Implement noise augmentation for SD 4X upscale model.

comfy/ldm/modules/diffusionmodules/openaimodel.py

@@ -498,7 +498,7 @@ class UNetModel(nn.Module):
 
         if self.num_classes is not None:
             if isinstance(self.num_classes, int):
-                self.label_emb = nn.Embedding(num_classes, time_embed_dim)
+                self.label_emb = nn.Embedding(num_classes, time_embed_dim, dtype=self.dtype, device=device)
             elif self.num_classes == "continuous":
                 print("setting up linear c_adm embedding layer")
                 self.label_emb = nn.Linear(1, time_embed_dim)

comfy/ldm/modules/diffusionmodules/upscaling.py

@@ -41,8 +41,12 @@ class AbstractLowScaleModel(nn.Module):
         self.register_buffer('sqrt_recip_alphas_cumprod', to_torch(np.sqrt(1. / alphas_cumprod)))
         self.register_buffer('sqrt_recipm1_alphas_cumprod', to_torch(np.sqrt(1. / alphas_cumprod - 1)))
 
-    def q_sample(self, x_start, t, noise=None):
+    def q_sample(self, x_start, t, noise=None, seed=None):
-        noise = default(noise, lambda: torch.randn_like(x_start))
+        if noise is None:
+            if seed is None:
+                noise = torch.randn_like(x_start)
+            else:
+                noise = torch.randn(x_start.size(), dtype=x_start.dtype, layout=x_start.layout, generator=torch.manual_seed(seed)).to(x_start.device)
         return (extract_into_tensor(self.sqrt_alphas_cumprod.to(x_start.device), t, x_start.shape) * x_start +
                 extract_into_tensor(self.sqrt_one_minus_alphas_cumprod.to(x_start.device), t, x_start.shape) * noise)
 
@@ -69,12 +73,12 @@ class ImageConcatWithNoiseAugmentation(AbstractLowScaleModel):
         super().__init__(noise_schedule_config=noise_schedule_config)
         self.max_noise_level = max_noise_level
 
-    def forward(self, x, noise_level=None):
+    def forward(self, x, noise_level=None, seed=None):
         if noise_level is None:
             noise_level = torch.randint(0, self.max_noise_level, (x.shape[0],), device=x.device).long()
         else:
             assert isinstance(noise_level, torch.Tensor)
-        z = self.q_sample(x, noise_level)
+        z = self.q_sample(x, noise_level, seed=seed)
         return z, noise_level
 
 

comfy/model_base.py

@@ -1,7 +1,7 @@
 import torch
-from comfy.ldm.modules.diffusionmodules.openaimodel import UNetModel
+from comfy.ldm.modules.diffusionmodules.openaimodel import UNetModel, Timestep
 from comfy.ldm.modules.encoders.noise_aug_modules import CLIPEmbeddingNoiseAugmentation
-from comfy.ldm.modules.diffusionmodules.openaimodel import Timestep
+from comfy.ldm.modules.diffusionmodules.upscaling import ImageConcatWithNoiseAugmentation
 import comfy.model_management
 import comfy.conds
 import comfy.ops
@@ -78,8 +78,9 @@ class BaseModel(torch.nn.Module):
         extra_conds = {}
         for o in kwargs:
             extra = kwargs[o]
-            if hasattr(extra, "to"):
+            if hasattr(extra, "dtype"):
-                extra = extra.to(dtype)
+                if extra.dtype != torch.int and extra.dtype != torch.long:
+                    extra = extra.to(dtype)
             extra_conds[o] = extra
 
         model_output = self.diffusion_model(xc, t, context=context, control=control, transformer_options=transformer_options, **extra_conds).float()
@@ -368,20 +369,31 @@ class Stable_Zero123(BaseModel):
 class SD_X4Upscaler(BaseModel):
     def __init__(self, model_config, model_type=ModelType.V_PREDICTION, device=None):
         super().__init__(model_config, model_type, device=device)
+        self.noise_augmentor = ImageConcatWithNoiseAugmentation(noise_schedule_config={"linear_start": 0.0001, "linear_end": 0.02}, max_noise_level=350)
 
     def extra_conds(self, **kwargs):
         out = {}
 
         image = kwargs.get("concat_image", None)
         noise = kwargs.get("noise", None)
+        noise_augment = kwargs.get("noise_augmentation", 0.0)
+        device = kwargs["device"]
+        seed = kwargs["seed"] - 10
+
+        noise_level = round((self.noise_augmentor.max_noise_level) * noise_augment)
 
         if image is None:
             image = torch.zeros_like(noise)[:,:3]
 
         if image.shape[1:] != noise.shape[1:]:
-            image = utils.common_upscale(image, noise.shape[-1], noise.shape[-2], "bilinear", "center")
+            image = utils.common_upscale(image.to(device), noise.shape[-1], noise.shape[-2], "bilinear", "center")
+
+        noise_level = torch.tensor([noise_level], device=device)
+        if noise_augment > 0:
+            image, noise_level = self.noise_augmentor(image.to(device), noise_level=noise_level, seed=seed)
 
         image = utils.resize_to_batch_size(image, noise.shape[0])
 
         out['c_concat'] = comfy.conds.CONDNoiseShape(image)
+        out['y'] = comfy.conds.CONDRegular(noise_level)
         return out

comfy/samplers.py

@@ -603,8 +603,8 @@ def sample(model, noise, positive, negative, cfg, device, sampler, sigmas, model
         latent_image = model.process_latent_in(latent_image)
 
     if hasattr(model, 'extra_conds'):
-        positive = encode_model_conds(model.extra_conds, positive, noise, device, "positive", latent_image=latent_image, denoise_mask=denoise_mask)
+        positive = encode_model_conds(model.extra_conds, positive, noise, device, "positive", latent_image=latent_image, denoise_mask=denoise_mask, seed=seed)
-        negative = encode_model_conds(model.extra_conds, negative, noise, device, "negative", latent_image=latent_image, denoise_mask=denoise_mask)
+        negative = encode_model_conds(model.extra_conds, negative, noise, device, "negative", latent_image=latent_image, denoise_mask=denoise_mask, seed=seed)
 
     #make sure each cond area has an opposite one with the same area
     for c in positive:

comfy/supported_models.py

@@ -290,6 +290,7 @@ class SD_X4Upscaler(SD20):
 
     unet_extra_config = {
         "disable_self_attentions": [True, True, True, False],
+        "num_classes": 1000,
         "num_heads": 8,
         "num_head_channels": -1,
     }

comfy_extras/nodes_sdupscale.py

@@ -9,7 +9,7 @@ class SD_4XUpscale_Conditioning:
                               "positive": ("CONDITIONING",),
                               "negative": ("CONDITIONING",),
                               "scale_ratio": ("FLOAT", {"default": 4.0, "min": 0.0, "max": 10.0, "step": 0.01}),
-                              # "noise_augmentation": ("FLOAT", {"default": 0.0, "min": 0.0, "max": 10.0, "step": 0.01}), #TODO
+                              "noise_augmentation": ("FLOAT", {"default": 0.0, "min": 0.0, "max": 1.0, "step": 0.001}),
                              }}
     RETURN_TYPES = ("CONDITIONING", "CONDITIONING", "LATENT")
     RETURN_NAMES = ("positive", "negative", "latent")
@@ -18,7 +18,7 @@ class SD_4XUpscale_Conditioning:
 
     CATEGORY = "conditioning/upscale_diffusion"
 
-    def encode(self, images, positive, negative, scale_ratio):
+    def encode(self, images, positive, negative, scale_ratio, noise_augmentation):
         width = max(1, round(images.shape[-2] * scale_ratio))
         height = max(1, round(images.shape[-3] * scale_ratio))
 
@@ -30,11 +30,13 @@ class SD_4XUpscale_Conditioning:
         for t in positive:
             n = [t[0], t[1].copy()]
             n[1]['concat_image'] = pixels
+            n[1]['noise_augmentation'] = noise_augmentation
             out_cp.append(n)
 
         for t in negative:
             n = [t[0], t[1].copy()]
             n[1]['concat_image'] = pixels
+            n[1]['noise_augmentation'] = noise_augmentation
             out_cn.append(n)
 
         latent = torch.zeros([images.shape[0], 4, height // 4, width // 4])