a d&@sdZddlmZmZmZddlZddlmZgdZGdddej Z ej ej ej ej ddd Z Gd d d ej Z ej eej efeej efej dd d Zdej eeeej dddZdS)z:Module containing functionals for intensity normalisation.)ListTupleUnionN) normalizenormalize_min_max denormalize Normalize DenormalizecsreZdZdZeejeee eefeejeee eefddfdd Z ejejdddZ d d Z Z S) racNormalize a tensor image with mean and standard deviation. .. math:: \text{input[channel] = (input[channel] - mean[channel]) / std[channel]} Where `mean` is :math:`(M_1, ..., M_n)` and `std` :math:`(S_1, ..., S_n)` for `n` channels, Args: mean: Mean for each channel. std: Standard deviations for each channel. Shape: - Input: Image tensor of size :math:`(*, C, ...)`. - Output: Normalised tensor with same size as input :math:`(*, C, ...)`. Examples: >>> x = torch.rand(1, 4, 3, 3) >>> out = Normalize(0.0, 255.)(x) >>> out.shape torch.Size([1, 4, 3, 3]) >>> x = torch.rand(1, 4, 3, 3) >>> mean = torch.zeros(4) >>> std = 255. * torch.ones(4) >>> out = Normalize(mean, std)(x) >>> out.shape torch.Size([1, 4, 3, 3]) Nmeanstdreturncsvtt|tr t|g}t|tr6t|g}t|ttfrNt|}t|ttfrft|}||_||_ dSN) super__init__ isinstancefloattorchtensortuplelistr r selfr r  __class__a/var/www/html/stable-diffusion-webui/venv/lib/python3.9/site-packages/kornia/enhance/normalize.pyr)s       zNormalize.__init__inputr cCst||j|jSr)rr r rrrrrforward?szNormalize.forwardcCs"d|jd|jd}|jj|SNz(mean=z, std=)r r r__name__rreprrrr__repr__BszNormalize.__repr__)r$ __module__ __qualname____doc__rrTensorrrrrr r' __classcell__rrrrr sr)datar r r cCs|j}t|jdks"|jddkr0||d}t|jdksL|jddkrZ||d}|jr|jddkr|jd|jdkr|jdd|jddkrtd|jd|jd|jr|jddkr|jd|jdkr|jdd|jddkrtd|jd|jdtj||j|jd }tj||j|jd }|jr`|d dddf}|jrz|d dddf}||d|dd ||}||S) azNormalize an image/video tensor with mean and standard deviation. .. math:: \text{input[channel] = (input[channel] - mean[channel]) / std[channel]} Where `mean` is :math:`(M_1, ..., M_n)` and `std` :math:`(S_1, ..., S_n)` for `n` channels, Args: data: Image tensor of size :math:`(B, C, *)`. mean: Mean for each channel. std: Standard deviations for each channel. Return: Normalised tensor with same size as input :math:`(B, C, *)`. Examples: >>> x = torch.rand(1, 4, 3, 3) >>> out = normalize(x, torch.tensor([0.0]), torch.tensor([255.])) >>> out.shape torch.Size([1, 4, 3, 3]) >>> x = torch.rand(1, 4, 3, 3) >>> mean = torch.zeros(4) >>> std = 255. * torch.ones(4) >>> out = normalize(x, mean, std) >>> out.shape torch.Size([1, 4, 3, 3]) rN5mean length and number of channels do not match. Got  and .4std length and number of channels do not match. Got devicedtype.) shapelenexpand ValueErrorr as_tensorr5r6viewr-r r r8outrrrrGs&04rcsZeZdZdZeejefeejefddfdd ZejejdddZ d d Z Z S) r awDenormalize a tensor image with mean and standard deviation. .. math:: \text{input[channel] = (input[channel] * std[channel]) + mean[channel]} Where `mean` is :math:`(M_1, ..., M_n)` and `std` :math:`(S_1, ..., S_n)` for `n` channels, Args: mean: Mean for each channel. std: Standard deviations for each channel. Shape: - Input: Image tensor of size :math:`(*, C, ...)`. - Output: Denormalised tensor with same size as input :math:`(*, C, ...)`. Examples: >>> x = torch.rand(1, 4, 3, 3) >>> out = Denormalize(0.0, 255.)(x) >>> out.shape torch.Size([1, 4, 3, 3]) >>> x = torch.rand(1, 4, 3, 3, 3) >>> mean = torch.zeros(1, 4) >>> std = 255. * torch.ones(1, 4) >>> out = Denormalize(mean, std)(x) >>> out.shape torch.Size([1, 4, 3, 3, 3]) Nr cst||_||_dSr)rrr r rrrrrs zDenormalize.__init__rcCst||j|jSr)rr r rrrrr szDenormalize.forwardcCs"d|jd|jd}|jj|Sr!r#r%rrrr'szDenormalize.__repr__) r$r(r)r*rrr+rrr r'r,rrrrr s*r cCs|j}t|tr.tj|g|d|j|jd}t|trVtj|g|d|j|jd}t|tjsttdt |t|tjstdt |t|tjstdt ||jr|jddkr|jd|jdkr|jdd |jdd krt d |jd |jd |jr||jddkr||jd|jdkr||jdd |jdd kr|t d |jd |jd tj ||j|jd}tj ||j|jd}|jr|ddddf}|jr|ddddf}| |d|dd||}| |S)aoDenormalize an image/video tensor with mean and standard deviation. .. math:: \text{input[channel] = (input[channel] * std[channel]) + mean[channel]} Where `mean` is :math:`(M_1, ..., M_n)` and `std` :math:`(S_1, ..., S_n)` for `n` channels, Args: input: Image tensor of size :math:`(B, C, *)`. mean: Mean for each channel. std: Standard deviations for each channel. Return: Denormalised tensor with same size as input :math:`(B, C, *)`. Examples: >>> x = torch.rand(1, 4, 3, 3) >>> out = denormalize(x, 0.0, 255.) >>> out.shape torch.Size([1, 4, 3, 3]) >>> x = torch.rand(1, 4, 3, 3, 3) >>> mean = torch.zeros(1, 4) >>> std = 255. * torch.ones(1, 4) >>> out = denormalize(x, mean, std) >>> out.shape torch.Size([1, 4, 3, 3, 3]) r.r4zdata should be a tensor. Got z(mean should be a tensor or a float. Got z%std should be a tensor or float. Got rNr/r0r1r2r3.r7) r8rrrrr5r6r+ TypeErrortyper;r<r=r>rrrrs2     44r?ư>)xmin_valmax_valepsr c Cs t|tjs tdt|dt|ts>tdt|dt|ts\tdt|dt|jdkr|td|jd|j}|d|d}}| ||d  d d ||d}| ||d  d d ||d}||| ||d |||||} | |S) axNormalise an image/video tensor by MinMax and re-scales the value between a range. The data is normalised using the following formulation: .. math:: y_i = (b - a) * \frac{x_i - \text{min}(x)}{\text{max}(x) - \text{min}(x)} + a where :math:`a` is :math:`\text{min_val}` and :math:`b` is :math:`\text{max_val}`. Args: x: The image tensor to be normalised with shape :math:`(B, C, *)`. min_val: The minimum value for the new range. max_val: The maximum value for the new range. eps: Float number to avoid zero division. Returns: The normalised image tensor with same shape as input :math:`(B, C, *)`. Example: >>> x = torch.rand(1, 5, 3, 3) >>> x_norm = normalize_min_max(x, min_val=-1., max_val=1.) >>> x_norm.min() tensor(-1.) >>> x_norm.max() tensor(1.0000) zdata should be a tensor. Got: r2z"'min_val' should be a float. Got: z'b' should be a float. Got: z/Input shape must be at least a 3d tensor. Got: rr.r7) rrr+rArBrr9r8r;r=minmax) rFrGrHrIr8BCZx_minZx_maxZx_outrrrrs   ""*r)rCrDrE)r*typingrrrrZtorch.nnnn__all__Modulerr+rr rrrrrrrs <:-,E