Cat Tower (NoobAI XL checkpoint)

Cat Tower is anime style NoobAI XL checkpoint.VAE is baked in.v-prediction versions are experimental models.You need to use the webui that support v-prediction.ComfyUI (recommended)Forge Classic (recommended)reForgeForge (maybe)AUTOMATIC1111 (from v1.4, and you need to place the config file in the same folder as the checkpoint)LicenseThis model adheres to the 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 SettingSteps: 25-30CFG scale: 5-7Sampler (vPred): I recommend using "Euler". Using "Euler a" sometimes generates colored skin and objects such as red, blue and so on.Sampler (Epsilon-pred): DPM++ 2M Karras, Euler aScheduler (vPred) : Normal, Simple ("Automatic" produce noisy images.)Use ADetailer as neededPositive Promptmasterpiece, best qualityNegative Promptworst quality, bad quality, low quality, lowres, scan artifacts, jpeg artifacts, sketch, light particlesMerge recipes (ChenkinNoob v1.1)Merging the difference between ChenkinNoob-XL-V0.2 and stable-diffusion-xl-base-1.0, CyberRealistic XL v7.0 and stable-diffusion-xl-base-1.0 to ChenkinNoob-XL-V0.2 by "Perpendicular Component" and "Add Difference" (Checkpoint A)Merging Checkpoint A to ChenkinNoob-XL-V0.2 by "SLERP", alpha = 0.5 (Checkpoint B)Merging Checkpoint B to ChenkinNoob-XL-V0.2 by "Rotate", alignment = 1.0, alpha = 0.0 (Checkpoint C)Replacing the CLIP of Checkpoint C by ChenkinNoob-XL-V0.2 (Checkpoint D)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 ChenkinNoob-XL-V0.2, ratio = 1.0 (Checkpoint E)Merging Checkpoint E to Checkpoint D by "SLERP" with block (0,0,0,0,0,0,0,0,0,0,0,0.7,0.7,0.7,0.7,0,0,0,0,0) (Checkpoint F)Merging Cat Carrier v8.0 to Checkpoint F by "Weighted Sum" with block (0,0,0,0,0,0,0.5,0.5,0.5,0.5,0.5,0.5,0.5,0.5,0.5,0,0,0,0,0) (Checkpoint G)Merging the difference between Cat Carrier v8.0 and Raehoshi illust XL v7.1 to Checkpoint G by "Add Difference" with block (0,0,0,0,0,0,0,0,0,0,0,1,1,0,0,0,0,0,0,0) (Checkpoint H)Merging the difference between LunarPeachMix v2.0 and Illustrious XL v1.1 to Checkpoint H by "Train Difference" with block (0,0,0,0,0,0,0,0,0,0,0,0.3,0.3,0,0,0,0,0,0,0) (Checkpoint I)Merging the difference between copycat-RouWei vpred-v0.42 and Rouwei 0.80 vpred to Checkpoint I by "Train Difference" with block (0,0,0,0,0,0,0,0,0,0,0,0.3,0.3,0,0,0,0,0,0,0) (Checkpoint J)Merging style LoRA to Checkpoint J (Cat Tower ChenkinNoob v1.1)Merge recipes (ChenkinNoob v1.0)Merging the difference between ChenkinNoob-XL-V0.1 and stable-diffusion-xl-base-1.0, CyberRealistic XL v7.0 and stable-diffusion-xl-base-1.0 to ChenkinNoob-XL-V0.1 by "Perpendicular Component" and "Add Difference" (Checkpoint A)Merging Checkpoint A to ChenkinNoob-XL-V0.1 by "SLERP", alpha = 0.5 (Checkpoint B)Merging Checkpoint B to ChenkinNoob-XL-V0.1 by "Rotate", alignment = 1.0, alpha = 0.0 (Checkpoint C)Replacing the CLIP of Checkpoint C by ChenkinNoob-XL-V0.1 (Checkpoint D)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 ChenkinNoob-XL-V0.1, ratio = 1.0 (Checkpoint E)Merging Checkpoint E to Checkpoint D by "SLERP" with block (0,0,0,0,0,0,0,0,0,0,0,0.7,0.7,0.7,0.7,0,0,0,0,0) (Checkpoint F)Merging Cat Carrier v8.0 to Checkpoint F by "Weighted Sum" with block (0,0,0,0,0,0,0.5,0.5,0.5,0.5,0.5,0.5,0.5,0.5,0.5,0,0,0,0,0) (Checkpoint G)Merging the difference between Cat Carrier v8.0 and Raehoshi illust XL v7.1 to Checkpoint G by "Add Difference" with block (0,0,0,0,0,0,0,0,0,0,0,1,1,0,0,0,0,0,0,0) (Checkpoint H)Merging the difference between LunarPeachMix v2.0 and Illustrious XL v1.1 to Checkpoint H by "Train Difference" with block (0,0,0,0,0,0,0,0,0,0,0,0.3,0.3,0,0,0,0,0,0,0) (Checkpoint I)Merging the difference between copycat-RouWei vpred-v0.42 and Rouwei 0.80 vpred to Checkpoint I by "Train Difference" with block (0,0,0,0,0,0,0,0,0,0,0,0.3,0.3,0,0,0,0,0,0,0) (Checkpoint J)Merging style LoRA to Checkpoint J (Cat Tower ChenkinNoob v1.0)Merge recipes (Obsession vPred v2.0)Merging the difference between Obsession (Illustrious-XL) v-pred_v2.0 and stable-diffusion-xl-base-1.0, CyberRealistic XL v5.6 and stable-diffusion-xl-base-1.0 to Obsession (Illustrious-XL) v-pred_v2.0 by "Perpendicular Component" and "Add Difference" (Checkpoint A)Merging Checkpoint A to Obsession (Illustrious-XL) v-pred_v2.0 by "SLERP", alpha = 0.5 (Checkpoint B)Merging Checkpoint B to Obsession (Illustrious-XL) v-pred_v2.0 by "Rotate", alignment = 1.0, alpha = 0.0 (Checkpoint C)Replacing the CLIP of Checkpoint C by Obsession (Illustrious-XL) v-pred_v2.0 (Checkpoint D)Finetuning Checkpoint D with anime style dataset A(2.1k) by full finetune, optimizer = AdamW8bit, scheduler = warmup stable decay, learning rate = 6e-6, scheduler min lr ratio = 0.1, 32 epoch, unet only, gradient accumulation steps = 1, batch size = 4. (Checkpoint E)Finetuning Checkpoint D with anime style dataset B(0.5k) by full finetune, optimizer = AdamW8bit, scheduler = warmup stable decay, learning rate = 3e-6, scheduler min lr ratio = 0.1, 60 epoch, unet only, gradient accumulation steps = 1, batch size = 4. (Checkpoint F)Merging Checkpoint F to Checkpoint E by "SLERP", alpha = 0.5 (Checkpoint G)Merging the difference between LunarPeachMix v2.0 and Illustrious XL v1.1 to Checkpoint G by "Add Difference" with block (0,0,0,0,0,0,0,0,0,0,0,1,1,1,1,0,0,0,0,0) (Checkpoint H)Merging the difference between JANKU v5.0 and RouWei v0.8 epsilon to Checkpoint H by "Add Difference" with block (0,0,0,0,0,0,0,0,0,0,0,1,1,1,1,0,0,0,0,0) (Checkpoint I)Merging the difference between copycat-RouWei vpred-v0.42 and Rouwei 0.80 vpred to Checkpoint I by "Add Difference" with block (0,0,0,0,0,0,0,0,0,0,0,1,1,1,1,0,0,0,0,0) (Checkpoint J)Merging the Checkpoint I by "Clamp", bounds = Checkpoint G, LunarPeachMix v2.0, JANKU v5.0 and copycat-RouWei vpred-v0.42 (Checkpoint K)Add v_pred and ztsnr keys to Checkpoint K by noob_v_pencil-XL merge recipes (Cat Tower vPred v2.0)Merge recipes (Epsilon-pred v1.4)Merging the difference between NoobAI-XL Epsilon-pred v1.1 and stable-diffusion-xl-base-1.0, CyberRealistic XL v5.6 and stable-diffusion-xl-base-1.0 to NoobAI-XL Epsilon-pred v1.1 by "Perpendicular Component" and "Add Difference" (Checkpoint A)Merging Checkpoint A to NoobAI-XL Epsilon-pred v1.1 by "SLERP", alpha = 0.5 (Checkpoint B)Merging Checkpoint B to NoobAI-XL Epsilon-pred v1.1 by "Rotate", alignment = 1.0, alpha = 0.0 (Checkpoint C)Replacing the CLIP of Checkpoint C by NoobAI-XL Epsilon-pred v1.1 (Checkpoint D)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)Merging Cat Carrier v6.0 to Checkpoint E by "Weighted Sum" with block (0,0,0,0,0,0,0.5,0.5,0.5,0.5,0.5,0.5,0.5,0.5,0.5,0,0,0,0,0) (Checkpoint F)Merging the difference between LunarPeachMix v2.0 and Illustrious XL v1.1 to Checkpoint F by "Add Difference" with block (0,0,0,0,0,0,0,0,0,0,0,1,1,1,1,0,0,0,0,0) (Checkpoint G)Merging the difference between JANKU v4.0 and RouWei v0.7 epred to Checkpoint G by "Add Difference" with block (0,0,0,0,0,0,0,0,0,0,0,1,1,1,1,0,0,0,0,0) (Checkpoint H)Merging the difference between copycat-RouWei vpred-v0.42 and Rouwei 0.80 vpred to Checkpoint H by "Add Difference" with block (0,0,0,0,0,0,0,0,0,0,0,1,1,1,1,0,0,0,0,0) (Checkpoint I)Merging the Checkpoint I by "Clamp", bounds = Checkpoint F, LunarPeachMix v2.0, JANKU v4.0 and copycat-RouWei vpred-v0.42 (Cat Tower Epsilon-pred v1.4)Merge recipes (vPred v1.8)Merging the difference between NoobAI-V-Pred-1.0-Version and stable-diffusion-xl-base-1.0, CyberRealistic XL v5.6 and stable-diffusion-xl-base-1.0 to NoobAI-V-Pred-1.0-Version by "Perpendicular Component" and "Add Difference" (Checkpoint A)Merging Checkpoint A to NoobAI-V-Pred-1.0-Version by "SLERP", alpha = 0.5 (Checkpoint B)Merging Checkpoint B to NoobAI-V-Pred-1.0-Version by "Rotate", alignment = 1.0, alpha = 0.0 (Checkpoint C)Replacing the CLIP of Checkpoint C by NoobAI-V-Pred-1.0-Version (Checkpoint D)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)Merging Cat Carrier v6.0 to Checkpoint E by "Weighted Sum" with block (0,0,0,0,0,0,0.5,0.5,0.5,0.5,0.5,0.5,0.5,0.5,0.5,0,0,0,0,0) (Checkpoint F)Merging the difference between LunarPeachMix v2.0 and Illustrious XL v1.1 to Checkpoint F by "Add Difference" with block (0,0,0,0,0,0,0,0,0,0,0,1,1,1,1,0,0,0,0,0) (Checkpoint G)Merging the difference between JANKU v4.0 and RouWei v0.7 epred to Checkpoint G by "Add Difference" with block (0,0,0,0,0,0,0,0,0,0,0,1,1,1,1,0,0,0,0,0) (Checkpoint H)Merging the difference between copycat-RouWei vpred-v0.42 and Rouwei 0.80 vpred to Checkpoint H by "Add Difference" with block (0,0,0,0,0,0,0,0,0,0,0,1,1,1,1,0,0,0,0,0) (Checkpoint I)Merging the Checkpoint I by "Clamp", bounds = Checkpoint F, LunarPeachMix v2.0, JANKU v4.0 and copycat-RouWei vpred-v0.42 (Checkpoint J)Add v_pred and ztsnr keys to Checkpoint J by noob_v_pencil-XL merge recipes (Cat Tower vPred v1.8)Merge recipes (Epsilon-pred v1.3)Merging Cat Tower Epsilon-pred v1.2 and Cat Tower v-pred v1.7 by "DARE-TIES" (Cat Tower Epsilon-pred v1.3)Merge recipes (vPred v1.7)Merging the difference between CyberRealistic XL v5 and stable-diffusion-xl-base-1.0 to NoobAI-V-Pred-1.0-Version by "Add Difference" (Checkpoint A)Replacing the CLIP of Checkpoint A by NoobAI-V-Pred-1.0-Version (Checkpoint B)Merging Cat Carrier v3.0 to Checkpoint B by "Weighted Sum" with block (0,0.2,0.2,0.2,0.2,0.2,0.2,0.2,0.2,0.4,0.4,0.4,0.4,0.4,0.4,0.0,0,0,0,0) (Checkpoint C)Merging Cat Tower vpred v1.5 to Checkpoint C by "Add Cosine B" with block (0,0.2,0.2,0.2,0.2,0.2,0.2,0.2,0.2,0.2,0.2,0.2,0.2,0.2,0.2,0,0,0,0,0) (Checkpoint D)Merging Ca