i jddlmZmZmZddlZddlZddlZddl m Z ddl m Z ddl mZGdde Zy) )OptionalTupleUnionN)register_to_config)VaeImageProcessor)PIL_INTERPOLATIONceZdZdZe ddedededeeefdededed ed ed ed e e ee ee d ffffd Z de jjdedede jjfdZ dde e jjej"ej&fde ede edeeeffdZxZS)WanAnimateImageProcessorat Image processor to preprocess the reference (character) image for the Wan Animate model. Args: do_resize (`bool`, *optional*, defaults to `True`): Whether to downscale the image's (height, width) dimensions to multiples of `vae_scale_factor`. Can accept `height` and `width` arguments from [`image_processor.VaeImageProcessor.preprocess`] method. vae_scale_factor (`int`, *optional*, defaults to `8`): VAE (spatial) scale factor. If `do_resize` is `True`, the image is automatically resized to multiples of this factor. vae_latent_channels (`int`, *optional*, defaults to `16`): VAE latent channels. spatial_patch_size (`Tuple[int, int]`, *optional*, defaults to `(2, 2)`): The spatial patch size used by the diffusion transformer. For Wan models, this is typically (2, 2). resample (`str`, *optional*, defaults to `lanczos`): Resampling filter to use when resizing the image. do_normalize (`bool`, *optional*, defaults to `True`): Whether to normalize the image to [-1,1]. do_binarize (`bool`, *optional*, defaults to `False`): Whether to binarize the image to 0/1. do_convert_rgb (`bool`, *optional*, defaults to be `False`): Whether to convert the images to RGB format. do_convert_grayscale (`bool`, *optional*, defaults to be `False`): Whether to convert the images to grayscale format. fill_color (`str` or `float` or `Tuple[float, ...]`, *optional*, defaults to `None`): An optional fill color when `resize_mode` is set to `"fill"`. This will fill the empty space with that color instead of filling with data from the image. Any valid `color` argument to `PIL.Image.new` is valid; if `None`, will default to filling with data from `image`. do_resizevae_scale_factorvae_latent_channelsspatial_patch_sizeresample reducing_gap do_normalize do_binarizedo_convert_rgbdo_convert_grayscale fill_color.c Dt || r| r tddyy)Nz`do_convert_rgb` and `do_convert_grayscale` can not both be set to `True`, if you intended to convert the image into RGB format, please set `do_convert_grayscale = False`.z` if you intended to convert the image into grayscale format, please set `do_convert_rgb = False`)super__init__ ValueError) selfr r rrrrrrrrr __class__s q/home/obispo/Crisostomo_bridge/mision_env/lib/python3.12/site-packages/diffusers/pipelines/wan/image_processor.pyrz!WanAnimateImageProcessor.__init__9s3  2tr 3>imagewidthheightreturnc :||z }|j|jz }|jjdu}|jjxsd}||kr|n|j|z|jz}||k\r|n|j|z|jz} |j || ft |jj } tjjd||f|} | j| |dz|dzz |dz| dzz f|r||kr|dz| dzz } | dkDrm| j| j || fdd|dfd| j| j || fd| j|| jfd| | zf| S||kDr}|dz|dzz } | dkDrm| j| j | |fddd|fd| j| j | |f| jd| j|f| |zdf| S) a* Resize the image to fit within the specified width and height, maintaining the aspect ratio, and then center the image within the dimensions, filling empty with data from image. Args: image (`PIL.Image.Image`): The image to resize and fill. width (`int`): The width to resize the image to. height (`int`): The height to resize the image to. Returns: `PIL.Image.Image`: The resized and filled image. Nr)rRGB)color)box)rr) r r!configrresizer rPILImagenewpaste)rrr r!ratio src_ratiofill_with_image_datarsrc_wsrc_hresizedres fill_height fill_widths r_resize_and_fillz)WanAnimateImageProcessor._resize_and_fillPs:.KK%,,. #{{55=[[++0q * f0D 0T9,%,,2F%++2U,,u~8I$++J^J^8_,`iimmEE6?*mE ' UaZ 71uPQz9QR S y $kEQJ6 ?IIgnne[-A1eUVGWnX^dIeII{';!W^^UZ\c\j\jAkl e 34 ""aZ%1*4 >IIgnnj&-A1aQWGXnY_eIfII F';'--QRT[TaTaciAjk'%/3  rc*|jt|tjjr |j}n9t|tj r|j d}n|j d}|jt|tjjr |j}n9t|tj r|j d}n|j d}||z}||z }|jj|jjdz}|jj|jjdz}ttj||z|z|z}ttj||z |z|z}||fS)aH Returns the height and width of the image, downscaled to the next integer multiple of `vae_scale_factor`. Args: image (`Union[PIL.Image.Image, np.ndarray, torch.Tensor]`): The image input, which can be a PIL image, NumPy array, or PyTorch tensor. If it is a NumPy array, it should have shape `[batch, height, width]` or `[batch, height, width, channels]`. If it is a PyTorch tensor, it should have shape `[batch, channels, height, width]`. height (`Optional[int]`, *optional*, defaults to `None`): The height of the preprocessed image. If `None`, the height of the `image` input will be used. width (`Optional[int]`, *optional*, defaults to `None`): The width of the preprocessed image. If `None`, the width of the `image` input will be used. Returns: `Tuple[int, int]`: A tuple containing the height and width, both resized to the nearest integer multiple of `vae_scale_factor * spatial_patch_size`. r&rr) isinstancer*r+r!torchTensorshaper r(r rroundnpsqrt)rrr!r max_area aspect_ratio mod_value_h mod_value_ws rget_default_height_widthz1WanAnimateImageProcessor.get_default_height_widths>2 >%1E5<<0QQ =%1 E5<<0 A A6>~ kk22T[[5S5STU5VV kk22T[[5S5STU5VV rwwx,678KG+Ubggh567;FTu}r) T)r&r&lanczosNTFFFr)NN)__name__ __module__ __qualname____doc__rboolintrstrrrfloatrr*r+r7r?ndarrayr;r<rE __classcell__)rs@rr r sh< !#%.4! !!$%*EF!  "#s(O   #U3uUCZ/@#@AB,5yy55 5  5t!%# 2SYY__bjj%,,>?2 2} 2 sCx 2rr )typingrrrnumpyr? PIL.Imager*r;configuration_utilsrimage_processorrutilsr r rrrZs.*) 50&_0_r