import os as OS import inspect as Inspect from collections import namedtuple as NamedTuple import os.path as Pathing import sys as Environment import cv2 as Vision # import trimesh as Meshing import gradio as Framework import mmcv as Vision_MM # from diffusers import StableDiffusionPipeline as Pipeline import numpy as Number import torch as Engine import anime_3dkenburns.kenburns_effect as Anime3D_KenBurns from anime_3dkenburns.kenburns_effect import KenBurnsPipeline as Pipeline import Web_UI.Common.Utility_Sampler as Sampler import Web_UI.Common.Utility_Style as Style # import Utility_Model as Model # # from Web_UI.Common.Utility_Model import Model as Model from ..Common.Utility_Model import Model as Model #region Script # # Debug # print ( F"{OS.getcwd()=}" ) #endregion Script # Text → Text Tab # Field PlaceHolder_Label_Camera=F"#### Horizon:　(Original:　), Vertical:　(Original:　)" Pipeline=Pipeline(F"configs/3dkenburns.yaml") Limit_Minimum=400 # It is better that does not less than 60% of original Width | Height Limit_Maxmum=4000 Path_List_Sample=F"examples" # Pipeline=Pipeline() # Configuration_AIS = Pipeline.cfg Device = Pipeline.device # Global Carrier Image_Carrier = None Image_Tensor_Carrier = None Instance_Carrier = None Image_Intermediate_Carrier = None Configuration_Argument_Carrier=None IsIntermediateResultSaving = False IsAllActivatedControlDisplayed=False Ratio_Crop_Camera_Start_Deault=0.6 Ratio_Crop_Camera_End_Deault=0.6 Ratio_Crop_Minimum=0 Ratio_Crop_AutoZoomed=0.97 Mask_α_Default=0.7 Progress_Default=F"Display the Progress" IsVerbosing_Global=False Height_Global=None Width_Global=None IsStartingCameraSetting_Global=True Location_Point=NamedTuple(F"Point",F"Horizon Vertical") Camera_Start_Global=None Camera_End_Global=None Ratio_Global=None Configuration_Global=None List_Frame_Global=None Ratio_Resize_Global=None Page_Global = None # Common control for multiple tabs Create_CheckPoint_ComboBox=lambda:Framework.Dropdown\ ( choices = Model.List_CheckPointTile (), label = F"Stable Diffusion Check-Point", # elem_id = F"ComboBox_CheckPoint", ) # Combo Box无法自行完成对File System中的Content的刷新、自行检查，需要手动处理 Create_RelaunchModelList_Button=lambda:Framework.Button\ ( # ！考虑改为Icon的Button F"Relaunch", # show_label = False, # lines = 2, # As known as Height # placeholder = "Negative prompt | input here (Ctrl | Alt + Enter or Button \"Generate\" to work )", ) Create_CheckPointLoadingStatus_Label=lambda:Framework.Label\ ( # choices = Model.List_CheckPointTile (), label = F"Check-Point Loading Status", value =F"None Model has been loaded." # 相当于default # show_label = False, # visible = False, # elem_id = F"ComboBox_CheckPoint", ) Create_Main_Textbox=lambda:Framework.Textbox\ ( label = "Prompt | Input", show_label = False, lines = 3, # As known as Height placeholder = "Prompt | input here (Ctrl | Alt + Enter or Button \"Generate\" to work )", # 设置上限+当前字符量额统计、限制 # Debug # value = F"Bill Gates rides on a horse." ) Create_Negative_TextBox=lambda:Framework.Textbox\ ( label = "Negative prompt | input", show_label = False, lines = 2, # As known as Height placeholder = "Negative prompt | input here (Ctrl | Alt + Enter or Button \"Generate\" to work )", # !Need to be classified ) Create_Interrupt_Button=lambda:Framework.Button\ ( F"Interrogate CLIP", # show_label = False, # lines = 2, # As known as Height # placeholder = "Negative prompt | input here (Ctrl | Alt + Enter or Button \"Generate\" to work )", ) Create_Skip_Button=lambda:Framework.Button\ ( F"Interrogate CLIP", # show_label = False, # lines = 2, # As known as Height # placeholder = "Negative prompt | input here (Ctrl | Alt + Enter or Button \"Generate\" to work )", ) Create_Generate_Button=lambda:Framework.Button\ ( F"Generate", # show_label = False, # lines = 2, # As known as Height # placeholder = "Negative prompt | input here (Ctrl | Alt + Enter or Button \"Generate\" to work )", variant = "primary", ) # Pending Create_Clear_Button=lambda:Framework.Button\ ( F"Clear Prompted | Input", # show_label = False, # lines = 2, # As known as Height # placeholder = "Negative prompt | input here (Ctrl | Alt + Enter or Button \"Generate\" to work )", # variant = "primary", ) Create_Style_ComboBox=lambda:Framework.Dropdown\ ( choices = Style.List_Style () , value = [], # 相当于default multiselect = True, label = F"Style List", ) # Control for Image → Image Tab Create_Normal_Input_ImageBox=lambda:Framework.Image \ ( label = "Image → Image", ) Create_Sketch_Input_ImageBox=lambda:Framework.Image \ ( label = "Sketch", ) Create_Inpaint_Input_ImageBox=lambda:Framework.Image \ ( label = "Inpaint", ) Create_InpaintSketch_Input_ImageBox=lambda:Framework.Image \ ( label = "Inpaint Sketch", ) Create_InpaintUpload_Input_ImageBox=lambda:Framework.Image \ ( label = "Inpaint Upload", ) Create_Batch_Input_ImageBox=lambda:Framework.Image \ ( label = "Batch", ) Create_InterrogateCLIP_Button=lambda:Framework.Button\ ( F"Interrogate CLIP", # show_label = False, # lines = 2, # As known as Height # placeholder = "Negative prompt | input here (Ctrl | Alt + Enter or Button \"Generate\" to work )", ) Create_Interrogate_DeepBooru_Button=lambda:Framework.Button\ ( F"Interrogate DeepBooru", # show_label = False, # lines = 2, # As known as Height # placeholder = "Negative prompt | input here (Ctrl | Alt + Enter or Button \"Generate\" to work )", ) # Pending to check Create_ResizeMode_RadioGroup=lambda:Framework.Radio\ ( choices = [ F"Just resize" , F"Crop and resize" , F"Resize and fill" , F"Just resize (Latent up-scale)" , ], # value = [], # 相当于default # multiselect = True, label = F"Resize mode", ) Create_MaskBlur_Slider=lambda:Framework.Slider\ ( # choices = Sampler.List_Sampler () , value = 4, # 相当于default minimum = 0, maximum = 64, step = 1, # multiselect = True, label = F"Mask blur", ) Create_MaskTransparency=lambda:Framework.Slider\ ( # choices = Sampler.List_Sampler () , value = 0, # 相当于default minimum = 0, maximum = 100, step = 1, # multiselect = True, label = F"Mask transparency", ) Create_MaskMode_RadioGroup=lambda:Framework.Radio\ ( choices = [ F"Inpaint masked", F"Inpaint \033[1mnot\033[0m masked" , # !Need to be optimised ], # value = [], # 相当于default # multiselect = True, label = F"Mask mode", ) Create_MaskedContent_RadioGroup=lambda:Framework.Radio\ ( choices = [ F"Fill", F"Original", F"Latent noise", F"Latent nothing", ], # value = [], # 相当于default # multiselect = True, label = F"Masked content", ) Create_InpaintArea_RadioGroup=lambda:Framework.Radio\ ( choices = [ F"Whole picture", F"Only Masked", ], # value = [], # 相当于default # multiselect = True, label = F"Inpaint area", ) Create_OnlyMaskedPadding_Pixel_Slider=lambda:Framework.Slider\ ( # choices = Style.List_Style () , value = 32, # 相当于default minimum = 0, maximum = 256, step = 4, # multiselect = True, label = F"Only masked padding (Pixel)", ) Create_Sampler_ComboBox=lambda:Framework.Dropdown\ ( choices = Sampler.List_Sampler () , value = [], # 相当于default # multiselect = True, label = F"Sampler | Sampling Method", ) Create_SamplingStep_Slider=lambda:Framework.Slider\ ( # choices = Style.List_Style () , value = 20, # 相当于default minimum = 1, maximum = 150, step = 1, # multiselect = True, label = F"Sampling Step", ) Create_CheckBox_Group=lambda :Framework.CheckboxGroup \ ( choices = [ F"Restore Face" , F"Tiling", F"High Resolution Fix", ], value = F"", ) Create_RestoreFace_CheckBox=lambda:Framework.Checkbox\ ( # choices = Style.List_Style () , # value = 20, # 相当于default # multiselect = True, label = F"Restore Face", ) Create_Tiling_CheckBox=lambda:Framework.Checkbox\ ( # choices = Style.List_Style () , # value = 20, # 相当于default # multiselect = True, label = F"Tiling", ) Create_HighResolutionFix_CheckBox=lambda:Framework.Checkbox\ ( # choices = Style.List_Style () , # value = 20, # 相当于default # multiselect = True, label = F"High Resolution Fix", ) Create_Step_CheckBox=lambda Name_In,Value_Default_In,Tip_In=None,IsVisible_In=True:Framework.Checkbox\ ( # choices = Style.List_Style () , # value = 20, # 相当于default # multiselect = True, label = Name_In, value = Value_Default_In, info = Tip_In, visible = IsVisible_In, ) Create_Width_Slider=lambda:Framework.Slider\ ( # choices = Style.List_Style () , value = 512, # 相当于default minimum = 64, maximum = 2048, step = 8, # multiselect = True, label = F"Width", ) Create_Height_Slider=lambda:Framework.Slider\ ( # choices = Style.List_Style () , value = 512, # 相当于default minimum = 64, maximum = 2048, step = 8, # multiselect = True, label = F"Height", ) # Or called Switch Create_Swap_Button=lambda:Framework.Button\ ( # choices = Style.List_Style () , # value = 512, # 相当于default # multiselect = True, F"Swap", ) Create_BatchCount_Slider=lambda:Framework.Slider\ ( # choices = Style.List_Style () , value = 1, # 相当于default minimum = 1, # maximum = 2048, step = 1, # multiselect = True, label = F"Batch Count", ) Create_BatchSize_Slider=lambda:Framework.Slider\ ( # choices = Style.List_Style () , value = 1, # 相当于default minimum = 1, maximum = 8, step = 1, # multiselect = True, label = F"Batch Size", ) Create_CFG_Scale_Slider=lambda:Framework.Slider\ ( # choices = Style.List_Style () , value = 7.0, # 相当于default minimum = 1.0, maximum = 30.0, step = 0.5, # multiselect = True, label = F"CFG Scale", ) Create_DenoisingStrength_Slider=lambda:Framework.Slider\ ( # choices = Style.List_Style () , value = 0.75, # 相当于default minimum = 0, maximum = 1, step = 0.01, # multiselect = True, label = F"Denoising strength", ) Create_Seed_TextBox=lambda:Framework.Textbox\ ( # choices = Style.List_Style () , value = F"-1", # 相当于default # multiselect = True, label = F"Seed", ) Create_AdditionalScript_ComboBox=lambda:Framework.Dropdown\ ( # choices = Style.List_Style () , value = F"None", # 相当于default # multiselect = True, label = F"Additional Script", ) Create_Output_BatchedImageBox=lambda \ Name_In=F"Output",\ Type_In=F"pil",\ Number_Column=None,\ Object_Fit_In=F"contain", \ IsPreviewMode_In=False,\ IsLabelDisplayed_In=False,\ Height_In=F"auto",\ :\ Framework.Gallery\ ( # choices = Style.List_Style () , # value = F"None", # 相当于default # multiselect = True, label = Name_In, show_label = IsLabelDisplayed_In, # 的确不好看，更像是个Holder Tool Tip # type = Type_In, # select=1, # _css= "overflow-x: scroll; overflow-y: hidden;", columns = Number_Column, object_fit = Object_Fit_In , preview = IsPreviewMode_In, height =Height_In, ) Create_Output_ImageBox_Single=lambda:Framework.Image\ ( # choices = Style.List_Style () , # value = F"None", # 相当于default # multiselect = True, label = F"Output", show_label = False, # 的确不好看，更像是个Holder Tool Tip type = "pil", ) Create_OpenInExplorer_Button=lambda:Framework.Button\ ( # choices = Style.List_Style () , # value = F"None", # 相当于default # multiselect = True, F"Open in ...", ) Create_OriginalSaveLink_Button=lambda:Framework.Button\ ( # choices = Style.List_Style () , # value = F"None", # 相当于default # multiselect = True, F"Save Link", ) Create_ZipSaveLink_Button=lambda:Framework.Button\ ( # choices = Style.List_Style () , # value = F"None", # 相当于default # multiselect = True, label = F"Zip Link", ) Create_ApplySetting_Button=lambda:Framework.Button\ ( # choices = Style.List_Style () , # value = F"None", # 相当于default # multiselect = True, F"Apply Setting", ) Create_Restart_Button=lambda:Framework.Button\ ( # choices = Style.List_Style () , # value = F"None", # 相当于default # multiselect = True, F"Restart", ) #region Function | Method #region Event Handler def DoOnButtonSubmit_AIS_Instance_Segmentation_Clicking (Image_In:Number.ndarray) : """ The 1st Process Args: Image_In (): Returns: """ # Pre-Processing # Pre-Setting Instance_Processing = None Image_Processing = None if isinstance ( Image_In , str ) : Image_Processing = Vision_MM.imread ( Image_In ) else: Image_Processing=Image_In with Engine.no_grad () : if Instance_Processing is None : Instance_Processing , Image_Return = Pipeline.run_instance_segmentation \ ( Image_Processing , scale_down_to_maxsize = False , ) else : pass Anime3D_KenBurns.torch_gc () global Image_Carrier Image_Carrier = Image_Processing global Instance_Carrier Instance_Carrier = Instance_Processing return Image_Return def DoOnButtonSubmit_AIS_Infer_Disparity_Depth_Estimation_Clicking () : """ The 2nd Process Returns: """ # Pre-Processing # Pre-Declaring global Image_Carrier global Instance_Carrier with Engine.no_grad () : Image_Processing = Anime3D_KenBurns.scaledown_maxsize\ ( Image_Carrier , Configuration_AIS.max_size, ) Instance_Carrier.resize \ ( Image_Processing.shape [ 0 ] , Image_Processing.shape [ 1 ] , ) Configuration_AIS.int_height , Configuration_AIS.int_width = Image_Processing.shape [ : 2 ] Image_Tensor=Engine.FloatTensor\ ( Number.ascontiguousarray\ ( Image_Processing\ .transpose(2,0,1)\ [None,:,:,:]\ .astype(Number.float32)\ *(1.0/255.0), ), )\ .to(Device) Configuration_Argument:Anime3D_KenBurns.KenBurnsConfig=Configuration_AIS.copy() if Instance_Carrier is None : # ！此处的Instance_Carrier指向是否为Global需要测试 Instance_Carrier , Image_Processing = Pipeline.run_instance_segmentation\ ( Image_Processing , scale_down_to_maxsize = False, ) Anime3D_KenBurns.torch_gc () else: pass # Coarse Depth if Image_Tensor is None : Image_Tensor = Engine.FloatTensor\ ( Number.ascontiguousarray\ ( Image_Processing.transpose ( 2 , 0 , 1 )\ [ None , : , : , : ]\ .astype ( Number.float32 )\ * (1.0 / 255.0) ) )\ .to ( Device ) else: pass Disparity_Return = Pipeline._depth_est ( Image_Tensor , Image_Processing ) Anime3D_KenBurns.torch_gc () # Debug print(F"Disparity generated successfully") # # global Image_Intermediate_Carrier # # Image_Intermediate_Carrier=Disparity_Return # global Image_Tensor_Carrier Image_Tensor_Carrier=Disparity_Return # # global Image_Carrier Image_Carrier=Image_Processing # global Disparity_Carrier Disparity_Carrier=Disparity_Return # # global Instance_Carrier # Instance_Carrier=Instance_Processing # global Configuration_Argument_Carrier Configuration_Argument_Carrier=Configuration_Argument return Disparity_Return # ？和前1函数合并，共用Depth Coarse的名，保持和后面的一致 def DoOnButtonSubmit_AIS_Infer_Disparity_Depth_Coarse_Clicking () : """ The 3rd Process Returns: """ with Engine.no_grad () : Image_With_Depth_Coarse_Return = Anime3D_KenBurns.colorize_depth\ ( Disparity_Carrier\ .cpu ()\ .numpy () , inverse = True , rgb2bgr = True , cmap = 'magma_r', ) if Configuration_Argument_Carrier is not None : Configuration_Argument_Carrier.stage_depth_coarse = Image_With_Depth_Coarse_Return else: pass if IsIntermediateResultSaving : # To Do pass else: pass Anime3D_KenBurns.torch_gc () # global Image_Intermediate_Carrier # Image_Intermediate_Carrier=Image_With_Depth_Coarse_Return return Image_With_Depth_Coarse_Return def DoOnButtonSubmit_AIS_Infer_Disparity_Depth_Adjusted_Depth_Clicking () : """ The 4th Process Returns: """ # Pre-Processing # Pre-Declaring global Disparity_Carrier with Engine.no_grad () : # Adjust Depth if Configuration_AIS.detector == F"maskrcnn" : Disparity_Return = Anime3D_KenBurns.depth_adjustment_maskrcnn\ ( Instance_Carrier , Disparity_Carrier , Image_Tensor_Carrier, ) else : Disparity_Return = Anime3D_KenBurns.depth_adjustment_animesseg\ ( Instance_Carrier , Disparity_Carrier , Image_Tensor_Carrier, ) Image_With_Depth_Adjusted_Return = Anime3D_KenBurns.colorize_depth\ ( Disparity_Return\ .cpu ()\ .numpy () , inverse = True , rgb2bgr = True , cmap = 'magma_r', ) if Configuration_Argument_Carrier is not None : Configuration_Argument_Carrier.stage_depth_adjusted = Image_With_Depth_Adjusted_Return else: pass if IsIntermediateResultSaving : # To Do pass else: pass Anime3D_KenBurns.torch_gc () # global Disparity_Carrier Disparity_Carrier=Disparity_Return return\ [ Image_With_Depth_Adjusted_Return, Disparity_Return, ] def DoOnButtonSubmit_AIS_Adjustment_Anime_Segmentation_TabClicking () : """ The 5th Process Returns: """ # Pre-Processing # Pre-Declaring global Disparity_Carrier with Engine.no_grad () : # Final Depth if Configuration_AIS.default_depth_refine : Disparity_Return = Pipeline.refine_depth\ ( Image_Tensor_Carrier , Disparity_Carrier, ) elif Configuration_AIS.refine_crf : Disparity_Return = Pipeline.refine_depth_crf\ ( Image_Carrier , Disparity_Carrier , Instance_Carrier, ) else: pass Image_With_Depth_Final_Return = Anime3D_KenBurns.colorize_depth\ ( Disparity_Return\ .cpu ()\ .numpy () , inverse = True , rgb2bgr = True , cmap = 'magma_r', ) if Configuration_Argument_Carrier is not None : Configuration_Argument_Carrier.stage_depth_final = Image_With_Depth_Final_Return else: pass if IsIntermediateResultSaving : # To Do pass else: pass Anime3D_KenBurns.torch_gc () # global Disparity_Carrier Disparity_Carrier=Disparity_Return return\ [ Image_With_Depth_Final_Return, Disparity_Return, ] def DoOnButton_Run_With_Steps_Stepping_TabClicking\ ( Image_In:Number.ndarray, IsBoundingBox_Required_In, IsInstanceMask_Required_In, IsInstanceContour_Required_In, IsTag_List_Required_In, Mask_α_In, ) : global Configuration_Global IsVerbosing_In=IsVerbosing_Global # Debug print(F"{IsVerbosing_In=} in {Inspect.currentframe().f_code.co_name}()") List_Return=[] # Carrier_Configuration = Pipeline.generate_kenburns_config\ ( img = Image_In , verbose = IsVerbosing_In , # savep = None, # Gradio的File Control提供了Download，此Argument意义不大 ) # Debug print(F"Scale Down in {Inspect.currentframe().f_code.co_name} after Configure Ken Burns") print(F"{Image_In.shape=}") print(F"{Carrier_Configuration.int_width=},{Carrier_Configuration.int_height=}") print(F"Ratio: {Carrier_Configuration.int_width/Carrier_Configuration.int_height}") print(F"Ratio Global | Original as contrast: {Ratio_Global=}") Image_Return_After_2nd_Process = Carrier_Configuration.instances.draw_instances\ ( img=Image_In , draw_bbox = IsBoundingBox_Required_In, draw_ins_mask = IsInstanceMask_Required_In, draw_ins_contour = IsInstanceContour_Required_In, # ！需要保持None，Empty不行 # draw_indices =\ # [ # # List of indice # ], draw_tags = IsTag_List_Required_In , mask_alpha=Mask_α_In, ) # Debug print(F"Scale Down in {Inspect.currentframe().f_code.co_name} after Draw Instance") print(F"{Image_In.shape=}") print(F"{Carrier_Configuration.int_width=},{Carrier_Configuration.int_height=}") print(F"Ratio: {Carrier_Configuration.int_width/Carrier_Configuration.int_height}") print(F"Ratio Global | Original as contrast: {Ratio_Global=}") List_Return.append(Image_Return_After_2nd_Process ) # ！此处为了抵消AIS代码中预先对Image进行的BGR→RGB，进行了1次转换 Image_Return_Aftewr_3rd_Process=Vision.cvtColor(Carrier_Configuration.stage_depth_coarse, Vision.COLOR_BGR2RGB) # List_Return.append(Image_Return_Aftewr_3rd_Process) # ！此处为了抵消AIS代码中预先对Image进行的BGR→RGB，进行了1次转换 Image_Return_Aftewr_4th_Process=Vision.cvtColor(Carrier_Configuration.stage_depth_adjusted, Vision.COLOR_BGR2RGB) # List_Return.append(Image_Return_Aftewr_4th_Process) # ！此处为了抵消AIS代码中预先对Image进行的BGR→RGB，进行了1次转换 Image_Return_Aftewr_5th_Process=Vision.cvtColor(Carrier_Configuration.stage_depth_final, Vision.COLOR_BGR2RGB) # List_Return.append(Image_Return_Aftewr_5th_Process) # Generate High Light Crops for Auto-zoomed Method List_Return.append(HighLight_Crop_AutoZoom(Image_In,Carrier_Configuration)) # List_Return.append(Image_In) # Update Control status List_Return.append(Framework.update(interactive = True)) # # # Update Progress List_Return.append(F"Instances generated successfully.") # Post-Processing Configuration_Global=Carrier_Configuration return List_Return def Adjust_Camera_Location ( Location_Camera_Original_In:Location_Point , Difference_Width_Original_In =0,Ratio_Scaling_In=1) : # ！需要使用更简洁的方式设定非null的default # Location_Camera_Original_Return:Location_Point=None # Location_Central_Original=Location_Point(Approximate_To_Integer( Width_Global * Ratio_Resize_Global/2),Approximate_To_Integer( Height_Global * Ratio_Resize_Global/2 ) ) # Difference_Height_Original=Approximate_To_Integer(Difference_Width_Original_In/Ratio_Global) # Horizon_Carrier=None Horizon_Carrier=Approximate_To_Integer(Location_Camera_Original_In.Horizon*Ratio_Scaling_In) # Vertical_Carrier=None Vertical_Carrier=Approximate_To_Integer(Location_Camera_Original_In.Vertical*Ratio_Scaling_In) # if(Location_Camera_Original_In.HorizonSize_Original.Horizon: Crop_Processing = Size_Original else: Crop_Processing=Crop_In # # Debug # print(F"{Inspect.currentframe().f_code.co_name}()") # print(F"{Limit_Lower=}" ) # if(Crop_Processing.HorizonHeight_Original\ else False Ratio_Global=Width_Original/Height_Original # Save Ratio of Image Input as Single # Processing: 计算Down Scale后的Ratio、Pixel坐标 if(IsHorizonRectangle): Ratio_Resize=Width_Original/Configuration_AIS.max_size # ！考虑更换获取来源，更统一一些，而不是当前的Legacy的Pipeline的残留来源 Width_Global=Configuration_AIS.max_size # 直接获取；或Original通过Ratio计算 Height_Global=Approximate_To_Integer(Height_Original/Ratio_Resize) # 直接Original通过Ratio计算；或给定值通过长宽Ratio计算 else: Ratio_Resize=Height_Original/Configuration_AIS.max_size # ！考虑更换获取来源，更统一一些，而不是当前的Legacy的Pipeline的残留来源 Width_Global=Approximate_To_Integer(Width_Original/Ratio_Resize) # 直接Original通过Ratio计算；或给定值通过长宽Ratio计算 Height_Global=Configuration_AIS.max_size # 直接获取；或Original通过Ratio计算 # Debug print ( F"After generating Down Scaled Image in {Inspect.currentframe ().f_code.co_name}()" ) print ( F"{Width_Global=}, {Height_Global=}" ) print ( F"{Width_Original=}, {Height_Original=}" ) print ( F"Ratio: {Width_Global / Height_Global}" ) print ( F"Ratio Global | Original as contrast: {Ratio_Global=}" ) Width_Preferred=Approximate_To_Integer( Width_Global * Ratio_Crop_Camera_Start_In ) Height_Preferred=Approximate_To_Integer( Height_Global * Ratio_Crop_Camera_End_In ) List_Return.append(Ratio_Crop_Camera_Start_Deault) List_Return.append(Width_Preferred) List_Return.append(Height_Preferred) # List_Return.append(Ratio_Crop_Camera_End_Deault) List_Return.append(Width_Preferred) List_Return.append(Height_Preferred) # List_Return.append ( Image_In ) List_Return.append ( Image_In ) # # # List_Return.append ( Progress_Default ) List_Return.append ( Progress_Default ) # Post-Processing Ratio_Resize_Global=Ratio_Resize # List_Return.append(Framework.update(width=Width_Global,height=Height_Global)) return List_Return # ！需要拆分Wrapper、合并内容 def DoOnImageBox_Stepping_TabSelecting\ ( Image_BackUp_In, Image_In, # Width_Crop_Camera_Start_In, # Width_Crop_Camera_End_In, # Ratio_Crop_Camera_Start_In, Ratio_Crop_Camera_End_In, # # Argument_System:Framework.SelectData, ) : global IsStartingCameraSetting_Global global Camera_Start_Global global Camera_End_Global List_Return=None Image_Return=None if(IsStartingCameraSetting_Global): # Pre-Processing # 清除前次的痕迹 Image_In=Number.copy(Image_BackUp_In) Location_Camera_Start_Original=Location_Point(Argument_System.index [ 0 ],Argument_System.index[1 ] ) # Debug print(F"Before Processing Start Camera Location in {Inspect.currentframe().f_code.co_name}()") print(F"{Location_Camera_Start_Original.Horizon=},{Location_Camera_Start_Original.Vertical=}") print(F"Ratio: {Location_Camera_Start_Original.Horizon/Location_Camera_Start_Original.Vertical}") print(F"Ratio Global | Original as contrast: {Ratio_Global=}") # print(F"Resized Image Size should be :") print(F"Horizon : {Location_Camera_Start_Original.Horizon/Ratio_Resize_Global}, Vertical : {Location_Camera_Start_Original.Vertical/Ratio_Resize_Global}") # Width_Crop_Camera_Start_Original=Approximate_To_Integer(Width_Crop_Camera_Start_In*Ratio_Resize_Global) Image_Return,Location_Camera_Start= HighLight_Crop_Using_Ratio ( Image_In , Location_Camera_Start_Original , Ratio_Crop_Camera_Start_In, Color_In = (255 , 165, 0) ) # ！需要Constant化：Orange List_Return=\ [ Image_Return, # Width_Crop_Camera_Start_Return, # Width_Crop_Camera_End_In, # 为了保证数值的精度，Original的坐标使用原始数据 # ！需要合并、提取函数+Constant F"#### Horizon: {Location_Camera_Start.Horizon} (Original: {Location_Camera_Start_Original.Horizon }), Vertical: {Location_Camera_Start.Vertical} (Original: {Location_Camera_Start_Original.Vertical })", PlaceHolder_Label_Camera, ] IsStartingCameraSetting_Global=False Camera_Start_Global=Location_Camera_Start else: Location_Camera_End_Original=Location_Point(Argument_System.index [ 0 ],Argument_System.index[1]) # Debug print(F"Before Processing End Camera Location in {Inspect.currentframe().f_code.co_name}()") print(F"{Location_Camera_End_Original.Horizon=},{Location_Camera_End_Original.Vertical=}") print(F"Ratio: {Location_Camera_End_Original.Horizon/Location_Camera_End_Original.Vertical}") print(F"Ratio Global | Original as contrast: {Ratio_Global=}") # print(F"Resized Image Size should be :") print(F"Horizon : {Location_Camera_End_Original.Horizon/Ratio_Resize_Global}, Vertical : {Location_Camera_End_Original.Vertical/Ratio_Resize_Global}") # Width_Crop_Camera_End_Original=Approximate_To_Integer(Width_Crop_Camera_End_In*Ratio_Resize_Global) Image_Return, Location_Camera_End= HighLight_Crop_Using_Ratio ( Image_In , Location_Camera_End_Original , Ratio_Crop_Camera_End_In, Color_In = (255 , 0 , 255) ) # ！需要Constant化：Magenta List_Return=\ [ Image_Return , # Width_Crop_Camera_Start_In, # Width_Crop_Camera_End_Return, F"#### Horizon: {Camera_Start_Global.Horizon} (Original: {Approximate_To_Integer(Camera_Start_Global.Horizon*Ratio_Resize_Global)}), Vertical: {Camera_Start_Global.Vertical} (Original: {Approximate_To_Integer(Camera_Start_Global.Vertical*Ratio_Resize_Global)})", # 为了保证数值的精度，Original的坐标使用原始数据 F"#### Horizon: {Location_Camera_End.Horizon} (Original: {Approximate_To_Integer(Location_Camera_End_Original.Horizon)}), Vertical: {Location_Camera_End.Vertical} (Original: {Approximate_To_Integer(Location_Camera_End_Original.Vertical)})", ] IsStartingCameraSetting_Global=True Camera_End_Global=Location_Camera_End return List_Return # ！考虑合并Event Handler到1个 def DoOnSlider_Camera_Start_Width_Stepping_TabChanged (Width_In) : return Adjust_Ratio(Width_In = Width_In) def DoOnSlider_Camera_Start_Height_Stepping_TabChanged (Height_In) : return Adjust_Ratio(Height_In=Height_In) def DoOnSlider_Camera_End_Width_Stepping_TabChanged (Width_In) : return Adjust_Ratio(Width_In=Width_In) def DoOnSlider_Camera_End_Height_Stepping_TabChanged (Height_In) : return Adjust_Ratio(Height_In=Height_In) #region Core def Adjust_Ratio\ ( Width_In=None, Height_In=None, ) : Value_Return=None if(Width_In is not None): Value_Return=Approximate_To_Integer(Width_In/Ratio_Global) else: Value_Return=Approximate_To_Integer(Height_In*Ratio_Global) return Value_Return #endregion Core #region Utility def Approximate_To_Integer ( Input_In ) : # return Mathemastics.floor(Input_In) # 比较合理的估算方式：保障坐标不会超出范围 # return int(Input_In) # 是截尾法，是比较合理的估算方式：保障坐标不会超出范围，同时对负数值的处理符合预期（趋向〇，而非-∞） return int(round(Input_In)) # AIS源码|调用的Python源码中主要的坐标的处理逻辑 def HighLight_Crop_Using_Ratio\ ( Image_In, Location_Camera_Original_In:Location_Point, Ratio_Crop_In, Color_In, ): Height_Original,Width_Original=Image_In.shape[:2] # Crop_Original_Input = Location_Point ( Width_Original_In , Approximate_To_Integer ( Width_Original_In / Ratio_Global ) ) Crop_Original_Input = Location_Point ( Approximate_To_Integer(Width_Original*Ratio_Crop_In) , Approximate_To_Integer ( Height_Original*Ratio_Crop_In ) ) Size_Original=Location_Point ( Width_Original , Height_Original ) return HighLight_Crop_Core(Image_In,Location_Camera_Original_In,Crop_Original_Input,Color_In,Size_Original) def HighLight_Crop_Using_Crop\ ( Image_In, Location_Camera_Original_In:Location_Point, Crop_Original_In:Location_Point, Color_In, ): Height_Original,Width_Original=Image_In.shape[:2] Size_Original=Location_Point ( Width_Original , Height_Original ) return HighLight_Crop_Core(Image_In,Location_Camera_Original_In,Crop_Original_In,Color_In,Size_Original) def HighLight_Crop_Core\ ( Image_In, Location_Camera_Original_In:Location_Point, Crop_Original_In:Location_Point, Color_In, Size_Original_In:Location_Point, ): # Debug print(F"{Inspect.currentframe().f_code.co_name}()") print(F"{Color_In=}") Image_Return=Number.copy(Image_In) Location_Camera_Return:Location_Point=None Width_Return=None Location_Camera_Original:Location_Point=None Crop_Original=None # Pre-Processing # Crop_Camera = Crop_Camera \ # if IsFullImageRequired ( Width_In,Ratio_Resize_Global ) \ # else Adjust_Crop ( Location_In = Location_Camera_In , Crop_In = Crop_Camera ,Ratio_Resize_In = Ratio_Resize_Global) # Location_Camera_Original,Crop_Original = Adjust_Crop ( Location_Camera_In = Location_Camera_Original_In , Crop_In = Crop_Original_Input , Ratio_Resize_Image_In = Ratio_Resize_Global , Limit_Ratio_Resize_Crop_In = Ratio_Crop_In ) Crop_Original=Crop_Original_In Location_Camera_Original=Centralise_Camera(Location_Camera_Original_In,Crop_Original_In,Size_Original_In) # Processing: Add Rectangle to mark Selection # ？是否需要进行Open CV的颜色顺序的转换 Location_Start_Original,Location_End_Original=ConvertLocation_From_Center_To_Corner(Location_Camera_Original,Crop_Original ) Vision.rectangle\ ( img = Image_Return , pt1 = Location_Start_Original , # Start Location pt2 = Location_End_Original , # End Location color = Color_In , # thickness = 2, # ！需要Constant化 # thickness = 5, # ！需要Constant化 thickness = 10, # ！需要Constant化 ) # ？Python语法是否对Structure的算术运算有简化 # ！需要提取函数 Location_Camera_Return=Location_Point ( Approximate_To_Integer ( Location_Camera_Original.Horizon / Ratio_Resize_Global ), Approximate_To_Integer(Location_Camera_Original.Vertical / Ratio_Resize_Global ) ) Width_Return=Approximate_To_Integer( Crop_Original.Horizon / Ratio_Resize_Global ) return\ [ Image_Return, Location_Camera_Return, # Width_Return, ] def ConvertLocation_From_Center_To_Corner\ ( Camera_In:Location_Point, Crop_In:Location_Point, ) : # 计算起止点坐标 Location_Start_Return = Location_Point\ ( Camera_In.Horizon - Approximate_To_Integer ( Crop_In.Horizon / 2 ) , Camera_In.Vertical - Approximate_To_Integer ( Crop_In.Vertical / 2 ) , ) Location_End_Return = Location_Point\ ( Camera_In.Horizon + Approximate_To_Integer ( Crop_In.Horizon / 2 ) , Camera_In.Vertical + Approximate_To_Integer ( Crop_In.Vertical / 2 ) , ) # 返回结果 return\ [ Location_Start_Return, Location_End_Return, ] def Centralise_Camera(Location_Camera_In:Location_Point,Crop_In:Location_Point,Size_In:Location_Point): Location_Camera_Return:Location_Point=None Location_Start:Location_Point Location_End:Location_Point Location_Start,Location_End=ConvertLocation_From_Center_To_Corner(Location_Camera_In,Crop_In) ΔHorizon=0 ΔVertical=0 if Location_Start.Horizon<0: ΔHorizon=-Location_Start.Horizon elif Location_End.Horizon>Size_In.Horizon: ΔHorizon= Size_In.Horizon - Location_End.Horizon else: pass if Location_Start.Vertical<0: ΔVertical=-Location_Start.Vertical elif Location_End.Vertical>Size_In.Vertical: ΔVertical= Size_In.Vertical - Location_End.Vertical else: pass Location_Camera_Return=Location_Point(Location_Camera_In.Horizon+ΔHorizon,Location_Camera_In.Vertical+ΔVertical) return Location_Camera_Return #endregion Utility #endregion Event Handler #region Customised #region Anime Instance Segmentation def HighLight_Crop_AutoZoom (Image_In,Configuration_In) : # with Engine.no_grad(): Image_Return=None # Image_Return=Image_In if \ (Image_In is not None)\ and (Configuration_In is not None): Height_Original , Width_Original = Image_In.shape [ :2 ] Location_Camera_Start_Original = Location_Point ( Approximate_To_Integer ( Width_Original / 2 ) , Approximate_To_Integer ( Height_Original / 2 ) ) Camera_Start = \ { 'fltCenterU' : Location_Camera_Start_Original.Horizon , 'fltCenterV' : Location_Camera_Start_Original.Vertical , 'intCropWidth' : Approximate_To_Integer ( Ratio_Crop_AutoZoomed * Configuration_In.int_width ) , 'intCropHeight' : Approximate_To_Integer ( Ratio_Crop_AutoZoomed * Configuration_In.int_height ) , } # Camera_End=Anime3D_KenBurns.process_autozoom\ # ( # { # 'fltShift' : 100.0 , # 'fltZoom' : 1.25 , # 'objFrom' : Camera_Start, # } , # Configuration_In, # ) # Location_Camera_End_Original=Location_Point(Camera_End[F"fltCenterU"],Camera_End[F"fltCenterV"]) Location_Camera_End_Original = Location_Point ( Location_Camera_Start_Original.Horizon + 20 , Location_Camera_Start_Original.Vertical + 20 ) # Crop_Camera_End_Original=Location_Point(Camera_End[F"intCropWidth"],Camera_End[F"intCropHeight"]) # Start Camera Image_Processing , _ = HighLight_Crop_Using_Ratio ( Image_In , Location_Camera_Start_Original , Ratio_Crop_AutoZoomed , Color_In = (255 , 165 , 0) ) # ！需要Constant化：Orange # End Camera # Image_Return , _ = HighLight_Crop_Using_Crop ( Image_Processing , Location_Camera_End_Original , Crop_Original_In = Crop_Camera_End_Original , Color_In = (255 , 0 , 255) ) # ！需要Constant化：Magenta Image_Return , _ = HighLight_Crop_Using_Ratio ( Image_Processing , Location_Camera_End_Original , Ratio_Crop_In = Ratio_Crop_AutoZoomed * 0.8 , Color_In = (255 , 0 , 255) ) # ！需要Constant化：Magenta elif Configuration_In is None: Image_Return=Image_In else: pass return Image_Return def DoOnCheckBox_Auto_Zoom_Stepping_TabChanged (IsAutoZoomed_In,Image_In) : Image_Return=None if(IsAutoZoomed_In): Image_Return=HighLight_Crop_AutoZoom(Image_In,Configuration_Global) else: Image_Return=Image_In return Image_Return def Create_Stepping_Demonstration_Tab () : Title=F"3D Ken Burns" with Framework.Tab ( Title ) as Page: # Pre-Processing Page_Global = Page # Framework.Markdown(Title) # Framework.Markdown() # Framework.Markdown() Framework.Markdown ( F"# Step 0. Choose Image") with Framework.Row(): with Framework.Column(): Title_Image_Input_Original=F"Input Image" Framework.Markdown ( F"#### {Title_Image_Input_Original}" ) # ！需要需要禁用更改Image功能，仅作为大屏展示|操作 ImageBox_Input_Original_Stepping_Tab = Framework.Image \ ( label = Title_Image_Input_Original , show_label = False, # # type=F"pil", type=F"numpy", # interactive = True, # editable = False, # height = 600, ) # with Framework.Row\ # ( # # wrap=False, # # min_width = 180, # ): # with Framework.Column\ # ( # min_width = 160, # ): # CheckBox_Convert_To_BGR_Stepping_Tab = Create_Step_CheckBox\ # ( # Name_In = F"RGB → BGR", # Value_Default_In = True, # ) # with Framework.Column\ # ( # min_width = 180, # ): # ImageBox_After_1st_Process_Stepping_Tab=Framework.Image\ # ( # label = F"Image-After 1st Process", # # type=F"pil", # type=F"numpy", # ) # Button_Submit_After_1st_Process_Stepping_Tab=Framework.Button\ # ( # F"Submit", # ) # CheckBox_After_1st_Process_Stepping_Tab=Create_Step_CheckBox\ # ( # Name_In = F"1st", # Value_Default_In = True, # ) # with Framework.Row(): with Framework.Column(): List_Example_Stepping_Tab=Framework.Examples\ ( examples=\ [ OS.path.join ( Path_List_Sample , Carrier_Path_Image ) for Carrier_Path_Image in OS.listdir ( Path_List_Sample ) ], inputs = \ [ ImageBox_Input_Original_Stepping_Tab, ], examples_per_page = 40, ) Framework.Markdown ( F"---" ) Framework.Markdown ( F"# Step 1. Generate Instance Masks") with Framework.Row(): with Framework.Column \ ( min_width = 400 , ) : Title_After_Drawing_Instances=F"Instances" Framework.Markdown(F"#### {Title_After_Drawing_Instances}") ImageBox_After_2nd_Process_Stepping_Tab = Framework.Image \ ( label = Title_After_Drawing_Instances , show_label = False, # type=F"pil", type = F"numpy" , interactive = False, # editable = False, height = 600, ) # Button_Submit_After_2nd_Process_Stepping_Tab=Framework.Button\ # ( # F"Submit", # ) # CheckBox_After_2nd_Process_Stepping_Tab = Create_Step_CheckBox\ # ( # Name_In = F"2nd", # Value_Default_In = True, # ) with Framework.Column \ ( min_width = 180 , ) : Framework.Markdown(F"#### Progress") Label_Progress_Instance_Generating=Framework.Label\ ( value =Progress_Default, # label = F"Progress", label = F"", # show_label = False, ) Framework.Markdown(F"---") Framework.Markdown(F"#### Options") CheckBox_BoundingBox_Stepping_Tab = Create_Step_CheckBox\ ( Name_In = F"Bounding Box", Value_Default_In = True, ) CheckBox_InstanceMask_Stepping_Tab = Create_Step_CheckBox\ ( Name_In = F"Instance Mask", Value_Default_In = True, ) CheckBox_InstanceContour_Stepping_Tab = Create_Step_CheckBox\ ( Name_In = F"Instance Contour", Value_Default_In = True, IsVisible_In = IsAllActivatedControlDisplayed, ) # 生成Text Tag CheckBox_Tag_List_Stepping_Tab = Create_Step_CheckBox\ ( Name_In = F"Tag", Value_Default_In = False, IsVisible_In = IsAllActivatedControlDisplayed, ) Slider_Mask_α_Stepping_Tab=Framework.Slider\ ( value = Mask_α_Default, # 相当于default minimum = 0, maximum = 1, step = 0.01, # multiselect = True, label = F"Transparency of Mask", # info = F"The higher the value set, the brighter the brightness is.", ) # with Framework.Row(): # with Framework.Column(): CheckBox_IsVerbosing_Stepping_Tab = Create_Step_CheckBox\ ( Name_In = F"Verbose", Value_Default_In = False, Tip_In = F"If checked, Programme will generate intermediate | temporary Image Results in its Root Directory.", IsVisible_In = IsAllActivatedControlDisplayed, ) Button_Run_With_Steps_Stepping_Tab=Framework.Button\ ( F"Run", ) with Framework.Row(): # ！需要命名为F"Intermediate Masks" # with Framework.Group(): with Framework.Column \ ( min_width = 160 , ) : Title_After_Generating_Coarse_Depth=F"Coarse Depth" Framework.Markdown(F"#### {Title_After_Generating_Coarse_Depth}") ImageBox_After_3rd_Process_Stepping_Tab = Framework.Image \ ( label = Title_After_Generating_Coarse_Depth , show_label = False, # type=F"pil", type = F"numpy" , interactive = False, # editable = False, ) # Button_Submit_After_3rd_Process_Stepping_Tab=Framework.Button\ # ( # F"Submit", # ) # CheckBox_After_3rd_Process_Stepping_Tab = Create_Step_CheckBox\ # ( # Name_In = F"3rd", # Value_Default_In = True, # ) with Framework.Column \ ( min_width = 160 , ) : Title_After_Generating_Adjusted_Depth=F"Adjusted Depth" Framework.Markdown(F"#### {Title_After_Generating_Adjusted_Depth}") ImageBox_After_4th_Process_Stepping_Tab = Framework.Image \ ( label = Title_After_Generating_Adjusted_Depth , show_label = False, # type=F"pil", type = F"numpy" , interactive = False, # editable = False, ) # Button_Submit_After_4th_Process_Stepping_Tab=Framework.Button\ # ( # F"Submit", # ) # CheckBox_After_4th_Process_Stepping_Tab = Create_Step_CheckBox\ # ( # Name_In = F"4th", # Value_Default_In = True, # ) with Framework.Column \ ( min_width = 160 , ) : Title_After_Generating_Final_Depth=F"Final Depth" Framework.Markdown(F"#### {Title_After_Generating_Final_Depth}") ImageBox_After_5th_Process_Stepping_Tab = Framework.Image \ ( label = Title_After_Generating_Final_Depth , show_label = False, # type=F"pil", type = F"numpy" , interactive = False, # editable = False, ) # Button_Submit_After_5th_Process_Stepping_Tab=Framework.Button\ # ( # F"Submit", # ) # CheckBox_After_5th_Process_Stepping_Tab = Create_Step_CheckBox\ # ( # Name_In = F"5th", # Value_Default_In = True, # ) Framework.Markdown ( F"---" ) # Frame List Generating Part Framework.Markdown ( F"# Step 2. Set Camera View at Start and End" ) Framework.Markdown(F"## The ORANGE box shows camera view at start. The MAGENTA box shows camera view at end.") # Framework.Markdown(F"## Crop of Start Camera in Orange, Crop of End Camera in Magenta.") with Framework.Row(): with Framework.Column () : Title_ImageBox_Output_Original=F"Input Image with Camera Views at Start and End" Framework.Markdown ( F"#### {Title_ImageBox_Output_Original}" ) # ！需要需要禁用更改Image功能，仅作为大屏展示|操作 ImageBox_Output_Original_Stepping_Tab=Framework.Image\ ( label = Title_ImageBox_Output_Original, show_label = False, # type=F"pil", type=F"numpy", # # interactive = False, # Due to the bug under solving, we have to work in interactive mode: https://github.com/gradio-app/gradio/issues/5945 interactive = True, # # editable = False, # height = 600, ) # with Framework.Row(): with Framework.Column () : Framework.Markdown(F"#### Progress:") Label_Progress_Camera_Generating=Framework.Label\ ( value =Progress_Default, # label = F"Progress:", label = F"", # show_label = False, ) Framework.Markdown(F"---") Framework.Markdown ( F"### Method 1: Auto Zoom (Pre-set Camera Views)" ) CheckBox_Auto_Zoom_Stepping_Tab = Create_Step_CheckBox\ ( Name_In = F"Auto Zoom", Value_Default_In = True, ) Framework.Markdown ( F"---" ) Framework.Markdown ( F"### Method 2: Set Camera Views Manually" ) # Label_Camera_Start_Stepping_Tab=Framework.Label\ # ( # value =PlaceHolder_Label_Camera, # label = F"Start Camera Location", # # elem_classes = F"Label-Display-Position", # ) Label_Camera_Start_Stepping_Tab=Framework.Markdown\ ( PlaceHolder_Label_Camera, visible = IsAllActivatedControlDisplayed, ) Title_Ratio_Crop_Start=F"Camera View at Start" Framework.Markdown ( F"{Title_Ratio_Crop_Start}" ) Slider_Ratio_Crop_Camera_Start = Framework.Slider \ ( value = Ratio_Crop_Camera_Start_Deault , # 相当于default minimum = Ratio_Crop_Minimum , # ！考虑改为Constant maximum = 1 , # ！考虑改为Constant step = 0.01 , # multiselect = True, # label = Title_Ratio_Crop_Start , label = F"Width of Camera View at Start: Width of Input Image" , # show_label = False, show_label = True, ) Framework.Markdown\ ( F"---", visible = IsAllActivatedControlDisplayed, ) # with Framework.Row(): # with Framework.Column () : Slider_Camera_Start_Width = Framework.Slider \ ( value = Width_Global , # 相当于default minimum = Limit_Minimum , # ！考虑改为Constant maximum = Limit_Maxmum , # ！考虑改为Constant step = 1 , # multiselect = True, label = F"Start Camera-Width (After Down Scaling)" , visible = IsAllActivatedControlDisplayed, ) Slider_Camera_Start_Height = Framework.Slider \ ( value = Height_Global , # 相当于default minimum = Limit_Minimum , # ！考虑改为Constant maximum = Limit_Maxmum , # ！考虑改为Constant step = 1 , # multiselect = True, label = F"Start Camera-Height (After Down Scaling)" , visible = False, ) # with Framework.Column () : # Label_Camera_End_Stepping_Tab=Framework.Label\ # ( # value = PlaceHolder_Label_Camera, # label = F"End Camera Location", # # elem_classes = F"Label-Display-Position", # ) Label_Camera_End_Stepping_Tab=Framework.Markdown\ ( PlaceHolder_Label_Camera, visible = IsAllActivatedControlDisplayed, ) Title_Ratio_Crop_End=F"Camera View at End" Framework.Markdown ( F"{Title_Ratio_Crop_End}" ) Slider_Ratio_Crop_Camera_End = Framework.Slider \ ( value = Ratio_Crop_Camera_End_Deault , # 相当于default minimum = Ratio_Crop_Minimum , # ！考虑改为Constant maximum = 1 , # ！考虑改为Constant step = 0.01 , # multiselect = True, # label = Title_Ratio_Crop_End , label = F"Width of Camera View at End: Width of Input Image" , # show_label = False, show_label = True, ) Slider_Camera_End_Width = Framework.Slider \ ( value = Width_Global , # 相当于default minimum = Limit_Minimum , maximum = Limit_Maxmum , step = 1 , # multiselect = True, label = F"End Camera-Width (After Down Scaling)" , visible = IsAllActivatedControlDisplayed, ) Slider_Camera_End_Height = Framework.Slider \ ( value = Height_Global , # 相当于default minimum = Limit_Minimum , maximum = Limit_Maxmum , step = 1 , # multiselect = True, label = F"End Camera-Height (After Down Scaling)" , visible = False, ) Framework.Markdown ( F"---" ) Button_Generate_Camera_View_Stepping_Tab=Framework.Button\ ( F"Generate Camera View", interactive = False, ) # Framework.Markdown(F"#### The closer the Cameras are, the steadier the effect is.") with Framework.Row\ ( visible = IsAllActivatedControlDisplayed, ): Slider_Number_Frame_Stepping_Tab = Framework.Slider \ ( value = 75 , # 相当于default minimum = 0 , maximum = 100 , # ！Pending step = 1 , # multiselect = True, label = F"Frame Number" , ) CheckBox_IsInpainting_Stepping_Tab = Create_Step_CheckBox\ ( Name_In = F"Inpainting", Value_Default_In = True, ) Slider_Depthor_Stepping_Tab = Framework.Slider \ ( value = 3 , # 相当于default minimum = 0 , maximum = 10 , # ！Pending step = 1 , # multiselect = True, label = F"Depth Factor" , visible = False, # 目前认为不需要进行手动设置 ) Framework.Markdown(F"#### Inpainted Backgrounds") with Framework.Row(): with Framework.Column(): Title_Camera_PositiveFilm_Start=F"Start" Framework.Markdown(F"{Title_Camera_PositiveFilm_Start}") ImageBox_Output_PositiveFilm_Start_Stepping_Tab=Framework.Image\ ( label = Title_Camera_PositiveFilm_Start , show_label = False, # type=F"pil", type=F"numpy", interactive = False, # editable = False, height = 600 , ) with Framework.Column () : Title_Camera_PositiveFilm_End = F"End" Framework.Markdown ( F"{Title_Camera_PositiveFilm_End}" ) ImageBox_Output_PositiveFilm_End_Stepping_Tab=Framework.Image\ ( label = Title_Camera_PositiveFilm_End , show_label = False, # type=F"pil", type=F"numpy", interactive = False, # editable = False, height = 600 , ) Framework.Markdown(F"#### Masks for Background Inpainting") with Framework.Row(): with Framework.Column () : Title_Camera_Mask_Start = F"Start" Framework.Markdown ( F"{Title_Camera_Mask_Start}" ) ImageBox_Output_Mask_Start_Stepping_Tab=Framework.Image\ ( label = Title_Camera_Mask_Start , show_label = False, # type=F"pil", type=F"numpy", interactive = False, # editable = False, height = 600 , ) with Framework.Column () : Title_Camera_Mask_End = F"End" Framework.Markdown ( F"{Title_Camera_Mask_End}" ) ImageBox_Output_Mask_End_Stepping_Tab=Framework.Image\ ( label = Title_Camera_Mask_End , show_label = False, # type=F"pil", type=F"numpy", interactive = False, # editable = False, height = 600 , ) Framework.Markdown ( F"---" ) # Video Encoding Part with Framework.Row(): with Framework.Column\ ( min_width = 200, ): Framework.Markdown ( F"# Step 3. 3D Ken Burns Result" ) CheckBox_Depth_Field_Stepping_Tab = Create_Step_CheckBox\ ( Name_In = F"Depth Field", Value_Default_In = False, IsVisible_In = False, ) with Framework.Column\ ( min_width = 200, ): Button_Encode_Video_Stepping_Tab=Framework.Button\ ( F"Encode Video", interactive = False, # disable = True, ) # Video Box已自带，不再需要，但是Video无法独立使用 with Framework.Column\ ( min_width = 200, visible = IsAllActivatedControlDisplayed, ): File_Output_Downloaing_Stepping_Tab=Framework.File \ ( # size=(300,100), ) with Framework.Row(): # ！需要缩小占用，最好以Button+浮层形式展示；或最佳的是类似Clip工具的横向Frame轴形式拉动（横向Scroll，纵向以单张Frame Height自适应） with Framework.Column \ ( min_width = 200 , visible = IsAllActivatedControlDisplayed, ) : Framework.Markdown\ ( F"#### After 1 Image clicked, use Left (**←**) | Right (**→**) to experience the animation effect.", # visible = IsAllActivatedControlDisplayed, ) BatchedImageBox_Output_ListFrame_Stepping_Tab = Create_Output_BatchedImageBox \ ( Name_In = F"Frame List" , Type_In = F"numpy" , # IsPreviewMode_In = True, # ！慎用，会导致UI Thread死循环，UI直接卡死 Height_In = 400 , ) with Framework.Column(min_width = 300,visible = IsAllActivatedControlDisplayed): Title_Image_Input_Original_Contrast=F"Original Input Image as Contrast" Framework.Markdown ( F"#### {Title_Image_Input_Original_Contrast}" ) # ！需要需要禁用更改Image功能，仅作为大屏展示|操作 ImageBox_Output_Original_Contrast = Framework.Image \ ( label = Title_Image_Input_Original_Contrast , show_label = False, # type=F"pil", type=F"numpy", interactive = False, # editable = False, height = 600 , ) with Framework.Column\ ( min_width = 100, # visible = False, ): Framework.Checkbox(visible = False) with Framework.Column\ ( # width=1000, ) : Framework.Markdown(F"### A **Five** Seconds Video in variant Frame Ratio.") # with Framework.Row(): # Choose 1 Video related Control, then combined with File Control in same Column VideoBox_Output_Stepping_Tab = Framework.PlayableVideo \ ( show_label = False, # style={ "height" : "600px" }, # height=600, # width=800, # width=F"60%", height = 600, ) with Framework.Column\ ( min_width = 100, # visible = False, ): Framework.Checkbox(visible = False) # !Add basic authority | permission, authors, references # Bind Event handler # Button_Submit_After_1st_Process_Stepping_Tab.click\ # ( # fn = DoOnButtonSubmit_AIS_Instance_Segmentation_Clicking, # inputs =\ # [ # ImageBox_Input_Original_Stepping_Tab, # ], # outputs =\ # [ # ImageBox_After_1st_Process_Stepping_Tab, # ], # ) # Button_Submit_After_2nd_Process_Stepping_Tab.click\ # ( # fn = DoOnButtonSubmit_AIS_Infer_Disparity_Depth_Estimation_Clicking, # inputs =\ # [ # # ImageBox_After_1st_Process_Stepping_Tab, # ], # outputs =\ # [ # ImageBox_After_2nd_Process_Stepping_Tab, # ], # ) # Button_Submit_After_3rd_Process_Stepping_Tab.click\ # ( # fn = DoOnButtonSubmit_AIS_Infer_Disparity_Depth_Coarse_Clicking, # inputs =\ # [ # # ImageBox_After_2nd_Process_Stepping_Tab, # ], # outputs =\ # [ # ImageBox_After_3rd_Process_Stepping_Tab, # ], # ) # Button_Submit_After_4th_Process_Stepping_Tab.click\ # ( # fn = DoOnButtonSubmit_AIS_Infer_Disparity_Depth_Adjusted_Depth_Clicking, # inputs =\ # [ # # ImageBox_After_3rd_Process_Stepping_Tab, # ], # outputs =\ # [ # ImageBox_Disparity_After_4th_Process_Stepping_Tab, # ImageBox_After_4th_Process_Stepping_Tab, # ], # ) # Button_Submit_After_5th_Process_Stepping_Tab.click\ # ( # fn = DoOnButtonSubmit_AIS_Adjustment_Anime_Segmentation_TabClicking, # inputs =\ # [ # # ImageBox_After_4th_Process_Stepping_Tab, # ], # outputs =\ # [ # ImageBox_Disparity_After_5th_Process_Stepping_Tab, # ImageBox_After_5th_Process_Stepping_Tab, # ], # ) # ImageBox_Input_Original_Stepping_Tab.change\ # ( # # fn=DoOnImageBox_Input_Original_Stepping_TabChanged, # fn=lambda Image_In_Out:[Image_In_Out,Image_In_Out,Image_In_Out,Image_In_Out,], # inputs =\ # [ # ImageBox_Input_Original_Stepping_Tab, # ], # outputs =\ # [ # # ImageBox_After_1st_Process_Stepping_Tab, # ImageBox_After_2nd_Process_Stepping_Tab, # ImageBox_After_3rd_Process_Stepping_Tab, # ImageBox_After_4th_Process_Stepping_Tab, # ImageBox_After_5th_Process_Stepping_Tab, # ], # ) # # ImageBox_Input_Original_Stepping_Tab.select\ # ( # fn=DoOnImageBox_Stepping_TabSelecting, # inputs = \ # [ # Slider_Camera_Start_Width, # Slider_Camera_End_Width, # ] , # outputs = \ # [ # Label_Camera_Start_Stepping_Tab, # Label_Camera_End_Stepping_Tab, # ] , # ) ImageBox_Input_Original_Stepping_Tab.change\ ( fn=DoOnImageBox_Input_Original_Stepping_TabChanged, inputs = \ [ ImageBox_Input_Original_Stepping_Tab, Slider_Ratio_Crop_Camera_Start, Slider_Ratio_Crop_Camera_End, ] , outputs = \ [ ImageBox_After_2nd_Process_Stepping_Tab, ImageBox_After_3rd_Process_Stepping_Tab, ImageBox_After_4th_Process_Stepping_Tab, ImageBox_After_5th_Process_Stepping_Tab, # CheckBox_BoundingBox_Stepping_Tab, CheckBox_InstanceMask_Stepping_Tab, CheckBox_InstanceContour_Stepping_Tab, CheckBox_Tag_List_Stepping_Tab, Slider_Mask_α_Stepping_Tab, # ImageBox_Output_PositiveFilm_Start_Stepping_Tab, ImageBox_Output_PositiveFilm_End_Stepping_Tab, ImageBox_Output_Mask_Start_Stepping_Tab, ImageBox_Output_Mask_End_Stepping_Tab, # CheckBox_Depth_Field_Stepping_Tab, BatchedImageBox_Output_ListFrame_Stepping_Tab, VideoBox_Output_Stepping_Tab, # Label_Camera_Start_Stepping_Tab, Label_Camera_End_Stepping_Tab, # Slider_Number_Frame_Stepping_Tab, CheckBox_IsInpainting_Stepping_Tab, # # CheckBox_Auto_Zoom_Stepping_Tab, # # Slider_Ratio_Crop_Camera_Start, Slider_Camera_Start_Width, Slider_Camera_Start_Height, # Slider_Ratio_Crop_Camera_End, Slider_Camera_End_Width, Slider_Camera_End_Height, # ImageBox_Output_Original_Stepping_Tab, ImageBox_Output_Original_Contrast, # # # # VideoBox_Output_Stepping_Tab, Label_Progress_Instance_Generating, Label_Progress_Camera_Generating, ] , ) # ImageBox_Output_Original_Stepping_Tab.change\ # ( # fn=DoOnImageBox_Output_Original_Stepping_TabChanged, # inputs = \ # [ # ImageBox_Output_Original_Stepping_Tab, # ] , # outputs = \ # [ # ImageBox_Output_Original_Stepping_Tab, # ] , # ) ImageBox_Output_Original_Stepping_Tab.select\ ( fn=DoOnImageBox_Stepping_TabSelecting, inputs = \ [ ImageBox_Input_Original_Stepping_Tab, ImageBox_Output_Original_Stepping_Tab, # Slider_Camera_Start_Width, # Slider_Camera_End_Width, # Slider_Ratio_Crop_Camera_Start, Slider_Ratio_Crop_Camera_End, ] , outputs = \ [ ImageBox_Output_Original_Stepping_Tab, # Slider_Camera_Start_Width, # Slider_Camera_End_Width, Label_Camera_Start_Stepping_Tab, Label_Camera_End_Stepping_Tab, ] , ) CheckBox_Auto_Zoom_Stepping_Tab.change\ ( fn=DoOnCheckBox_Auto_Zoom_Stepping_TabChanged, inputs = \ [ CheckBox_Auto_Zoom_Stepping_Tab, ImageBox_Input_Original_Stepping_Tab, ] , outputs = \ [ ImageBox_Output_Original_Stepping_Tab, ] , ) # Slider_Camera_Start_Width.change\ # ( # fn = DoOnSlider_Camera_Start_Width_Stepping_TabChanged, # inputs =\ # [ # Slider_Camera_Start_Width, # ], # outputs =\ # [ # # Slider_Camera_Start_Width, # ], # ) # Slider_Camera_Start_Height.change\ # ( # fn = DoOnSlider_Camera_Start_Height_Stepping_TabChanged, # inputs =\ # [ # Slider_Camera_Start_Height, # ], # outputs =\ # [ # # Slider_Camera_Start_Height, # ], # ) # Slider_Camera_End_Width.change\ # ( # fn = DoOnSlider_Camera_End_Width_Stepping_TabChanged, # inputs =\ # [ # Slider_Camera_End_Width, # ], # outputs =\ # [ # # Slider_Camera_End_Width, # ], # ) # Slider_Camera_End_Height.change\ # ( # fn = DoOnSlider_Camera_End_Height_Stepping_TabChanged, # inputs =\ # [ # Slider_Camera_End_Height, # ], # outputs =\ # [ # # Slider_Camera_End_Height, # ], # ) Button_Run_With_Steps_Stepping_Tab.click\ ( fn = DoOnButton_Run_With_Steps_Stepping_TabClicking, inputs =\ [ ImageBox_Input_Original_Stepping_Tab, CheckBox_BoundingBox_Stepping_Tab, CheckBox_InstanceMask_Stepping_Tab, CheckBox_InstanceContour_Stepping_Tab, CheckBox_Tag_List_Stepping_Tab, Slider_Mask_α_Stepping_Tab, ], outputs =\ [ # ImageBox_After_1st_Process_Stepping_Tab, # ImageBox_After_2nd_Process_Stepping_Tab, # ImageBox_After_3rd_Process_Stepping_Tab, # ImageBox_After_4th_Process_Stepping_Tab, # ImageBox_After_5th_Process_Stepping_Tab, # # ImageBox_Output_Original_Stepping_Tab, # # Button_Generate_Camera_View_Stepping_Tab, # # Label_Progress_Instance_Generating, ], ) Button_Generate_Camera_View_Stepping_Tab.click\ ( fn = DoOnButton_Generate_Camera_View_Stepping_TabClicking, inputs =\ [ CheckBox_IsInpainting_Stepping_Tab, # Slider_Ratio_Crop_Camera_Start, # Slider_Camera_Start_Width, # Slider_Camera_Start_Height, # Slider_Ratio_Crop_Camera_End, # Slider_Camera_End_Width, # Slider_Camera_End_Height, # Slider_Number_Frame_Stepping_Tab, CheckBox_Auto_Zoom_Stepping_Tab, # Slider_Depthor_Stepping_Tab, CheckBox_Depth_Field_Stepping_Tab, ], outputs =\ [ ImageBox_Output_PositiveFilm_Start_Stepping_Tab, ImageBox_Output_PositiveFilm_End_Stepping_Tab, ImageBox_Output_Mask_Start_Stepping_Tab, ImageBox_Output_Mask_End_Stepping_Tab, BatchedImageBox_Output_ListFrame_Stepping_Tab, # # Button_Encode_Video_Stepping_Tab, VideoBox_Output_Stepping_Tab, # # Label_Progress_Camera_Generating, ], ) Button_Encode_Video_Stepping_Tab.click\ ( fn = DoOnButton_Encode_Video_Stepping_TabClicking, # inputs =\ # [ # # To Do # ], outputs =\ [ VideoBox_Output_Stepping_Tab, File_Output_Downloaing_Stepping_Tab, ], # outputs = VideoBox_Output_Stepping_Tab, ) # ！Gradio不能直接呈现URL，只能呈现HTML，但是Flask是基于URL，仅HTML则只有静态Framework def Create_Original_Demonstration_Tab () : with Framework.Tab ( F"Original Demonstration-Ken Burns" ) : Framework.HTML(F"Original Demonstration") #endregion Anime Instance Segmentation #endregion Customised def Create_Information_Tab () : with Framework.Tab ( F"Information" ): # Including Image information and Model information with Framework.Row(): ImageBox_Input_TabInformation=Create_Batch_Input_ImageBox() ImageBox_Output=Create_Output_BatchedImageBox() # !Need to be changed to Status Control|Component def DoOnCheckBox_IsVerbosing_Setting_TabChanged (IsVerbosing_In) : global IsVerbosing_Global IsVerbosing_Global=IsVerbosing_In # Debug print(F"{IsVerbosing_In=} in {Inspect.currentframe().f_code.co_name}()") print(F"{IsVerbosing_Global=} in {Inspect.currentframe().f_code.co_name}()") def Create_Setting_Tab () : with Framework.Tab ( F"Setting" ): # with Framework.Row(): # Button_Apply_Setting = Create_ApplySetting_Button () # Button_Restart = Create_Restart_Button () with Framework.Row\ ( Title=F"Option", # wrap=False, # min_width = 180, ): with Framework.Column(): CheckBox_IsVerbosing_Stepping_Tab = Create_Step_CheckBox\ ( Name_In = F"Verbose", Value_Default_In = False, Tip_In = F"If checked, Programme will generate intermediate | temporary Image Results in its Root Directory.", ) # CheckBox_IsVerbosing_Stepping_Tab.change\ ( fn = DoOnCheckBox_IsVerbosing_Setting_TabChanged, inputs =\ [ CheckBox_IsVerbosing_Stepping_Tab, ], outputs =\ [ # To Do ], ) #endregion Function | Method