# ----- 链接模型 ------------ # cosmos ln -s /kaggle/input/cosmos-predict2-2b-video2world-480p-16fps/cosmos_predict2_2B_video2world_480p_16fps.safetensors ./models/diffusion_models/cosmos_predict2_2B_video2world_480p_16fps.safetensors # sd lora ln -s /kaggle/input/moxinv1/MoXinV1.safetensors ./models/loras/MoXinV1.safetensors ln -s /kaggle/input/blindbox-v1-mix/blindbox_v1_mix.safetensors ./models/loras/blindbox_v1_mix.safetensors ln -s /kaggle/input/dreamshaper-8/dreamshaper_8.safetensors ./models/checkpoints/dreamshaper_8.safetensors #wan t2v ln -s /kaggle/input/wan-2-1-vae/wan_2.1_vae.safetensors ./models/vae/wan_2.1_vae.safetensors ln -s /kaggle/input/umt5-xxl-fp8-e4m3fn-scaled/umt5_xxl_fp8_e4m3fn_scaled.safetensors ./models/text_encoders/umt5_xxl_fp8_e4m3fn_scaled.safetensors ln -s /kaggle/input/wan2-1-t2v-1-3b-fp16/wan2.1_t2v_1.3B_fp16.safetensors ./models/diffusion_models/wan2.1_t2v_1.3B_fp16.safetensors # wan vace #wan2.1_vace_1.3B_fp16.safetensors, 目前仅支持480P的视频，不能支持720P的。 ln -s /kaggle/input/wan2-1-vace-1-3b-fp16/wan2.1_vace_1.3B_fp16.safetensors ./models/diffusion_models/wan2.1_vace_1.3B_fp16.safetensors ln -s /kaggle/input/wan21-causvid-bidirect2-t2v-1-3b-lora-rank32/Wan21_CausVid_bidirect2_T2V_1_3B_lora_rank32.safetensors ./models/loras/Wan21_CausVid_bidirect2_T2V_1_3B_lora_rank32.safetensors ln -s /kaggle/input/umt5-xxl-fp16/umt5_xxl_fp16.safetensors ./models/text_encoders/umt5_xxl_fp16.safetensors #wan2.1_vace_14B_fp16.safetensors, T4 16G 爆了 # wget -c https://huggingface.co/Comfy-Org/Wan_2.1_ComfyUI_repackaged/resolve/main/split_files/diffusion_models/wan2.1_vace_14B_fp16.safetensors -P ./models/diffusion_models # ln -s /kaggle/input/wan21-causvid-14b-t2v-lora-rank32/Wan21_CausVid_14B_T2V_lora_rank32.safetensors ./models/loras/Wan21_CausVid_14B_T2V_lora_rank32.safetensors #Flux kontext ln -s /kaggle/input/flux-ae/ae.safetensors ./models/vae/ae.safetensors ln -s /kaggle/input/clip-l/clip_l.safetensors ./models/text_encoders/clip_l.safetensors ln -s /kaggle/input/t5xxl-fp8-e4m3fn-scaled/t5xxl_fp8_e4m3fn_scaled.safetensors ./models/text_encoders/t5xxl_fp8_e4m3fn_scaled.safetensors # wget -c https://huggingface.co/Comfy-Org/flux1-kontext-dev_ComfyUI/resolve/main/split_files/diffusion_models/flux1-dev-kontext_fp8_scaled.safetensors -P ./models/diffusion_models #Flux dev ln -s /kaggle/input/flux1-dev-fp8/flux1-dev-fp8.safetensors ./models/checkpoints/flux1-dev-fp8.safetensors #nunchaku ln -s /kaggle/input/svdq-int4-r32-flux-1-kontext-dev/svdq-int4_r32-flux.1-kontext-dev.safetensors -P ./models/diffusion_models #SD3 wget -c https://huggingface.co/Comfy-Org/stable-diffusion-3.5-fp8/resolve/main/sd3.5_large_fp8_scaled.safetensors -P /kaggle/working/ ln -s /kaggle/working/sd3.5_large_fp8_scaled.safetensors ./models/checkpoints/sd3.5_large_fp8_scaled.safetensors #omnigen2 # ln -s /kaggle/input/qwen-2-5-vl-fp16/qwen_2.5_vl_fp16.safetensors ./models/text_encoders/qwen_2.5_vl_fp16.safetensors # ln -s /kaggle/input/omnigen2-fp16/omnigen2_fp16.safetensors ./models/diffusion_models/omnigen2_fp16.safetensors # wget -c https://huggingface.co/Comfy-Org/Wan_2.1_ComfyUI_repackaged/blob/main/split_files/diffusion_models/wan2.1_i2v_480p_14B_fp16.safetensors -P ./models/diffusion_models ln -s /kaggle/input/clip-vision-h/clip_vision_h.safetensors ./models/clip_vision/clip_vision_h.safetensors # ---------------- 安装自定义插件节点 ---------------- # 1 ComfyUI-Manager cd custom_nodes git clone https://github.com/ltdrdata/ComfyUI-Manager comfyui-manager cd /kaggle/ComfyUI # 2 安装 nunchaku pip install facexlib pip install onnxruntime pip install insightface pip install https://huggingface.co/mit-han-lab/nunchaku/resolve/main/nunchaku-0.3.1+torch2.6-cp311-cp311-linux_x86_64.whl # wget -c https://huggingface.co/mit-han-lab/nunchaku-flux.1-kontext-dev/resolve/main/svdq-int4_r32-flux.1-kontext-dev.safetensors -P ./models/diffusion_models cd /kaggle/ComfyUI cd custom_nodes git clone https://github.com/mit-han-lab/ComfyUI-nunchaku nunchaku_nodes cd /kaggle/ComfyUI # 3 ComfyUI-GGUF # wget -c https://huggingface.co/city96/FLUX.1-dev-gguf/resolve/main/flux1-dev-Q3_K_S.gguf -P ./models/unet # cd custom_nodes # git clone https://github.com/city96/ComfyUI-GGUF # cd ComfyUI-GGUF # pip install --upgrade gguf # 4 encrypt image cd custom_nodes git clone https://github.com/Vander-Bilt/comfyui-encrypt-image.git cd /kaggle/ComfyUI #5 Prompts Generator cd custom_nodes git clone https://github.com/fairy-root/Flux-Prompt-Generator.git cd /kaggle/ComfyUI # 6 Custom-Scripts cd custom_nodes git clone https://github.com/Vander-Bilt/ComfyUI-Custom-Scripts.git cd /kaggle/ComfyUI # 7 save2hf cd custom_nodes git clone https://github.com/Vander-Bilt/save2hf.git cd save2hf pip install -r requirements.txt cd /kaggle/ComfyUI #class MyHTMLNode: cd custom_nodes git clone https://github.com/Vander-Bilt/MyHTMLNode.git cd /kaggle/ComfyUI REMOTE_PORT="$1" # 传入参数1,2... 这样便于扩展，如果以后用了其他的frp，也好调整。 # frp execution binary only. 这里放frpc执行文件，如果版本有变也好改。配置文件模版，也在项目中, 模版文件名为template_frpc wget -O /kaggle/working/frp_0.54.0_linux_amd64.tar.gz https://github.com/fatedier/frp/releases/download/v0.54.0/frp_0.54.0_linux_amd64.tar.gz tar -xzvf /kaggle/working/frp_0.54.0_linux_amd64.tar.gz -C /kaggle/working cp -p /kaggle/working/frp_0.54.0_linux_amd64/frpc /kaggle/working/frpc cp -p /kaggle/ComfyUI/template_frpc /kaggle/working/frpc.toml FRP_CONFIG_FILE="/kaggle/working/frpc.toml" CHOICE="$1" if [ "$CHOICE" -eq 1 ]; then TARGET_REMOTE_PORT="21663" elif [ "$CHOICE" -eq 2 ]; then TARGET_REMOTE_PORT="21664" else echo "Invalid CHOICE: $CHOICE" echo "Only 1 or 2 are supported." exit 1 # 退出并返回错误码 fi sed -i '' "s/REMOTE_PORT/$TARGET_REMOTE_PORT/" "$FRP_CONFIG_FILE" sleep 2