Merge branch 'master' into flipflop-stream

Author: Kosinkadink

README.md

@@ -206,14 +206,32 @@ Put your SD checkpoints (the huge ckpt/safetensors files) in: models/checkpoints
 Put your VAE in: models/vae
 
 
-### AMD GPUs (Linux only)
+### AMD GPUs (Linux)
+
 AMD users can install rocm and pytorch with pip if you don't have it already installed, this is the command to install the stable version:
 
 ```pip install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/rocm6.4```
 
-This is the command to install the nightly with ROCm 6.4 which might have some performance improvements:
+This is the command to install the nightly with ROCm 7.0 which might have some performance improvements:
+
+```pip install --pre torch torchvision torchaudio --index-url https://download.pytorch.org/whl/nightly/rocm7.0```
+
+
+### AMD GPUs (Experimental: Windows and Linux), RDNA 3, 3.5 and 4 only.
+
+These have less hardware support than the builds above but they work on windows. You also need to install the pytorch version specific to your hardware.
+
+RDNA 3 (RX 7000 series):
+
+```pip install --pre torch torchvision torchaudio --index-url https://rocm.nightlies.amd.com/v2/gfx110X-dgpu/```
 
-```pip install --pre torch torchvision torchaudio --index-url https://download.pytorch.org/whl/nightly/rocm6.4```
+RDNA 3.5 (Strix halo/Ryzen AI Max+ 365):
+
+```pip install --pre torch torchvision torchaudio --index-url https://rocm.nightlies.amd.com/v2/gfx1151/```
+
+RDNA 4 (RX 9000 series):
+
+```pip install --pre torch torchvision torchaudio --index-url https://rocm.nightlies.amd.com/v2/gfx120X-all/```
 
 ### Intel GPUs (Windows and Linux)
 
@@ -270,12 +288,6 @@ You can install ComfyUI in Apple Mac silicon (M1 or M2) with any recent macOS ve
 
 > **Note**: Remember to add your models, VAE, LoRAs etc. to the corresponding Comfy folders, as discussed in [ComfyUI manual installation](#manual-install-windows-linux).
 
-#### DirectML (AMD Cards on Windows)
-
-This is very badly supported and is not recommended. There are some unofficial builds of pytorch ROCm on windows that exist that will give you a much better experience than this. This readme will be updated once official pytorch ROCm builds for windows come out.
-
-```pip install torch-directml``` Then you can launch ComfyUI with: ```python main.py --directml```
-
 #### Ascend NPUs
 
 For models compatible with Ascend Extension for PyTorch (torch_npu). To get started, ensure your environment meets the prerequisites outlined on the [installation](https://ascend.github.io/docs/sources/ascend/quick_install.html) page. Here's a step-by-step guide tailored to your platform and installation method:

comfy/ldm/ace/vae/music_dcae_pipeline.py

@@ -23,8 +23,6 @@ def __init__(self, source_sample_rate=None, dcae_config={}, vocoder_config={}):
         else:
             self.source_sample_rate = source_sample_rate
 
-        # self.resampler = torchaudio.transforms.Resample(source_sample_rate, 44100)
-
         self.transform = transforms.Compose([
             transforms.Normalize(0.5, 0.5),
         ])
@@ -37,10 +35,6 @@ def __init__(self, source_sample_rate=None, dcae_config={}, vocoder_config={}):
         self.scale_factor = 0.1786
         self.shift_factor = -1.9091
 
-    def load_audio(self, audio_path):
-        audio, sr = torchaudio.load(audio_path)
-        return audio, sr
-
     def forward_mel(self, audios):
         mels = []
         for i in range(len(audios)):
@@ -73,10 +67,8 @@ def encode(self, audios, audio_lengths=None, sr=None):
             latent = self.dcae.encoder(mel.unsqueeze(0))
             latents.append(latent)
         latents = torch.cat(latents, dim=0)
-        # latent_lengths = (audio_lengths / sr * 44100 / 512 / self.time_dimention_multiple).long()
         latents = (latents - self.shift_factor) * self.scale_factor
         return latents
-        # return latents, latent_lengths
 
     @torch.no_grad()
     def decode(self, latents, audio_lengths=None, sr=None):
@@ -91,17 +83,14 @@ def decode(self, latents, audio_lengths=None, sr=None):
             wav = self.vocoder.decode(mels[0]).squeeze(1)
 
             if sr is not None:
-                # resampler = torchaudio.transforms.Resample(44100, sr).to(latents.device).to(latents.dtype)
                 wav = torchaudio.functional.resample(wav, 44100, sr)
-                # wav = resampler(wav)
             else:
                 sr = 44100
             pred_wavs.append(wav)
 
         if audio_lengths is not None:
             pred_wavs = [wav[:, :length].cpu() for wav, length in zip(pred_wavs, audio_lengths)]
         return torch.stack(pred_wavs)
-        # return sr, pred_wavs
 
     def forward(self, audios, audio_lengths=None, sr=None):
         latents, latent_lengths = self.encode(audios=audios, audio_lengths=audio_lengths, sr=sr)

comfy/ldm/mmaudio/vae/__init__.py

@@ -0,0 +1,120 @@
+# Implementation adapted from https://github.com/EdwardDixon/snake under the MIT license.
+#   LICENSE is in incl_licenses directory.
+
+import torch
+from torch import nn, sin, pow
+from torch.nn import Parameter
+import comfy.model_management
+
+class Snake(nn.Module):
+    '''
+    Implementation of a sine-based periodic activation function
+    Shape:
+        - Input: (B, C, T)
+        - Output: (B, C, T), same shape as the input
+    Parameters:
+        - alpha - trainable parameter
+    References:
+        - This activation function is from this paper by Liu Ziyin, Tilman Hartwig, Masahito Ueda:
+        https://arxiv.org/abs/2006.08195
+    Examples:
+        >>> a1 = snake(256)
+        >>> x = torch.randn(256)
+        >>> x = a1(x)
+    '''
+    def __init__(self, in_features, alpha=1.0, alpha_trainable=True, alpha_logscale=False):
+        '''
+        Initialization.
+        INPUT:
+            - in_features: shape of the input
+            - alpha: trainable parameter
+            alpha is initialized to 1 by default, higher values = higher-frequency.
+            alpha will be trained along with the rest of your model.
+        '''
+        super(Snake, self).__init__()
+        self.in_features = in_features
+
+        # initialize alpha
+        self.alpha_logscale = alpha_logscale
+        if self.alpha_logscale:
+            self.alpha = Parameter(torch.empty(in_features))
+        else:
+            self.alpha = Parameter(torch.empty(in_features))
+
+        self.alpha.requires_grad = alpha_trainable
+
+        self.no_div_by_zero = 0.000000001
+
+    def forward(self, x):
+        '''
+        Forward pass of the function.
+        Applies the function to the input elementwise.
+        Snake ∶= x + 1/a * sin^2 (xa)
+        '''
+        alpha = comfy.model_management.cast_to(self.alpha, dtype=x.dtype, device=x.device).unsqueeze(0).unsqueeze(-1) # line up with x to [B, C, T]
+        if self.alpha_logscale:
+            alpha = torch.exp(alpha)
+        x = x + (1.0 / (alpha + self.no_div_by_zero)) * pow(sin(x * alpha), 2)
+
+        return x
+
+
+class SnakeBeta(nn.Module):
+    '''
+    A modified Snake function which uses separate parameters for the magnitude of the periodic components
+    Shape:
+        - Input: (B, C, T)
+        - Output: (B, C, T), same shape as the input
+    Parameters:
+        - alpha - trainable parameter that controls frequency
+        - beta - trainable parameter that controls magnitude
+    References:
+        - This activation function is a modified version based on this paper by Liu Ziyin, Tilman Hartwig, Masahito Ueda:
+        https://arxiv.org/abs/2006.08195
+    Examples:
+        >>> a1 = snakebeta(256)
+        >>> x = torch.randn(256)
+        >>> x = a1(x)
+    '''
+    def __init__(self, in_features, alpha=1.0, alpha_trainable=True, alpha_logscale=False):
+        '''
+        Initialization.
+        INPUT:
+            - in_features: shape of the input
+            - alpha - trainable parameter that controls frequency
+            - beta - trainable parameter that controls magnitude
+            alpha is initialized to 1 by default, higher values = higher-frequency.
+            beta is initialized to 1 by default, higher values = higher-magnitude.
+            alpha will be trained along with the rest of your model.
+        '''
+        super(SnakeBeta, self).__init__()
+        self.in_features = in_features
+
+        # initialize alpha
+        self.alpha_logscale = alpha_logscale
+        if self.alpha_logscale:
+            self.alpha = Parameter(torch.empty(in_features))
+            self.beta = Parameter(torch.empty(in_features))
+        else:
+            self.alpha = Parameter(torch.empty(in_features))
+            self.beta = Parameter(torch.empty(in_features))
+
+        self.alpha.requires_grad = alpha_trainable
+        self.beta.requires_grad = alpha_trainable
+
+        self.no_div_by_zero = 0.000000001
+
+    def forward(self, x):
+        '''
+        Forward pass of the function.
+        Applies the function to the input elementwise.
+        SnakeBeta ∶= x + 1/b * sin^2 (xa)
+        '''
+        alpha = comfy.model_management.cast_to(self.alpha, dtype=x.dtype, device=x.device).unsqueeze(0).unsqueeze(-1) # line up with x to [B, C, T]
+        beta = comfy.model_management.cast_to(self.beta, dtype=x.dtype, device=x.device).unsqueeze(0).unsqueeze(-1)
+        if self.alpha_logscale:
+            alpha = torch.exp(alpha)
+            beta = torch.exp(beta)
+        x = x + (1.0 / (beta + self.no_div_by_zero)) * pow(sin(x * alpha), 2)
+
+        return x

comfy/ldm/mmaudio/vae/activations.py

@@ -0,0 +1,157 @@
+import torch
+import torch.nn as nn
+import torch.nn.functional as F
+import math
+import comfy.model_management
+
+if 'sinc' in dir(torch):
+    sinc = torch.sinc
+else:
+    # This code is adopted from adefossez's julius.core.sinc under the MIT License
+    # https://adefossez.github.io/julius/julius/core.html
+    #   LICENSE is in incl_licenses directory.
+    def sinc(x: torch.Tensor):
+        """
+        Implementation of sinc, i.e. sin(pi * x) / (pi * x)
+        __Warning__: Different to julius.sinc, the input is multiplied by `pi`!
+        """
+        return torch.where(x == 0,
+                           torch.tensor(1., device=x.device, dtype=x.dtype),
+                           torch.sin(math.pi * x) / math.pi / x)
+
+
+# This code is adopted from adefossez's julius.lowpass.LowPassFilters under the MIT License
+# https://adefossez.github.io/julius/julius/lowpass.html
+#   LICENSE is in incl_licenses directory.
+def kaiser_sinc_filter1d(cutoff, half_width, kernel_size): # return filter [1,1,kernel_size]
+    even = (kernel_size % 2 == 0)
+    half_size = kernel_size // 2
+
+    #For kaiser window
+    delta_f = 4 * half_width
+    A = 2.285 * (half_size - 1) * math.pi * delta_f + 7.95
+    if A > 50.:
+        beta = 0.1102 * (A - 8.7)
+    elif A >= 21.:
+        beta = 0.5842 * (A - 21)**0.4 + 0.07886 * (A - 21.)
+    else:
+        beta = 0.
+    window = torch.kaiser_window(kernel_size, beta=beta, periodic=False)
+
+    # ratio = 0.5/cutoff -> 2 * cutoff = 1 / ratio
+    if even:
+        time = (torch.arange(-half_size, half_size) + 0.5)
+    else:
+        time = torch.arange(kernel_size) - half_size
+    if cutoff == 0:
+        filter_ = torch.zeros_like(time)
+    else:
+        filter_ = 2 * cutoff * window * sinc(2 * cutoff * time)
+        # Normalize filter to have sum = 1, otherwise we will have a small leakage
+        # of the constant component in the input signal.
+        filter_ /= filter_.sum()
+        filter = filter_.view(1, 1, kernel_size)
+
+    return filter
+
+
+class LowPassFilter1d(nn.Module):
+    def __init__(self,
+                 cutoff=0.5,
+                 half_width=0.6,
+                 stride: int = 1,
+                 padding: bool = True,
+                 padding_mode: str = 'replicate',
+                 kernel_size: int = 12):
+        # kernel_size should be even number for stylegan3 setup,
+        # in this implementation, odd number is also possible.
+        super().__init__()
+        if cutoff < -0.:
+            raise ValueError("Minimum cutoff must be larger than zero.")
+        if cutoff > 0.5:
+            raise ValueError("A cutoff above 0.5 does not make sense.")
+        self.kernel_size = kernel_size
+        self.even = (kernel_size % 2 == 0)
+        self.pad_left = kernel_size // 2 - int(self.even)
+        self.pad_right = kernel_size // 2
+        self.stride = stride
+        self.padding = padding
+        self.padding_mode = padding_mode
+        filter = kaiser_sinc_filter1d(cutoff, half_width, kernel_size)
+        self.register_buffer("filter", filter)
+
+    #input [B, C, T]
+    def forward(self, x):
+        _, C, _ = x.shape
+
+        if self.padding:
+            x = F.pad(x, (self.pad_left, self.pad_right),
+                      mode=self.padding_mode)
+        out = F.conv1d(x, comfy.model_management.cast_to(self.filter.expand(C, -1, -1), dtype=x.dtype, device=x.device),
+                       stride=self.stride, groups=C)
+
+        return out
+
+
+class UpSample1d(nn.Module):
+    def __init__(self, ratio=2, kernel_size=None):
+        super().__init__()
+        self.ratio = ratio
+        self.kernel_size = int(6 * ratio // 2) * 2 if kernel_size is None else kernel_size
+        self.stride = ratio
+        self.pad = self.kernel_size // ratio - 1
+        self.pad_left = self.pad * self.stride + (self.kernel_size - self.stride) // 2
+        self.pad_right = self.pad * self.stride + (self.kernel_size - self.stride + 1) // 2
+        filter = kaiser_sinc_filter1d(cutoff=0.5 / ratio,
+                                      half_width=0.6 / ratio,
+                                      kernel_size=self.kernel_size)
+        self.register_buffer("filter", filter)
+
+    # x: [B, C, T]
+    def forward(self, x):
+        _, C, _ = x.shape
+
+        x = F.pad(x, (self.pad, self.pad), mode='replicate')
+        x = self.ratio * F.conv_transpose1d(
+            x, comfy.model_management.cast_to(self.filter.expand(C, -1, -1), dtype=x.dtype, device=x.device), stride=self.stride, groups=C)
+        x = x[..., self.pad_left:-self.pad_right]
+
+        return x
+
+
+class DownSample1d(nn.Module):
+    def __init__(self, ratio=2, kernel_size=None):
+        super().__init__()
+        self.ratio = ratio
+        self.kernel_size = int(6 * ratio // 2) * 2 if kernel_size is None else kernel_size
+        self.lowpass = LowPassFilter1d(cutoff=0.5 / ratio,
+                                       half_width=0.6 / ratio,
+                                       stride=ratio,
+                                       kernel_size=self.kernel_size)
+
+    def forward(self, x):
+        xx = self.lowpass(x)
+
+        return xx
+
+class Activation1d(nn.Module):
+    def __init__(self,
+                 activation,
+                 up_ratio: int = 2,
+                 down_ratio: int = 2,
+                 up_kernel_size: int = 12,
+                 down_kernel_size: int = 12):
+        super().__init__()
+        self.up_ratio = up_ratio
+        self.down_ratio = down_ratio
+        self.act = activation
+        self.upsample = UpSample1d(up_ratio, up_kernel_size)
+        self.downsample = DownSample1d(down_ratio, down_kernel_size)
+
+    # x: [B,C,T]
+    def forward(self, x):
+        x = self.upsample(x)
+        x = self.act(x)
+        x = self.downsample(x)
+
+        return x