o ۷is@s$dZddlZddlZddlmZmZmZddlmZddl m Z ddl m Z m Z mZmZddlmZmZmZdd lmZdd lmZdd lmZd d lmZmZd dlmZmZm Z ddlm!Z!ddl"m#Z#m$Z$m%Z%m&Z&m'Z'm(Z(m)Z)m*Z*m+Z+e,e-Z.dej/dej/fddZ0de*e+ddfde1e2e1Bde1de3de3dej4dBf ddZ5dde&e'dfde1e2e1Bdej/dBde1de3dej4dBf dd Z6dde$e#e%dfde1e2e1Bdej/e2ej7j7ej7j7fBdBde1d!e1de3dej4dBf d"d#Z8 $dOd%ej/d&ej9dBd'e1fd(d)Z: * +dPdej/d,ed&ej9dej4d-ej;d.e3d'e1fd/d0ZGd5d6d6eZ?Gd7d8d8eZ@Gd9d:d:eZAGd;d<dd>eZCGd?d@d@eZDGdAdBdBeZEGdCdDdDeZFGdEdFdFeZGGdGdHdHeZHGdIdJdJeZIGdKdLdLeZJGdMdNdNeZKdS)Qz6 Text and VAE encoder blocks for QwenImage pipelines. N)"Qwen2_5_VLForConditionalGenerationQwen2TokenizerQwen2VLProcessor) FrozenDict)ClassifierFreeGuidance)InpaintProcessorVaeImageProcessoris_valid_imageis_valid_image_imagelist)AutoencoderKLQwenImageQwenImageControlNetModelQwenImageMultiControlNetModel)calculate_dimensions)logging) unwrap_module)ModularPipelineBlocks PipelineState) ComponentSpec InputParam OutputParam)QwenImageModularPipeline) QWENIMAGE_EDIT_PLUS_IMG_TEMPLATE#QWENIMAGE_EDIT_PLUS_PROMPT_TEMPLATE-QWENIMAGE_EDIT_PLUS_PROMPT_TEMPLATE_START_IDXQWENIMAGE_EDIT_PROMPT_TEMPLATE(QWENIMAGE_EDIT_PROMPT_TEMPLATE_START_IDX#QWENIMAGE_LAYERED_CAPTION_PROMPT_CN#QWENIMAGE_LAYERED_CAPTION_PROMPT_ENQWENIMAGE_PROMPT_TEMPLATE#QWENIMAGE_PROMPT_TEMPLATE_START_IDX hidden_statesmaskcCs4|}|jdd}||}tj||dd}|S)Nrdimr)boolsumtorchsplittolist)r#r$ bool_mask valid_lengthsselected split_resultr0d/home/ubuntu/vllm_env/lib/python3.10/site-packages/diffusers/modular_pipelines/qwenimage/encoders.py_extract_masked_hidden1s  r2promptprompt_template_encode prompt_template_encode_start_idxtokenizer_max_lengthdevicecst|tr|gn|}||fdd|D}|||dddd|}||j|jdd} | jd} t| |j} fdd| D} d d| D} td d| Dt fd d| D} t fd d| D}| j|d } | |fS)Ncg|]}|qSr0format.0etemplater0r1 Fz*get_qwen_prompt_embeds..Tpt) max_lengthpadding truncationreturn_tensors) input_idsattention_maskoutput_hidden_statescg|]}|dqSNr0r<drop_idxr0r1rARcS&g|]}tj|dtj|jdqSr)dtyper8r)onessizelongr8r<r0r0r1rAS&cSg|]}|dqSrrVr<r0r0r1rATrBc 2g|]}t|||d|dgqSrrr)cat new_zerosrVr=u max_seq_lenr0r1rAV2c *g|]}t|||dgqSrZr^rarcr0r1rAY*r8) isinstancestrtorHrIr#r2maxr)stack) text_encoder tokenizerr4r5r6r7r8txt txt_tokensencoder_hidden_statesr#split_hidden_statesattn_mask_list prompt_embedsencoder_attention_maskr0rOrdr@r1get_qwen_prompt_embeds9s6    rximagecst|tr|gn|}||fdd|D}|||ddd|}||j|j|j|jdd} | jd} t| |j} fdd| D} d d| D} t d d| Dt fd d| D} t fd d| D}| j|d } | |fS)Ncr9r0r:r<r?r0r1rAnrBz/get_qwen_prompt_embeds_edit..TrCtextimagesrErGrHrI pixel_valuesimage_grid_thwrJrKcrLrMr0r<rNr0r1rArPcSrQrRrTr<r0r0r1rArXcSrYrZr[r<r0r0r1rArBc r\r]r^rarcr0r1rArec rfrZr^rarcr0r1rArgrh) rirjrkrHrIr~rr#r2rlr)rm)rn processorr4ryr5r6r8rp model_inputsoutputsr#rsrtrurvr0rwr1get_qwen_prompt_embeds_editas@    rimg_template_encodecs8t|tr|gn|}t|tr$dt|D] \}} ||d7qn |dur.|dnd||fdd|D} || |ddd|} || j| j| j| j dd} | j d } t | | j}fd d|D}d d|D}t d d|Dt fd d|D}t fdd|D}|j|d}||fS)Nrcsg|] }|qSr0r:r<)base_img_promptr@r0r1rAsz4get_qwen_prompt_embeds_edit_plus..TrCrzr}rKcrLrMr0r<rNr0r1rArPcSrQrRrTr<r0r0r1rArXcSrYrZr[r<r0r0r1rArBc r\r]r^rarcr0r1rArec rfrZr^rarcr0r1rArgrh)rirjlist enumerater;rkrHrIr~rr#r2rlr)rm)rnrr4ryr5rr6r8iimgrprrr#rsrtrurvr0)rrOrdr@r1 get_qwen_prompt_embeds_edit_plussP     rsampleencoder_output generator sample_modecCsRt|dr|dkr|j|St|dr|dkr|jSt|dr%|jStd)N latent_distrargmaxlatentsz3Could not access latents of provided encoder_output)hasattrrrmoderAttributeError)rrrr0r0r1retrieve_latentss   rrvaerSlatent_channelsc sttjstdtddkrdndkr,tddj||dttrQfdd t j d D}tj |d d }n t  d }tjjd |d d d |j|j}tjjd |d d d |j|j} ||| }|S)Nz#Expected image to be a tensor, got .rz Expected image dims 4 or 5, got )r8rSc s0g|]}t||d|dqS)rrr)rencode)r=rrryrrr0r1rAs"z$encode_vae_image..rr%rr)rir)Tensor ValueErrortyper& unsqueezerkrrangeshaper_rrtensorconfig latents_meanviewr8rS latents_std) ryrrr8rSrr image_latentsrrr0rr1encode_vae_images.           rc@eZdZdZdZedefddZedee fddZ edee fdd Z edee fd d Zed ed efddZdS)QwenImageEditResizeStepa Image Resize step that resize the image to target area while maintaining the aspect ratio. Components: image_resize_processor (`VaeImageProcessor`) Inputs: image (`Image | list`): Reference image(s) for denoising. Can be a single image or list of images. Outputs: resized_image (`list`): The resized images qwenimage-editreturncCdS)NzZImage Resize step that resize the image to target area while maintaining the aspect ratio.r0selfr0r0r1 description*z#QwenImageEditResizeStep.descriptioncCtdttddiddgSNimage_resize_processorvae_scale_factorr from_configrdefault_creation_methodrr rrr0r0r1expected_components. z+QwenImageEditResizeStep.expected_componentscC tdgSNryrr@rr0r0r1inputs9 zQwenImageEditResizeStep.inputscCtdttjjddgSN resized_imagezThe resized imagesname type_hintrrrPILImagerr0r0r1intermediate_outputs=  z,QwenImageEditResizeStep.intermediate_outputs componentsstatec s||}|j}t|stdt|t|r|g}|dj\}}td||\}fdd|D}||_| |||fS)N/Images must be image or list of images but are rcg|] }jj|dqSheightwidthrresizer=rycalculated_heightcalculated_widthrr0r1rAVz4QwenImageEditResizeStep.__call__..) get_block_stateryr rrr rVrrset_block_state) rrr block_stater| image_width image_height_resized_imagesr0rr1__call__Gs  z QwenImageEditResizeStep.__call__N__name__ __module__ __qualname____doc__ model_namepropertyrjrrrrrrrrr)no_gradrrrr0r0r0r1rs  rc@seZdZdZdZedefddZedee fddZ edee fdd Z edee fd d Zed efd dZededefddZdS)QwenImageLayeredResizeStepaE Image Resize step that resize the image to a target area (defined by the resolution parameter from user) while maintaining the aspect ratio. Components: image_resize_processor (`VaeImageProcessor`) Inputs: image (`Image | list`): Reference image(s) for denoising. Can be a single image or list of images. resolution (`int`, *optional*, defaults to 640): The target area to resize the image to, can be 1024 or 640 Outputs: resized_image (`list`): The resized images qwenimage-layeredrcCr)NzImage Resize step that resize the image to a target area (defined by the resolution parameter from user) while maintaining the aspect ratio.r0rr0r0r1rvrz&QwenImageLayeredResizeStep.descriptioncCrrrrr0r0r1rzrz.QwenImageLayeredResizeStep.expected_componentscCstdtddtddgS)Nry resolutionz:The target area to resize the image to, can be 1024 or 640rdefaultrr)rr@intrr0r0r1rsz!QwenImageLayeredResizeStep.inputscCrrrrr0r0r1rrz/QwenImageLayeredResizeStep.intermediate_outputsrcCs|dvr td|dS)N)r3rz&Resolution must be 1024 or 640 but is rrr0r0r1 check_inputssz'QwenImageLayeredResizeStep.check_inputsrrc s||}|j|jd|j}t|stdt|t|r#|g}|dj\}}|j|j}t |||\}fdd|D} | |_ | |||fS)Nrrrcrrrrrr0r1rArz7QwenImageLayeredResizeStep.__call__..) rrrryr rrr rVrrr) rrrrr|rr target_arearrr0rr1rs   z#QwenImageLayeredResizeStep.__call__N)rrrrrrrjrrrrrrrr staticmethodrrr)rrrrr0r0r0r1ras   rc@r)QwenImageEditPlusResizeStepa Resize images for QwenImage Edit Plus pipeline. Produces two outputs: resized_image (1024x1024) for VAE encoding, resized_cond_image (384x384) for VL text encoding. Each image is resized independently based on its own aspect ratio. Components: image_resize_processor (`VaeImageProcessor`) Inputs: image (`Image | list`): Reference image(s) for denoising. Can be a single image or list of images. Outputs: resized_image (`list`): Images resized to 1024x1024 target area for VAE encoding resized_cond_image (`list`): Images resized to 384x384 target area for VL text encoding qwenimage-edit-plusrcC dS)NzResize images for QwenImage Edit Plus pipeline. Produces two outputs: resized_image (1024x1024) for VAE encoding, resized_cond_image (384x384) for VL text encoding. Each image is resized independently based on its own aspect ratio.r0rr0r0r1rz'QwenImageEditPlusResizeStep.descriptioncCrrrrr0r0r1rrz/QwenImageEditPlusResizeStep.expected_componentscCrrrrr0r0r1rs z"QwenImageEditPlusResizeStep.inputscCs,tdttjjddtdttjjddgS)Nrz8Images resized to 1024x1024 target area for VAE encodingrresized_cond_imagez:Images resized to 384x384 target area for VL text encodingrrr0r0r1rs  z0QwenImageEditPlusResizeStep.intermediate_outputsrrcCs||}|j}t|stdt|t|r|g}g}g}|D]3}|j\}} td|| \} } } ||j j || | dtd|| \} }} ||j j ||| dq"||_ ||_ | ||||fS)Nrrri@)rryr rrr rVrappendrrrrr)rrrrr|rresized_cond_imagesryrr vae_width vae_heightrvl_width vl_heightr0r0r1rs(   z$QwenImageEditPlusResizeStep.__call__Nrr0r0r0r1rs rcseZdZdZdZfddZedefddZede e fdd Z ede e fd d Z ede efd d ZedededefddZZS)"QwenImageLayeredGetImagePromptStepae Auto-caption step that generates a text prompt from the input image if none is provided. Uses the VL model (text_encoder) to generate a description of the image. If prompt is already provided, this step passes through unchanged. Components: text_encoder (`Qwen2_5_VLForConditionalGeneration`) processor (`Qwen2VLProcessor`) Inputs: prompt (`str`, *optional*): The prompt or prompts to guide image generation. resized_image (`Image`): The image to generate caption from, should be resized use the resize step use_en_prompt (`bool`, *optional*, defaults to False): Whether to use English prompt template Outputs: prompt (`str`): The prompt or prompts to guide image generation. If not provided, updated using image caption rct|_t|_tdSrM)r image_caption_prompt_enrimage_caption_prompt_cnsuper__init__r __class__r0r1r ;z+QwenImageLayeredGetImagePromptStep.__init__rcCr)NzAuto-caption step that generates a text prompt from the input image if none is provided. Uses the VL model (text_encoder) to generate a description of the image. If prompt is already provided, this step passes through unchanged.r0rr0r0r1r@rz.QwenImageLayeredGetImagePromptStep.descriptioncCstdttdtgS)Nrnr)rrrrr0r0r1rHsz6QwenImageLayeredGetImagePromptStep.expected_componentscCs0tjdddtddtjjddtddtd d gS) Nr4F)requiredrTzIThe image to generate caption from, should be resized use the resize steprrrr use_en_promptz&Whether to use English prompt templater)rr@rrr'rr0r0r1rOs z)QwenImageLayeredGetImagePromptStep.inputscCstdtddgS)Nr4z]The prompt or prompts to guide image generation. If not provided, updated using image captionr)rrjrr0r0r1rcs z7QwenImageLayeredGetImagePromptStep.intermediate_outputsrrc Cs||}|j}|jdus|jdks|jdkrV|jr|j}n|j}|j||jddd|}|j j d i|ddi}dd t |j |D}|jj |dd d d } | |_|||||fS)Nr TrCrzmax_new_tokensicSs g|] \}}|t|dqSrM)len)r=in_idsout_idsr0r0r1rAsz?QwenImageLayeredGetImagePromptStep.__call__..F)skip_special_tokensclean_up_tokenization_spacesrr0)r_execution_devicer4rr r rrrkrngenerateziprH batch_decodestripr) rrrrr8caption_promptr generated_idsgenerated_ids_trimmed output_textr0r0r1rms4    z+QwenImageLayeredGetImagePromptStep.__call__)rrrrrr rrjrrrrrrrrr)rrrr __classcell__r0r0r r1r#s  rceZdZdZdZfddZedefddZede e fdd Z ede e fd d Z ede efd d ZeddZededefddZZS)QwenImageTextEncoderStepa Text Encoder step that generates text embeddings to guide the image generation. Components: text_encoder (`Qwen2_5_VLForConditionalGeneration`): The text encoder to use tokenizer (`Qwen2Tokenizer`): The tokenizer to use guider (`ClassifierFreeGuidance`) Inputs: prompt (`str`): The prompt or prompts to guide image generation. negative_prompt (`str`, *optional*): The prompt or prompts not to guide the image generation. max_sequence_length (`int`, *optional*, defaults to 1024): Maximum sequence length for prompt encoding. Outputs: prompt_embeds (`Tensor`): The prompt embeddings. prompt_embeds_mask (`Tensor`): The encoder attention mask. negative_prompt_embeds (`Tensor`): The negative prompt embeddings. negative_prompt_embeds_mask (`Tensor`): The negative prompt embeddings mask. qwenimagecs t|_t|_d|_tdS)Nr3)r!r5r"r6r7r r rr r0r1r z!QwenImageTextEncoderStep.__init__rcCr)NzOText Encoder step that generates text embeddings to guide the image generation.r0rr0r0r1rrz$QwenImageTextEncoderStep.descriptioncCs2tdtddtdtddtdttddid d gS) NrnzThe text encoder to use)rrozThe tokenizer to useguiderguidance_scale@rr)rrrrrrr0r0r1rs   z,QwenImageTextEncoderStep.expected_componentscCs tdtdtjdddgS)Nr4negative_promptmax_sequence_lengthr3)rrrr0r0r1rs zQwenImageTextEncoderStep.inputscC$tdtdtdtdgSNruprompt_embeds_masknegative_prompt_embedsnegative_prompt_embeds_maskrr@rr0r0r1r z-QwenImageTextEncoderStep.intermediate_outputscCszt|tst|tstdt||dur*t|ts*t|ts*tdt||dur9|dkr;td|dSdS)N2`prompt` has to be of type `str` or `list` but is ;`negative_prompt` has to be of type `str` or `list` but is r3z9`max_sequence_length` cannot be greater than 1024 but is rirjrrr)r4r+r,r0r0r1rsz%QwenImageTextEncoderStep.check_inputsrrc Cs||}|j}||j|j|jt|j|j|j|j |j |j |d\|_ |_ |j ddd|jf|_ |j ddd|jf|_ d|_d|_|jr}|jpNd}t|j|j||j |j |j |d\|_|_|jddd|jf|_|jddd|jf|_|||||fS)N)r4r5r6r7r8r)rrrr4r+r,rxrnror5r6r7rur/r0r1requires_unconditional_embedsrrrrrr8r+r0r0r1rsF     z!QwenImageTextEncoderStep.__call__rrrrrr rrjrrrrrrrrrrr)rrrrr#r0r0r r1r%s    r%cr$)QwenImageEditTextEncoderStepa Text Encoder step that processes both prompt and image together to generate text embeddings for guiding image generation. Components: text_encoder (`Qwen2_5_VLForConditionalGeneration`) processor (`Qwen2VLProcessor`) guider (`ClassifierFreeGuidance`) Inputs: prompt (`str`): The prompt or prompts to guide image generation. negative_prompt (`str`, *optional*): The prompt or prompts not to guide the image generation. resized_image (`Image`): The image prompt to encode, should be resized using resize step Outputs: prompt_embeds (`Tensor`): The prompt embeddings. prompt_embeds_mask (`Tensor`): The encoder attention mask. negative_prompt_embeds (`Tensor`): The negative prompt embeddings. negative_prompt_embeds_mask (`Tensor`): The negative prompt embeddings mask. r&crrM)rr5rr6r r rr r0r1r 4rz%QwenImageEditTextEncoderStep.__init__rcCr)NzyText Encoder step that processes both prompt and image together to generate text embeddings for guiding image generation.r0rr0r0r1r9rz(QwenImageEditTextEncoderStep.descriptioncC*tdttdttdttddiddgSNrnrr(r)r*rrrrrrrrr0r0r1r= z0QwenImageEditTextEncoderStep.expected_componentscCs&tdtdtddtjjddgS)Nr4r+rTz?The image prompt to encode, should be resized using resize steprrr@rrrr0r0r1rJsz#QwenImageEditTextEncoderStep.inputscCr-r.r2rr0r0r1rWr3z1QwenImageEditTextEncoderStep.intermediate_outputscC`t|tst|tstdt||dur*t|ts,t|ts.tdt|dSdSdSNr4r5r6r4r+r0r0r1r` z)QwenImageEditTextEncoderStep.check_inputsrrc Cs||}||j|j|j}t|j|j|j|j|j |j |d\|_ |_ d|_ d|_|jrG|jp2d}t|j|j||j|j |j |d\|_ |_|||||fS)N)r4ryr5r6r8r)rrr4r+rrrnrrr5r6rur/r0r1r7rr8r0r0r1rls6    z%QwenImageEditTextEncoderStep.__call__r9r0r0r r1r:s      r:cr$) QwenImageEditPlusTextEncoderStepa? Text Encoder step for QwenImage Edit Plus that processes prompt and multiple images together to generate text embeddings for guiding image generation. Components: text_encoder (`Qwen2_5_VLForConditionalGeneration`) processor (`Qwen2VLProcessor`) guider (`ClassifierFreeGuidance`) Inputs: prompt (`str`): The prompt or prompts to guide image generation. negative_prompt (`str`, *optional*): The prompt or prompts not to guide the image generation. resized_cond_image (`Tensor`): The image(s) to encode, can be a single image or list of images, should be resized to 384x384 using resize step Outputs: prompt_embeds (`Tensor`): The prompt embeddings. prompt_embeds_mask (`Tensor`): The encoder attention mask. negative_prompt_embeds (`Tensor`): The negative prompt embeddings. negative_prompt_embeds_mask (`Tensor`): The negative prompt embeddings mask. rcs t|_t|_t|_tdSrM)rr5rrrr6r r rr r0r1r r'z)QwenImageEditPlusTextEncoderStep.__init__rcCr)NzText Encoder step for QwenImage Edit Plus that processes prompt and multiple images together to generate text embeddings for guiding image generation.r0rr0r0r1rrz,QwenImageEditPlusTextEncoderStep.descriptioncCr;r<r=rr0r0r1rr>z4QwenImageEditPlusTextEncoderStep.expected_componentscCs$tdtdtddtjddgS)Nr4r+rTzoThe image(s) to encode, can be a single image or list of images, should be resized to 384x384 using resize stepr)rr@r)rrr0r0r1rsz'QwenImageEditPlusTextEncoderStep.inputscCr-r.r2rr0r0r1rr3z5QwenImageEditPlusTextEncoderStep.intermediate_outputscCr@rAr6rBr0r0r1rrCz-QwenImageEditPlusTextEncoderStep.check_inputsrrc Cs||}||j|j|j}t|j|j|j|j|j |j |j |d\|_ |_ d|_d|_|jrK|jp4d}t|j|j||j|j |j |j |d\|_|_|||||fS)N)r4ryr5rr6r8r)rrr4r+rrrnrrr5rr6rur/r0r1r7rr8r0r0r1rs<    z)QwenImageEditPlusTextEncoderStep.__call__r9r0r0r r1rDs      rDc@eZdZdZdZedefddZedee fddZ edee fdd Z edee fd d Zed d ZededefddZdS)&QwenImageInpaintProcessImagesInputStepa Image Preprocess step for inpainting task. This processes the image and mask inputs together. Images will be resized to the given height and width. Components: image_mask_processor (`InpaintProcessor`) Inputs: mask_image (`Image`): Mask image for inpainting. image (`Image | list`): Reference image(s) for denoising. Can be a single image or list of images. height (`int`, *optional*): The height in pixels of the generated image. width (`int`, *optional*): The width in pixels of the generated image. padding_mask_crop (`int`, *optional*): Padding for mask cropping in inpainting. Outputs: processed_image (`Tensor`): The processed image processed_mask_image (`Tensor`): The processed mask image mask_overlay_kwargs (`dict`): The kwargs for the postprocess step to apply the mask overlay r&rcCr)NzImage Preprocess step for inpainting task. This processes the image and mask inputs together. Images will be resized to the given height and width.r0rr0r0r1r9rz2QwenImageInpaintProcessImagesInputStep.descriptioncCrNimage_mask_processorrrrrrrrrr0r0r1r=rz:QwenImageInpaintProcessImagesInputStep.expected_componentscCs,tdtdtdtdtdgS)N mask_imageryrrpadding_mask_croprrr0r0r1rHs z-QwenImageInpaintProcessImagesInputStep.inputscC,tdtjddtdtjddtdtddgSNprocessed_imageThe processed imagerprocessed_mask_imagezThe processed mask imagemask_overlay_kwargsz=The kwargs for the postprocess step to apply the mask overlayrr)rdictrr0r0r1rRs z;QwenImageInpaintProcessImagesInputStep.intermediate_outputscCh|dur||ddkrtd|dd||dur0||ddkr2td|dd|dSdSNrrzHeight must be divisible by z but is zWidth must be divisible by rrrrr0r0r1rf z3QwenImageInpaintProcessImagesInputStep.check_inputsrrcCst||}|j|j|j|jd|jp|j}|jp|j}|jj|j |j |||j d\|_ |_ |_|||||fS)NrVryr$rrrK)rrrrrdefault_height default_widthrH preprocessryrJrKrNrPrQrrrrrrrr0r0r1rns     z/QwenImageInpaintProcessImagesInputStep.__call__Nrrrrrrrjrrrrrrrrrrr)rrrrr0r0r0r1rFs   rFc@r)*QwenImageEditInpaintProcessImagesInputStepa Image Preprocess step for inpainting task. This processes the image and mask inputs together. Images should be resized first. Components: image_mask_processor (`InpaintProcessor`) Inputs: mask_image (`Image`): Mask image for inpainting. resized_image (`Image`): The resized image. should be generated using a resize step padding_mask_crop (`int`, *optional*): Padding for mask cropping in inpainting. Outputs: processed_image (`Tensor`): The processed image processed_mask_image (`Tensor`): The processed mask image mask_overlay_kwargs (`dict`): The kwargs for the postprocess step to apply the mask overlay rrcCr)Nz}Image Preprocess step for inpainting task. This processes the image and mask inputs together. Images should be resized first.r0rr0r0r1rrz6QwenImageEditInpaintProcessImagesInputStep.descriptioncCrrGrIrr0r0r1rrz>QwenImageEditInpaintProcessImagesInputStep.expected_componentscCs&tdtddtjjddtdgS)NrJrT:The resized image. should be generated using a resize steprrKr?rr0r0r1rsz1QwenImageEditInpaintProcessImagesInputStep.inputscCrLrMrRrr0r0r1rsz?QwenImageEditInpaintProcessImagesInputStep.intermediate_outputsrrcCsV||}|jdj\}}|jj|j|j|||jd\|_|_|_ | ||||fS)NrrX) rrrVrHr[rJrKrNrPrQrrrrrrrr0r0r1rs  z3QwenImageEditInpaintProcessImagesInputStep.__call__Nrr0r0r0r1r^s  r^c@rE)QwenImageProcessImagesInputStepa: Image Preprocess step. will resize the image to the given height and width. Components: image_processor (`VaeImageProcessor`) Inputs: image (`Image | list`): Reference image(s) for denoising. Can be a single image or list of images. height (`int`, *optional*): The height in pixels of the generated image. width (`int`, *optional*): The width in pixels of the generated image. Outputs: processed_image (`Tensor`): The processed image r&rcCr)NzKImage Preprocess step. will resize the image to the given height and width.r0rr0r0r1rrz+QwenImageProcessImagesInputStep.descriptioncCrNimage_processorrrrrrrr0r0r1rrz3QwenImageProcessImagesInputStep.expected_componentscCstdtdtdgS)Nryrrrrr0r0r1rsz&QwenImageProcessImagesInputStep.inputscCtdtjddgSNrNrOrrr)rrr0r0r1r z4QwenImageProcessImagesInputStep.intermediate_outputscCrTrUrrVr0r0r1rrWz,QwenImageProcessImagesInputStep.check_inputsrrcCsb||}|j|j|j|jd|jp|j}|jp|j}|jj|j ||d|_ | ||||fS)NrVryrr) rrrrrrYrZrcr[ryrNrr\r0r0r1r"s     z(QwenImageProcessImagesInputStep.__call__Nr]r0r0r0r1ras  rac@r)#QwenImageEditProcessImagesInputStepaW Image Preprocess step. Images needs to be resized first. Components: image_processor (`VaeImageProcessor`) Inputs: resized_image (`list`): The resized image. should be generated using a resize step Outputs: processed_image (`Tensor`): The processed image rrcCr)Nz8Image Preprocess step. Images needs to be resized first.r0rr0r0r1rIrz/QwenImageEditProcessImagesInputStep.descriptioncCrrbrrr0r0r1rMrz7QwenImageEditProcessImagesInputStep.expected_componentscCtddttjjddgSNrTr_rrrrrrr0r0r1rXrz*QwenImageEditProcessImagesInputStep.inputscCrdrerfrr0r0r1rcrgz8QwenImageEditProcessImagesInputStep.intermediate_outputsrrcCsD||}|jdj\}}|jj|j||d|_|||||fS)Nrrh)rrrVrcr[rNrr`r0r0r1rms  z,QwenImageEditProcessImagesInputStep.__call__Nrr0r0r0r1ri7s   ric@r)'QwenImageEditPlusProcessImagesInputStepa Image Preprocess step. Images can be resized first. If a list of images is provided, will return a list of processed images. Components: image_processor (`VaeImageProcessor`) Inputs: resized_image (`list`): The resized image. should be generated using a resize step Outputs: processed_image (`Tensor`): The processed image rrcCr)Nz|Image Preprocess step. Images can be resized first. If a list of images is provided, will return a list of processed images.r0rr0r0r1rrz3QwenImageEditPlusProcessImagesInputStep.descriptioncCrrbrrr0r0r1rrz;QwenImageEditPlusProcessImagesInputStep.expected_componentscCrjrkrlrr0r0r1rrz.QwenImageEditPlusProcessImagesInputStep.inputscCrdrerfrr0r0r1rrgzs      ,    +  2  9  BI\fr{ m\TGWl