[Bug]: 生成图片的时候经常被kill，但是gpu，显存，内存都没有达到

[MRYO-ga]

Is there an existing issue for this?

• I have searched the existing issues and checked the recent builds/commits of both this extension and the webui

Is EasyPhoto the latest version?

• I have updated EasyPhoto to the latest version and the bug still exists.

What happened?

生成图片的时候经常被kill，但是gpu，显存，内存都没有达到

Steps to reproduce the problem

选择模板，生成图像

What should have happened?

不会发生kill

Commit where the problem happens

webui: bef51aed032c0aaa5cfd80445bc4cf0d85b408b5
EastPhoto: 78696eb

What browsers do you use to access the UI ?

No response

Command Line Arguments

autodl平台：
参数：
cd /root/stable-diffusion-webui && SD_WEBUI_CACHE_FILE=/root/autodl-tmp/cache.json IGNORE_CMD_ARGS_ERRORS=true PATH=/root/miniconda3/envs/xl_env/bin:$PATH TF_CPP_MIN_LOG_LEVEL=0 HF_ENDPOINT=https://hf-mirror.com python  launch.py --ckpt-dir /root/autodl-tmp/models/Stable-diffusion --embeddings-dir /root/autodl-tmp/models/embeddings --hypernetwork-dir /root/autodl-tmp/models/hypernetwork --lora-dir /root/autodl-tmp/models/Lora --vae-dir /root/autodl-tmp/models/VAE --controlnet-dir /root/autodl-tmp/models/ControlNet --controlnet-annotator-models-path /root/autodl-tmp/models/ControlNet_annotator --dreambooth-models-path=/root/autodl-tmp/models/dreambooth --codeformer-models-path /root/autodl-tmp/models/Codeformer --gfpgan-models-path /root/autodl-tmp/models/GFPGAN --esrgan-models-path /root/autodl-tmp/models/ESRGAN --bsrgan-models-path /root/autodl-tmp/models/BSRGAN --realesrgan-models-path /root/autodl-tmp/models/RealESRGAN --scunet-models-path /root/autodl-tmp/models/ScuNET --swinir-models-path /root/autodl-tmp/models/SwinIR --ldsr-models-path /root/autodl-tmp/models/LDSR --port=6006 --theme dark --skip-install --xformers --disable-safe-unpickle --enable-insecure-extension-access --no-half-vae --disable-nan-check --max-batch-count=16 --api

List of enabled extensions

无

Console logs

2024-03-16 00:47:06,609 - EasyPhoto - ControlNet unit number: 8
2024-03-16 00:47:06,609 - EasyPhoto - Found 1 user id(s), but only 0 image prompt(s) for IP-Adapter Control. Use the reference image corresponding to the user instead.
2024-03-16 00:47:07,510 - EasyPhoto - Start templates and user_ids preprocess.
Loading VAE weights specified in settings: /root/autodl-tmp/models/VAE/vae-ft-mse-840000-ema-pruned.ckpt
Applying attention optimization: xformers... done.
VAE weights loaded.
2024-03-16 00:47:10,932 - EasyPhoto - 
            Start Generate template                 : 1;
            user_ids                                : ('mayi', 'none', 'none', 'none', 'none');
            sd_model_checkpoint                     : majicMIXrealisticV7.safetensors;
            input_prompts                           : ['easyphoto_face, easyphoto, 1person, <lora:mayi:0.9>, <lora:FilmVelvia3:0.65>, masterpiece, beauty<lora:lcm_lora_sd15:0.80>, ', 'none', 'none', 'none', 'none'];
            before_face_fusion_ratio                : 0.6;
            after_face_fusion_ratio                 : 0.6;
            first_diffusion_steps                   : 12;
            first_denoising_strength                : 0.5;
            second_diffusion_steps                  : 8;
            second_denoising_strength               : 0.3;
            seed                                    : 21843
            crop_face_preprocess                    : True
            apply_face_fusion_before                : True
            apply_face_fusion_after                 : True
            color_shift_middle                      : True
            color_shift_last                        : True
            super_resolution                        : True
            super_resolution_method                 : gpen
            display_score                           : True
            background_restore                      : False
            background_restore_denoising_strength   : 0.35
            makeup_transfer                         : False
            makeup_transfer_ratio                   : 0.5
            skin_retouching_bool                    : False
            face_shape_match                        : False
            id_control                              : True
            id_control_method                       : IP-Adapter Face
            ipa_weight                              : 0.3
            instantid_id_weight                     : 0.5
            instantid_ipa_weight                    : 0.5
            instantid_image_path                    : None
            ipa_image_path                          : None
            ref_mode_choose                         : Infer with User Lora
            no_user_lora_mode                       : IP-Adapter Face
            ipa_only_weight                         : 0.6
            ipa_only_image_path                     : None
            instantid_only_id_weight                : 0.5
            instantid_only_ipa_weight               : 0.5
            instantid_only_image_path               : None
        
2024-03-16 00:47:11,440 - EasyPhoto - Start Image resize to 512.0.
2024-03-16 00:47:11,460 - EasyPhoto - Start face detect.
2024-03-16 00:47:14,346 - EasyPhoto - Start First diffusion.
[-] ADetailer: img2img inpainting detected. adetailer disabled.
Warning: field include_init_images in API payload not found in <modules.processing.StableDiffusionProcessingImg2Img object at 0x7f57f2352950>.
Warning: field infotext in API payload not found in <modules.processing.StableDiffusionProcessingImg2Img object at 0x7f57f2352950>.
2024-03-16 00:47:15,496 - ControlNet - INFO - unit_separate = False, style_align = False
2024-03-16 00:47:15,519 - ControlNet - WARNING - [canny.processor_res] Invalid value(-1), using default value 512.
2024-03-16 00:47:16,256 - ControlNet - INFO - Loading model: control_v11p_sd15_canny [d14c016b]
2024-03-16 00:47:16,994 - ControlNet - INFO - Loaded state_dict from [/root/autodl-tmp/models/ControlNet/control_v11p_sd15_canny.pth]
2024-03-16 00:47:16,994 - ControlNet - INFO - controlnet_default_config
2024-03-16 00:47:21,552 - ControlNet - INFO - ControlNet model control_v11p_sd15_canny [d14c016b](ControlModelType.ControlNet) loaded.
2024-03-16 00:47:21,604 - ControlNet - INFO - Using preprocessor: canny
2024-03-16 00:47:21,604 - ControlNet - INFO - preprocessor resolution = 512
2024-03-16 00:47:21,610 - ControlNet - WARNING - [openpose_full.processor_res] Invalid value(-1), using default value 512.
2024-03-16 00:47:22,346 - ControlNet - INFO - Loading model: control_v11p_sd15_openpose [cab727d4]
2024-03-16 00:47:23,203 - ControlNet - INFO - Loaded state_dict from [/root/autodl-tmp/models/ControlNet/control_v11p_sd15_openpose.pth]
2024-03-16 00:47:23,204 - ControlNet - INFO - controlnet_default_config
2024-03-16 00:47:27,785 - ControlNet - INFO - ControlNet model control_v11p_sd15_openpose [cab727d4](ControlModelType.ControlNet) loaded.
2024-03-16 00:47:27,826 - ControlNet - INFO - Using preprocessor: openpose_full
2024-03-16 00:47:27,826 - ControlNet - INFO - preprocessor resolution = 512
2024-03-16 00:47:29,839 - ControlNet - INFO - Loading model: control_sd15_random_color [3afa6fba]
Killed

Additional information

No response

[YASSA202020]

README.md