mirror of
https://github.com/comfyanonymous/ComfyUI.git
synced 2026-02-03 02:00:29 +08:00
ae20354b69
Mark widgets as advanced across core, comfy_extras, and comfy_api_nodes to support the new collapsible advanced inputs section in the frontend. Changes: - 267 advanced markers in comfy_extras/ - 162 advanced markers in comfy_api_nodes/ - All files pass python3 -m py_compile verification Widgets marked advanced (hidden by default): - Scheduler internals: sigma_max, sigma_min, rho, mu, beta, alpha - Sampler internals: eta, s_noise, order, rtol, atol, h_init, pcoeff, etc. - Memory optimization: tile_size, overlap, temporal_size, temporal_overlap - Pipeline controls: add_noise, start_at_step, end_at_step - Timing controls: start_percent, end_percent - Layer selection: stop_at_clip_layer, layers, block_number - Video encoding: codec, crf, format - Device/dtype: device, noise_device, dtype, weight_dtype Widgets kept basic (always visible): - Core params: strength, steps, cfg, denoise, seed, width, height - Model selectors: ckpt_name, lora_name, vae_name, sampler_name - Common controls: upscale_method, crop, batch_size, fps, opacity Related: frontend PR #11939 Amp-Thread-ID: https://ampcode.com/threads/T-019c1734-6b61-702e-b333-f02c399963fc
152 lines
6.4 KiB
Python
152 lines
6.4 KiB
Python
from typing_extensions import override
|
|
|
|
from comfy_api.latest import ComfyExtension, io
|
|
|
|
|
|
def attention_multiply(attn, model, q, k, v, out):
|
|
m = model.clone()
|
|
sd = model.model_state_dict()
|
|
|
|
for key in sd:
|
|
if key.endswith("{}.to_q.bias".format(attn)) or key.endswith("{}.to_q.weight".format(attn)):
|
|
m.add_patches({key: (None,)}, 0.0, q)
|
|
if key.endswith("{}.to_k.bias".format(attn)) or key.endswith("{}.to_k.weight".format(attn)):
|
|
m.add_patches({key: (None,)}, 0.0, k)
|
|
if key.endswith("{}.to_v.bias".format(attn)) or key.endswith("{}.to_v.weight".format(attn)):
|
|
m.add_patches({key: (None,)}, 0.0, v)
|
|
if key.endswith("{}.to_out.0.bias".format(attn)) or key.endswith("{}.to_out.0.weight".format(attn)):
|
|
m.add_patches({key: (None,)}, 0.0, out)
|
|
|
|
return m
|
|
|
|
|
|
class UNetSelfAttentionMultiply(io.ComfyNode):
|
|
@classmethod
|
|
def define_schema(cls) -> io.Schema:
|
|
return io.Schema(
|
|
node_id="UNetSelfAttentionMultiply",
|
|
category="_for_testing/attention_experiments",
|
|
inputs=[
|
|
io.Model.Input("model"),
|
|
io.Float.Input("q", default=1.0, min=0.0, max=10.0, step=0.01, advanced=True),
|
|
io.Float.Input("k", default=1.0, min=0.0, max=10.0, step=0.01, advanced=True),
|
|
io.Float.Input("v", default=1.0, min=0.0, max=10.0, step=0.01, advanced=True),
|
|
io.Float.Input("out", default=1.0, min=0.0, max=10.0, step=0.01, advanced=True),
|
|
],
|
|
outputs=[io.Model.Output()],
|
|
is_experimental=True,
|
|
)
|
|
|
|
@classmethod
|
|
def execute(cls, model, q, k, v, out) -> io.NodeOutput:
|
|
m = attention_multiply("attn1", model, q, k, v, out)
|
|
return io.NodeOutput(m)
|
|
|
|
|
|
class UNetCrossAttentionMultiply(io.ComfyNode):
|
|
@classmethod
|
|
def define_schema(cls) -> io.Schema:
|
|
return io.Schema(
|
|
node_id="UNetCrossAttentionMultiply",
|
|
category="_for_testing/attention_experiments",
|
|
inputs=[
|
|
io.Model.Input("model"),
|
|
io.Float.Input("q", default=1.0, min=0.0, max=10.0, step=0.01, advanced=True),
|
|
io.Float.Input("k", default=1.0, min=0.0, max=10.0, step=0.01, advanced=True),
|
|
io.Float.Input("v", default=1.0, min=0.0, max=10.0, step=0.01, advanced=True),
|
|
io.Float.Input("out", default=1.0, min=0.0, max=10.0, step=0.01, advanced=True),
|
|
],
|
|
outputs=[io.Model.Output()],
|
|
is_experimental=True,
|
|
)
|
|
|
|
@classmethod
|
|
def execute(cls, model, q, k, v, out) -> io.NodeOutput:
|
|
m = attention_multiply("attn2", model, q, k, v, out)
|
|
return io.NodeOutput(m)
|
|
|
|
|
|
class CLIPAttentionMultiply(io.ComfyNode):
|
|
@classmethod
|
|
def define_schema(cls) -> io.Schema:
|
|
return io.Schema(
|
|
node_id="CLIPAttentionMultiply",
|
|
search_aliases=["clip attention scale", "text encoder attention"],
|
|
category="_for_testing/attention_experiments",
|
|
inputs=[
|
|
io.Clip.Input("clip"),
|
|
io.Float.Input("q", default=1.0, min=0.0, max=10.0, step=0.01, advanced=True),
|
|
io.Float.Input("k", default=1.0, min=0.0, max=10.0, step=0.01, advanced=True),
|
|
io.Float.Input("v", default=1.0, min=0.0, max=10.0, step=0.01, advanced=True),
|
|
io.Float.Input("out", default=1.0, min=0.0, max=10.0, step=0.01, advanced=True),
|
|
],
|
|
outputs=[io.Clip.Output()],
|
|
is_experimental=True,
|
|
)
|
|
|
|
@classmethod
|
|
def execute(cls, clip, q, k, v, out) -> io.NodeOutput:
|
|
m = clip.clone()
|
|
sd = m.patcher.model_state_dict()
|
|
|
|
for key in sd:
|
|
if key.endswith("self_attn.q_proj.weight") or key.endswith("self_attn.q_proj.bias"):
|
|
m.add_patches({key: (None,)}, 0.0, q)
|
|
if key.endswith("self_attn.k_proj.weight") or key.endswith("self_attn.k_proj.bias"):
|
|
m.add_patches({key: (None,)}, 0.0, k)
|
|
if key.endswith("self_attn.v_proj.weight") or key.endswith("self_attn.v_proj.bias"):
|
|
m.add_patches({key: (None,)}, 0.0, v)
|
|
if key.endswith("self_attn.out_proj.weight") or key.endswith("self_attn.out_proj.bias"):
|
|
m.add_patches({key: (None,)}, 0.0, out)
|
|
return io.NodeOutput(m)
|
|
|
|
|
|
class UNetTemporalAttentionMultiply(io.ComfyNode):
|
|
@classmethod
|
|
def define_schema(cls) -> io.Schema:
|
|
return io.Schema(
|
|
node_id="UNetTemporalAttentionMultiply",
|
|
category="_for_testing/attention_experiments",
|
|
inputs=[
|
|
io.Model.Input("model"),
|
|
io.Float.Input("self_structural", default=1.0, min=0.0, max=10.0, step=0.01, advanced=True),
|
|
io.Float.Input("self_temporal", default=1.0, min=0.0, max=10.0, step=0.01, advanced=True),
|
|
io.Float.Input("cross_structural", default=1.0, min=0.0, max=10.0, step=0.01, advanced=True),
|
|
io.Float.Input("cross_temporal", default=1.0, min=0.0, max=10.0, step=0.01, advanced=True),
|
|
],
|
|
outputs=[io.Model.Output()],
|
|
is_experimental=True,
|
|
)
|
|
|
|
@classmethod
|
|
def execute(cls, model, self_structural, self_temporal, cross_structural, cross_temporal) -> io.NodeOutput:
|
|
m = model.clone()
|
|
sd = model.model_state_dict()
|
|
|
|
for k in sd:
|
|
if (k.endswith("attn1.to_out.0.bias") or k.endswith("attn1.to_out.0.weight")):
|
|
if '.time_stack.' in k:
|
|
m.add_patches({k: (None,)}, 0.0, self_temporal)
|
|
else:
|
|
m.add_patches({k: (None,)}, 0.0, self_structural)
|
|
elif (k.endswith("attn2.to_out.0.bias") or k.endswith("attn2.to_out.0.weight")):
|
|
if '.time_stack.' in k:
|
|
m.add_patches({k: (None,)}, 0.0, cross_temporal)
|
|
else:
|
|
m.add_patches({k: (None,)}, 0.0, cross_structural)
|
|
return io.NodeOutput(m)
|
|
|
|
|
|
class AttentionMultiplyExtension(ComfyExtension):
|
|
@override
|
|
async def get_node_list(self) -> list[type[io.ComfyNode]]:
|
|
return [
|
|
UNetSelfAttentionMultiply,
|
|
UNetCrossAttentionMultiply,
|
|
CLIPAttentionMultiply,
|
|
UNetTemporalAttentionMultiply,
|
|
]
|
|
|
|
async def comfy_entrypoint() -> AttentionMultiplyExtension:
|
|
return AttentionMultiplyExtension()
|