsilon_fp16 and stable-diffusion-xl-base-1.0, CyberRealistic XL v5.6 and stable-diffusion-xl-base-1.0 to rouwei_080_epsilon_fp16 by "Perpendicular Component" and "Add Difference" (Checkpoint A)

2. Merging Checkpoint A to rouwei_080_epsilon_fp16 by "SLERP", alpha = 0.5 (Checkpoint B)

3. Merging Checkpoint B to rouwei_080_epsilon_fp16 by "Rotate", alignment = 1.0, alpha = 0.0 (Checkpoint C)

4. Replacing the CLIP of Checkpoint C by rouwei_080_epsilon_fp16 (Checkpoint D)

5. Finetuning Checkpoint D with anime style dataset(0.5k) by OFT method, dim = 4, alpha = 1e-3, learning rate = 2.5e-6, 15000 steps. And merging the OFT model to Checkpoint D, ratio = 1.0 (Checkpoint E)

6. Merging Checkpoiint E to Checkpoint D by "Weighted Sum" with block (0,0,0,0,0,0,0,0,0,0,0,1,1,0,0,0,0,0,0,0) (Checkpoint F)

7. Merging the Cat Tower epred v1.3 to Checkpoint F by "SLERP" with block (0,0,0.5,0.5,0.5,0.5,0.5,0.5,0.5,0.5,0.5,0.5,0.5,0.5,0.5,0,0,0,0,0) (Checkpoint G)

8. Merging the difference between JANKU v4.0 and RouWei v0.7 epred to Checkpoint G by "Train Difference" with block (0,0,0,0,0,0,0,0,0,0,0,0.2,0.2,0.2,0.2,0,0,0,0,0) (Calico Cat Tower e-pred v2.0)

Merge recipes (v-pred v1.0)

1. Finetuning RouWei v0.7 vpred with anime style dataset(0.5k) by OFT method, dim = 8, alpha = 1e-3, learning rate = 1e-4, 15000 steps

2. Merging the OFT model to RouWei v0.7 vpred, ratio = 1.0 (Checkpoint A)

3. Merging Checkpoiint A to RouWei v0.7 vpred by "Weighted Sum", alpha = 0.3 (Checkpoint B)

4. Merging the difference between Cat Carrier v3.0 and Illustrious XL v1.0 to Checkpoint B by "Add Difference" with block (0,0,0,0,0,0,0,0,0,0,0,0.5,0.5,0.5,0.5,0,0,0,0,0) (Checkpoint C)

5. Merging the difference between Cat Tower vpred v1.6 and NoobAI-V-Pred-1.0-Version to Checkpoint C by "Train Difference" with block (0,0,0,0,0,0,0,0,0,0,0,0.2,0.2,0.2,0.2,0,0,0,0,0) (Checkpoint D)

6. Merging the difference between Catloaf and ponyDiffusionV6XL to Checkpoint D by "Train Difference" with block (0,0,0,0,0,0,0,0,0,0,0,0.15,0.15,0.15,0.15,0,0,0,0,0) (Checkpoint E)

7. Merging the difference between copycat-noob Vpred_v1.01 and NoobAI-V-Pred-1.0-Version to Checkpoint E by "Train Difference" with block (0,0,0,0,0,0,0,0,0,0,0,0.15,0.15,0.15,0.15,0,0,0,0,0) (Checkpoint F)

8. Merging style LoRA to Checkpoint F (Checkpoint G)

9. Add v_pred and ztsnr keys to Checkpoint G by noob_v_pencil-XL merge recipes (Calico Cat Tower v-pred v1.0)