Refactor and improve the sag node.

Moved all the sag related code to comfy_extras/nodes_sag.py

comfy/model_patcher.py

@@ -61,6 +61,9 @@ class ModelPatcher:
         else:
             self.model_options["sampler_cfg_function"] = sampler_cfg_function
 
+    def set_model_sampler_post_cfg_function(self, post_cfg_function):
+        self.model_options["sampler_post_cfg_function"] = self.model_options.get("sampler_post_cfg_function", []) + [post_cfg_function]
+
     def set_model_unet_function_wrapper(self, unet_wrapper_function):
         self.model_options["model_function_wrapper"] = unet_wrapper_function
 
@@ -70,13 +73,17 @@ class ModelPatcher:
             to["patches"] = {}
         to["patches"][name] = to["patches"].get(name, []) + [patch]
 
-    def set_model_patch_replace(self, patch, name, block_name, number):
+    def set_model_patch_replace(self, patch, name, block_name, number, transformer_index=None):
         to = self.model_options["transformer_options"]
         if "patches_replace" not in to:
             to["patches_replace"] = {}
         if name not in to["patches_replace"]:
             to["patches_replace"][name] = {}
-        to["patches_replace"][name][(block_name, number)] = patch
+        if transformer_index is not None:
+            block = (block_name, number, transformer_index)
+        else:
+            block = (block_name, number)
+        to["patches_replace"][name][block] = patch
 
     def set_model_attn1_patch(self, patch):
         self.set_model_patch(patch, "attn1_patch")
@@ -84,11 +91,11 @@ class ModelPatcher:
     def set_model_attn2_patch(self, patch):
         self.set_model_patch(patch, "attn2_patch")
 
-    def set_model_attn1_replace(self, patch, block_name, number):
+    def set_model_attn1_replace(self, patch, block_name, number, transformer_index=None):
-        self.set_model_patch_replace(patch, "attn1", block_name, number)
+        self.set_model_patch_replace(patch, "attn1", block_name, number, transformer_index)
 
-    def set_model_attn2_replace(self, patch, block_name, number):
+    def set_model_attn2_replace(self, patch, block_name, number, transformer_index=None):
-        self.set_model_patch_replace(patch, "attn2", block_name, number)
+        self.set_model_patch_replace(patch, "attn2", block_name, number, transformer_index)
 
     def set_model_attn1_output_patch(self, patch):
         self.set_model_patch(patch, "attn1_output_patch")

comfy/samplers.py

@@ -1,7 +1,6 @@
 from .k_diffusion import sampling as k_diffusion_sampling
 from .extra_samplers import uni_pc
 import torch
-import torch.nn.functional as F
 import enum
 from comfy import model_management
 import math
@@ -9,10 +8,6 @@ from comfy import model_base
 import comfy.utils
 import comfy.conds
 
-
-#The main sampling function shared by all the samplers
-#Returns denoised
-def sampling_function(model, x, timestep, uncond, cond, cond_scale, model_options={}, seed=None):
 def get_area_and_mult(conds, x_in, timestep_in):
     area = (x_in.shape[2], x_in.shape[3], 0, 0)
     strength = 1.0
@@ -246,73 +241,27 @@ def sampling_function(model, x, timestep, uncond, cond, cond_scale, model_option
     del out_uncond_count
     return out_cond, out_uncond
 
+#The main sampling function shared by all the samplers
+#Returns denoised
+def sampling_function(model, x, timestep, uncond, cond, cond_scale, model_options={}, seed=None):
+        if math.isclose(cond_scale, 1.0):
+            uncond_ = None
+        else:
+            uncond_ = uncond
 
-        # if we're doing SAG, we still need to do uncond guidance, even though the cond and uncond will cancel out.
+        cond_pred, uncond_pred = calc_cond_uncond_batch(model, cond, uncond_, x, timestep, model_options)
-        if math.isclose(cond_scale, 1.0) and "sag" not in model_options:
-            uncond = None
-
-        cond_pred, uncond_pred = calc_cond_uncond_batch(model, cond, uncond, x, timestep, model_options)
         cfg_result = uncond_pred + (cond_pred - uncond_pred) * cond_scale
         if "sampler_cfg_function" in model_options:
             args = {"cond": x - cond_pred, "uncond": x - uncond_pred, "cond_scale": cond_scale, "timestep": timestep, "input": x, "sigma": timestep}
             cfg_result = x - model_options["sampler_cfg_function"](args)
 
-        if "sag" in model_options:
+        for fn in model_options.get("sampler_post_cfg_function", []):
-            assert uncond is not None, "SAG requires uncond guidance"
+            args = {"denoised": cfg_result, "cond": cond, "uncond": uncond, "model": model, "uncond_denoised": uncond_pred, "cond_denoised": cond_pred,
-            sag_scale = model_options["sag_scale"]
+                    "sigma": timestep, "model_options": model_options, "input": x}
-            sag_sigma = model_options["sag_sigma"]
+            cfg_result = fn(args)
-            sag_threshold = model_options.get("sag_threshold", 1.0)
 
-            # these methods are added by the sag patcher
-            uncond_attn = model.get_attn_scores()
-            mid_shape = model.get_mid_block_shape()
-            # create the adversarially blurred image
-            degraded = create_blur_map(uncond_pred, uncond_attn, mid_shape, sag_sigma, sag_threshold)
-            degraded_noised = degraded + x - uncond_pred
-            # call into the UNet
-            (sag, _) = calc_cond_uncond_batch(model, uncond, None, degraded_noised, timestep, model_options)
-            cfg_result += (degraded - sag) * sag_scale
         return cfg_result
 
-def create_blur_map(x0, attn, mid_shape, sigma=3.0, threshold=1.0):
-    # reshape and GAP the attention map
-    _, hw1, hw2 = attn.shape
-    b, _, lh, lw = x0.shape
-    attn = attn.reshape(b, -1, hw1, hw2)
-    # Global Average Pool
-    mask = attn.mean(1, keepdim=False).sum(1, keepdim=False) > threshold
-    # Reshape
-    mask = (
-        mask.reshape(b, *mid_shape)
-        .unsqueeze(1)
-        .type(attn.dtype)
-    )
-    # Upsample
-    mask = F.interpolate(mask, (lh, lw))
-
-    blurred = gaussian_blur_2d(x0, kernel_size=9, sigma=sigma)
-    blurred = blurred * mask + x0 * (1 - mask)
-    return blurred
-
-def gaussian_blur_2d(img, kernel_size, sigma):
-    ksize_half = (kernel_size - 1) * 0.5
-
-    x = torch.linspace(-ksize_half, ksize_half, steps=kernel_size)
-
-    pdf = torch.exp(-0.5 * (x / sigma).pow(2))
-
-    x_kernel = pdf / pdf.sum()
-    x_kernel = x_kernel.to(device=img.device, dtype=img.dtype)
-
-    kernel2d = torch.mm(x_kernel[:, None], x_kernel[None, :])
-    kernel2d = kernel2d.expand(img.shape[-3], 1, kernel2d.shape[0], kernel2d.shape[1])
-
-    padding = [kernel_size // 2, kernel_size // 2, kernel_size // 2, kernel_size // 2]
-
-    img = F.pad(img, padding, mode="reflect")
-    img = F.conv2d(img, kernel2d, groups=img.shape[-3])
-    return img
-
 class CFGNoisePredictor(torch.nn.Module):
     def __init__(self, model):
         super().__init__()

comfy_extras/nodes_sag.py

@@ -1,8 +1,12 @@
 import torch
 from torch import einsum
+import torch.nn.functional as F
+import math
+
 from einops import rearrange, repeat
 import os
 from comfy.ldm.modules.attention import optimized_attention, _ATTN_PRECISION
+import comfy.samplers
 
 # from comfy/ldm/modules/attention.py
 # but modified to return attention scores as well as output
@@ -49,7 +53,49 @@ def attention_basic_with_sim(q, k, v, heads, mask=None):
     )
     return (out, sim)
 
-class SagNode:
+def create_blur_map(x0, attn, sigma=3.0, threshold=1.0):
+    # reshape and GAP the attention map
+    _, hw1, hw2 = attn.shape
+    b, _, lh, lw = x0.shape
+    attn = attn.reshape(b, -1, hw1, hw2)
+    # Global Average Pool
+    mask = attn.mean(1, keepdim=False).sum(1, keepdim=False) > threshold
+    ratio = round(math.sqrt(lh * lw / hw1))
+    mid_shape = [math.ceil(lh / ratio), math.ceil(lw / ratio)]
+
+    # Reshape
+    mask = (
+        mask.reshape(b, *mid_shape)
+        .unsqueeze(1)
+        .type(attn.dtype)
+    )
+    # Upsample
+    mask = F.interpolate(mask, (lh, lw))
+
+    blurred = gaussian_blur_2d(x0, kernel_size=9, sigma=sigma)
+    blurred = blurred * mask + x0 * (1 - mask)
+    return blurred
+
+def gaussian_blur_2d(img, kernel_size, sigma):
+    ksize_half = (kernel_size - 1) * 0.5
+
+    x = torch.linspace(-ksize_half, ksize_half, steps=kernel_size)
+
+    pdf = torch.exp(-0.5 * (x / sigma).pow(2))
+
+    x_kernel = pdf / pdf.sum()
+    x_kernel = x_kernel.to(device=img.device, dtype=img.dtype)
+
+    kernel2d = torch.mm(x_kernel[:, None], x_kernel[None, :])
+    kernel2d = kernel2d.expand(img.shape[-3], 1, kernel2d.shape[0], kernel2d.shape[1])
+
+    padding = [kernel_size // 2, kernel_size // 2, kernel_size // 2, kernel_size // 2]
+
+    img = F.pad(img, padding, mode="reflect")
+    img = F.conv2d(img, kernel2d, groups=img.shape[-3])
+    return img
+
+class SelfAttentionGuidance:
     @classmethod
     def INPUT_TYPES(s):
         return {"required": { "model": ("MODEL",),
@@ -63,15 +109,9 @@ class SagNode:
 
     def patch(self, model, scale, blur_sigma):
         m = model.clone()
-        # set extra options on the model
-        m.model_options["sag"] = True
-        m.model_options["sag_scale"] = scale
-        m.model_options["sag_sigma"] = blur_sigma
 
         attn_scores = None
         mid_block_shape = None
-        m.model.get_attn_scores = lambda: attn_scores
-        m.model.get_mid_block_shape = lambda: mid_block_shape
 
         # TODO: make this work properly with chunked batches
         #       currently, we can only save the attn from one UNet call
@@ -92,24 +132,41 @@ class SagNode:
             else:
                 return optimized_attention(q, k, v, heads=heads)
 
+        def post_cfg_function(args):
+            nonlocal attn_scores
+            nonlocal mid_block_shape
+            uncond_attn = attn_scores
+
+            sag_scale = scale
+            sag_sigma = blur_sigma
+            sag_threshold = 1.0
+            model = args["model"]
+            uncond_pred = args["uncond_denoised"]
+            uncond = args["uncond"]
+            cfg_result = args["denoised"]
+            sigma = args["sigma"]
+            model_options = args["model_options"]
+            x = args["input"]
+
+            # create the adversarially blurred image
+            degraded = create_blur_map(uncond_pred, uncond_attn, sag_sigma, sag_threshold)
+            degraded_noised = degraded + x - uncond_pred
+            # call into the UNet
+            (sag, _) = comfy.samplers.calc_cond_uncond_batch(model, uncond, None, degraded_noised, sigma, model_options)
+            return cfg_result + (degraded - sag) * sag_scale
+
+        m.set_model_sampler_post_cfg_function(post_cfg_function)
+
         # from diffusers:
         # unet.mid_block.attentions[0].transformer_blocks[0].attn1.patch
-        def set_model_patch_replace(patch, name, key):
+        m.set_model_attn1_replace(attn_and_record, "middle", 0, 0)
-            to = m.model_options["transformer_options"]
+
-            if "patches_replace" not in to:
-                to["patches_replace"] = {}
-            if name not in to["patches_replace"]:
-                to["patches_replace"][name] = {}
-            to["patches_replace"][name][key] = patch
-        set_model_patch_replace(attn_and_record, "attn1", ("middle", 0, 0))
-        # from diffusers:
-        # unet.mid_block.attentions[0].register_forward_hook()
-        def forward_hook(m, inp, out):
-            nonlocal mid_block_shape
-            mid_block_shape = out[0].shape[-2:]
-        m.model.diffusion_model.middle_block[0].register_forward_hook(forward_hook)
         return (m, )
 
 NODE_CLASS_MAPPINGS = {
-    "Self-Attention Guidance": SagNode,
+    "SelfAttentionGuidance": SelfAttentionGuidance,
+}
+
+NODE_DISPLAY_NAME_MAPPINGS = {
+    "SelfAttentionGuidance": "Self-Attention Guidance",
 }