Merge PR #387 from Kosinkadink/develop - new fuse_methods by Quasimondo

Merging fuse_methods by Quasimondo

animatediff/context.py

@@ -11,9 +11,15 @@ class ContextFuseMethod:
     FLAT = "flat"
     PYRAMID = "pyramid"
     RELATIVE = "relative"
+    RANDOM = "random"
+    GAUSS_SIGMA = "gauss-sigma"
+    GAUSS_SIGMA_INV = "gauss-sigma inverse"
+    DELAYED_REVERSE_SAWTOOTH = "delayed reverse sawtooth"
+    PYRAMID_SIGMA = "pyramid-sigma"
+    PYRAMID_SIGMA_INV = "pyramid-sigma inverse"
 
-    LIST = [PYRAMID, FLAT]
-    LIST_STATIC = [PYRAMID, RELATIVE, FLAT]
+    LIST = [PYRAMID, FLAT, DELAYED_REVERSE_SAWTOOTH, PYRAMID_SIGMA, PYRAMID_SIGMA_INV, GAUSS_SIGMA, GAUSS_SIGMA_INV, RANDOM]
+    LIST_STATIC = [PYRAMID, RELATIVE, FLAT, DELAYED_REVERSE_SAWTOOTH, PYRAMID_SIGMA, PYRAMID_SIGMA_INV, GAUSS_SIGMA, GAUSS_SIGMA_INV, RANDOM]
 
 
 class ContextType:
@@ -308,18 +314,17 @@ def get_context_windows(num_frames: int, opts: Union[ContextOptionsGroup, Contex
 }
 
 
-def get_context_weights(num_frames: int, fuse_method: str):
+def get_context_weights(num_frames: int, fuse_method: str, sigma: Tensor = None):
     weights_func = FUSE_MAPPING.get(fuse_method, None)
     if not weights_func:
         raise ValueError(f"Unknown fuse_method '{fuse_method}'.")
-    return weights_func(num_frames)
+    return weights_func(num_frames, sigma=sigma )
 
 
 def create_weights_flat(length: int, **kwargs) -> list[float]:
     # weight is the same for all
     return [1.0] * length
 
-
 def create_weights_pyramid(length: int, **kwargs) -> list[float]:
     # weight is based on the distance away from the edge of the context window;
     # based on weighted average concept in FreeNoise paper
@@ -331,11 +336,90 @@ def create_weights_pyramid(length: int, **kwargs) -> list[float]:
         weight_sequence = list(range(1, max_weight, 1)) + [max_weight] + list(range(max_weight - 1, 0, -1))
     return weight_sequence
 
+def create_weights_random(length: int, **kwargs) -> list[float]:
+    if length % 2 == 0:
+        max_weight = length // 2
+    else:
+        max_weight = (length + 1) // 2
+    return list(np.random.random(length)*max_weight+0.001)
+
+def create_weights_gauss_sigma(length: int, **kwargs) -> list[float]:
+    sigma = 1.0 + 8.0*(min(4.0, kwargs["sigma"].mean().cpu()) / 4.0)
+    ax = np.linspace(-(length - 1) / 2., (length - 1) / 2., length)
+    w = np.exp(-0.5 * np.square(ax) / np.square(sigma))
+    if length % 2 == 0:
+        max_weight = length // 2
+    else:
+        max_weight = (length + 1) // 2
+    w *= max_weight / np.linalg.norm(w)
+    #print("create_weights_gauss_sigma sigma",sigma,w)
+    return list(w)
+
+def create_weights_gauss_sigma_inv(length: int, **kwargs) -> list[float]:
+    sigma = 1.0 + 8.0*(1.0-min(4.0, kwargs["sigma"].mean().cpu()) / 4.0)
+    ax = np.linspace(-(length - 1) / 2., (length - 1) / 2., length)
+    w = np.exp(-0.5 * np.square(ax) / np.square(sigma))
+    if length % 2 == 0:
+        max_weight = length // 2
+    else:
+        max_weight = (length + 1) // 2
+    w *= max_weight / np.linalg.norm(w)
+    #print("create_weights_gauss_sigma_inv sigma",sigma,w)
+    return list(w)
+
+def create_weights_pyramid_sigma_inv(length: int, **kwargs) -> list[float]:
+    sigma = min(4.0, kwargs["sigma"].mean().cpu()) / 4.0
+
+    if length % 2 == 0:
+        max_weight = length // 2
+        weight_sequence = np.array(list(range(1, max_weight + 1, 1)) + list(range(max_weight, 0, -1)))
+        weight_sequence2 = np.array([-max_weight]*(max_weight-1) +[max_weight,max_weight] + [-max_weight]*(max_weight-1))
+    else:
+        max_weight = (length + 1) // 2
+        weight_sequence = list(range(1, max_weight, 1)) + [max_weight] + list(range(max_weight - 1, 0, -1))
+        weight_sequence2 = np.array([-max_weight]*(max_weight) +[max_weight] + [-max_weight]*(max_weight-1))
+    weight_sequence = (sigma * weight_sequence2 + (1.0-sigma) * weight_sequence).clip(0.001,max_weight)
+    #print("create_weights_pyramid_sigma_inv",kwargs["sigma"].mean(),sigma, len(weight_sequence),weight_sequence)
+    return list(weight_sequence)
+
+def create_weights_pyramid_sigma(length: int, **kwargs) -> list[float]:
+    sigma = min(4.0, kwargs["sigma"].mean().cpu()) / 4.0
+
+    if length % 2 == 0:
+        max_weight = length // 2
+        weight_sequence = np.array(list(range(1, max_weight + 1, 1)) + list(range(max_weight, 0, -1)))
+        weight_sequence2 = np.array([-max_weight]*(max_weight-1) +[max_weight,max_weight] + [-max_weight]*(max_weight-1))
+    else:
+        max_weight = (length + 1) // 2
+        weight_sequence = list(range(1, max_weight, 1)) + [max_weight] + list(range(max_weight - 1, 0, -1))
+        weight_sequence2 = np.array([-max_weight]*(max_weight) +[max_weight] + [-max_weight]*(max_weight-1))
+    weight_sequence = (sigma * weight_sequence + (1.0-sigma) * weight_sequence2).clip(0.001,max_weight)
+    #print("create_weights_pyramid_sigma",kwargs["sigma"].mean(),sigma, len(weight_sequence),weight_sequence)
+    return list(weight_sequence)
+
+def create_weights_delayed_reverse_sawtooth(length: int, **kwargs) -> list[float]:
+    # assigns 0.01 to first half (or half-1 if even) of weights, then the rest of the weights are basically
+    # based on distance from context edge
+    if length % 2 == 0:
+        max_weight = length // 2
+        weight_sequence = [0.01]*(max_weight-1) + [max_weight] + list(range(max_weight, 0, -1))
+    else:
+        max_weight = (length + 1) // 2
+        weight_sequence = [0.01]*max_weight + [max_weight] + list(range(max_weight - 1, 0, -1))
+    #print("create_weights_delayed_falling_edge",len(weight_sequence),weight_sequence)
+    return weight_sequence
+
 
 FUSE_MAPPING = {
     ContextFuseMethod.FLAT: create_weights_flat,
     ContextFuseMethod.PYRAMID: create_weights_pyramid,
     ContextFuseMethod.RELATIVE: create_weights_pyramid,
+    ContextFuseMethod.GAUSS_SIGMA: create_weights_gauss_sigma,
+    ContextFuseMethod.GAUSS_SIGMA_INV: create_weights_gauss_sigma_inv,
+    ContextFuseMethod.RANDOM: create_weights_random,
+    ContextFuseMethod.DELAYED_REVERSE_SAWTOOTH: create_weights_delayed_reverse_sawtooth,
+    ContextFuseMethod.PYRAMID_SIGMA: create_weights_pyramid_sigma,
+    ContextFuseMethod.PYRAMID_SIGMA_INV: create_weights_pyramid_sigma_inv,
 }
 
 

animatediff/sampling.py

@@ -609,7 +609,7 @@ def get_resized_cond(cond_in, full_idxs: list[int], context_length: int) -> list
                         biases_final[i][(full_length*n)+idx] = bias_total + bias
         else:
             # add conds and counts based on weights of fuse method
-            weights = get_context_weights(len(ctx_idxs), ADGS.params.context_options.fuse_method) * batched_conds
+            weights = get_context_weights(len(ctx_idxs), ADGS.params.context_options.fuse_method, sigma=timestep) * batched_conds
             weights_tensor = torch.Tensor(weights).to(device=x_in.device).unsqueeze(-1).unsqueeze(-1).unsqueeze(-1)
             for i in range(len(sub_conds_out)):
                 conds_final[i][full_idxs] += sub_conds_out[i] * weights_tensor

pyproject.toml

@@ -1,7 +1,7 @@
 [project]
 name = "comfyui-animatediff-evolved"
 description = "Improved AnimateDiff integration for ComfyUI."
-version = "1.0.1"
+version = "1.0.2"
 license = "LICENSE"
 dependencies = []