(ti `ddlZddlZddlZddlZddlmcmZddl m Z m Z m Z Gdde Z y)N)VaeImageProcessoris_valid_imageis_valid_image_imagelistc JeZdZdZddedzdedzdej fdZ ddej dede jej ze e jjzfd Zededed edeeeffd Zed ej d ededej fdZy)VideoProcessorzSimple video processor.Nheightwidthreturnc t|tr`t|dtjrC|djdk(r1t j dttj|d}t|tr`t|dtjrC|djdk(r1t j dttj|d}t|tjtjfr|jdk(r t|}nYt|trt|ds t|r|g}n,t|trt|dr|}n tdtj|Dcgc]}|j |f||d|c}d}|j#dd d d d }|Scc}w) a Preprocesses input video(s). Keyword arguments will be forwarded to `VaeImageProcessor.preprocess`. Args: video (`list[PIL.Image]`, `list[list[PIL.Image]]`, `torch.Tensor`, `np.array`, `list[torch.Tensor]`, `list[np.array]`): The input video. It can be one of the following: * list of the PIL images. * list of list of PIL images. * 4D Torch tensors (expected shape for each tensor `(num_frames, num_channels, height, width)`). * 4D NumPy arrays (expected shape for each array `(num_frames, height, width, num_channels)`). * list of 4D Torch tensors (expected shape for each tensor `(num_frames, num_channels, height, width)`). * list of 4D NumPy arrays (expected shape for each array `(num_frames, height, width, num_channels)`). * 5D NumPy arrays: expected shape for each array `(batch_size, num_frames, height, width, num_channels)`. * 5D Torch tensors: expected shape for each array `(batch_size, num_frames, num_channels, height, width)`. height (`int`, *optional*, defaults to `None`): The height in preprocessed frames of the video. If `None`, will use the `get_default_height_width()` to get default height. width (`int`, *optional*`, defaults to `None`): The width in preprocessed frames of the video. If `None`, will use get_default_height_width()` to get the default width. Returns: `torch.Tensor` of shape `(batch_size, num_channels, num_frames, height, width)`: A 5D tensor holding the batched channels-first video(s). rzPassing `video` as a list of 5d np.ndarray is deprecated.Please concatenate the list along the batch dimension and pass it as a single 5d np.ndarray)axiszPassing `video` as a list of 5d torch.Tensor is deprecated.Please concatenate the list along the batch dimension and pass it as a single 5d torch.TensorzeInput is in incorrect format. Currently, we only support numpy.ndarray, torch.Tensor, PIL.Image.Image)r r )dimr) isinstancelistnpndarrayndimwarningswarn FutureWarning concatenatetorchTensorcatrr ValueErrorstack preprocesspermute)selfvideor r kwargsimgs h/Volumes/fast/ai/experiments/MLX_z-image/.venv/lib/python3.12/site-packages/diffusers/video_processor.pypreprocess_videozVideoProcessor.preprocess_videos: eT "z%(BJJ'GERSHMM]^L^ MMn  NN5q1E eT "z%(ELL'IeTUhmm_`N` MMp  IIe!,E ebjj%,,7 8UZZ1_KE t $a)AE]^cEdGE t $)A%()KEw  chich\__T__SWuWPVWchiopq aAq!, jsGr$ output_typec V|jd}g}t|D]>}||jdddd}|j||fi|}|j |@|dk(rt j |}|S|dk(rtj |}|S|dk(st|d|S) aI Converts a video tensor to a list of frames for export. Keyword arguments will be forwarded to `VaeImageProcessor.postprocess`. Args: video (`torch.Tensor`): The video as a tensor. output_type (`str`, defaults to `"np"`): Output type of the postprocessed `video` tensor. rrrrrptpilz9 does not exist. Please choose one of ['np', 'pt', 'pil']) shaperanger" postprocessappendrr rr) r#r$r)r% batch_sizeoutputs batch_idx batch_vid batch_outputs r'postprocess_videoz VideoProcessor.postprocess_video]s[[^ z*Ii(00Aq!z:VideoProcessor.classify_height_width_bin..sSuPRAR=Sr7)keyrr)r<minkeysint)r r r8 closest_ratio default_hwr>s @r'classify_height_width_binz(VideoProcessor.classify_height_width_binxsM6E> "FKKM/ST M* :a=!3z!}#555r7samples new_width new_heightc|jd|jd}}||k7s||k7rt||z ||z }t||z}t||z}|j\}} } } } |jdddddj || z| | | }t j |||fdd}||z dz} | |z}||z dz}||z}|d d d d ||| |f}|j || | ||jddddd}|S) a4 Resizes and crops a tensor of videos to the specified dimensions. Args: samples (`torch.Tensor`): A tensor of shape (N, C, T, H, W) where N is the batch size, C is the number of channels, T is the number of frames, H is the height, and W is the width. new_width (`int`): The desired width of the output videos. new_height (`int`): The desired height of the output videos. Returns: `torch.Tensor`: A tensor containing the resized and cropped videos. rrrrrbilinearF)sizemode align_cornersN)r-maxrCr"reshapeF interpolate)rGrHrI orig_height orig_widthr= resized_widthresized_heightncthwstart_xend_xstart_yend_ys r'resize_and_crop_tensorz%VideoProcessor.resize_and_crop_tensorsF#*--"2GMM!4DZ  * $ i(? [0)j2HIE U 23M u!45N$MMMAq!QooaAq!4<rqs. XX[&[r7