spandrel.architectures.GFPGAN.GFPGAN

class documentation

class GFPGAN(nn.Module): (source)

Constructor: GFPGAN(out_size, num_style_feat, channel_multiplier, decoder_load_path, ...)

The GFPGAN architecture: Unet + StyleGAN2 decoder with SFT. It is the clean version without custom compiled CUDA extensions used in StyleGAN2. Ref: GFP-GAN: Towards Real-World Blind Face Restoration with Generative Facial Prior.

Parameters
out_size	The spatial size of outputs.
num_style_feat	Channel number of style features. Default: 512.
channel_multiplier	Channel multiplier for large networks of StyleGAN2. Default: 2.
decoder_load_path	The path to the pre-trained decoder model (usually, the StyleGAN2). Default: None.
fix_decoder	Whether to fix the decoder. Default: True.
num_mlp	Layer number of MLP style layers. Default: 8.
input_is_latent	Whether input is latent style. Default: False.
different_w	Whether to use different latent w for different layers. Default: False.
narrow	The narrow ratio for channels. Default: 1.
sft_half	Whether to apply SFT on half of the input channels. Default: False.

Method	`__init__`	Undocumented
Method	`forward`	Forward function for GFPGANv1Clean.
Class Variable	`hyperparameters`	Undocumented
Instance Variable	`condition_scale`	Undocumented
Instance Variable	`condition_shift`	Undocumented
Instance Variable	`conv_body_down`	Undocumented
Instance Variable	`conv_body_first`	Undocumented
Instance Variable	`conv_body_up`	Undocumented
Instance Variable	`different_w`	Undocumented
Instance Variable	`final_conv`	Undocumented
Instance Variable	`final_linear`	Undocumented
Instance Variable	`input_is_latent`	Undocumented
Instance Variable	`log_size`	Undocumented
Instance Variable	`num_style_feat`	Undocumented
Instance Variable	`stylegan_decoder`	Undocumented
Instance Variable	`toRGB`	Undocumented

def __init__(self, *, out_size=512, num_style_feat=512, channel_multiplier=2, decoder_load_path=None, fix_decoder=False, num_mlp=8, input_is_latent=True, different_w=True, narrow=1, sft_half=True): (source) ¶

Undocumented

def forward(self, x, return_latents=False, return_rgb=True, randomize_noise=True, **kwargs): (source) ¶

Forward function for GFPGANv1Clean.

Parameters
x:`Tensor`	Input images.
return_latents:`bool`	Whether to return style latents. Default: False.
return_rgb:`bool`	Whether return intermediate rgb images. Default: True.
randomize_noise:`bool`	Randomize noise, used when 'noise' is False. Default: True.
**kwargs	Undocumented