soaska/my-sd
soaska/my-sd
1
0
Fork 0
Code Issues Pull Requests Actions Packages Activity

update examples

Browse Source
This commit is contained in:
lllyasviel 2024-01-31 23:29:56 -08:00
parent a203113f43
commit 412bc10a01
3 changed files with 21 additions and 3 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

11
README.md
View File

@ -352,6 +352,7 @@ The memory optimization in this example is fully automatic. You do not need to c
import cv2
import gradio as gr
import torch
from modules import scripts
from modules.shared_cmd_options import cmd_opts
@ -472,18 +473,26 @@ class ControlNetExampleForge(scripts.Script):
sigma_min = unet.model.model_sampling.sigma_min
advanced_sigma_weighting = lambda s: (s - sigma_min) / (sigma_max - sigma_min)
# You can even input a tensor to mask all control injections
# The mask will be automatically resized during inference in UNet.
# The size should be B 1 H W and the H and W are not important
# because they will be resized automatically
advanced_mask_weighting = torch.ones(size=(1, 1, 512, 512))
# But in this simple example we do not use them
positive_advanced_weighting = None
negative_advanced_weighting = None
advanced_frame_weighting = None
advanced_sigma_weighting = None
advanced_mask_weighting = None
unet = apply_controlnet_advanced(unet=unet, controlnet=self.model, image_bchw=control_image_bchw,
strength=0.6, start_percent=0.0, end_percent=0.8,
positive_advanced_weighting=positive_advanced_weighting,
negative_advanced_weighting=negative_advanced_weighting,
advanced_frame_weighting=advanced_frame_weighting,
advanced_sigma_weighting=advanced_sigma_weighting)
advanced_sigma_weighting=advanced_sigma_weighting,
advanced_mask_weighting=advanced_mask_weighting)
p.sd_model.forge_objects.unet = unet

11
extensions-builtin/sd_forge_controlnet_example/scripts/sd_forge_controlnet_example.py
View File

@ -2,6 +2,7 @@
import cv2
import gradio as gr
import torch
from modules import scripts
from modules.shared_cmd_options import cmd_opts
@ -122,18 +123,26 @@ class ControlNetExampleForge(scripts.Script):
sigma_min = unet.model.model_sampling.sigma_min
advanced_sigma_weighting = lambda s: (s - sigma_min) / (sigma_max - sigma_min)
# You can even input a tensor to mask all control injections
# The mask will be automatically resized during inference in UNet.
# The size should be B 1 H W and the H and W are not important
# because they will be resized automatically
advanced_mask_weighting = torch.ones(size=(1, 1, 512, 512))
# But in this simple example we do not use them
positive_advanced_weighting = None
negative_advanced_weighting = None
advanced_frame_weighting = None
advanced_sigma_weighting = None
advanced_mask_weighting = None
unet = apply_controlnet_advanced(unet=unet, controlnet=self.model, image_bchw=control_image_bchw,
strength=0.6, start_percent=0.0, end_percent=0.8,
positive_advanced_weighting=positive_advanced_weighting,
negative_advanced_weighting=negative_advanced_weighting,
advanced_frame_weighting=advanced_frame_weighting,
advanced_sigma_weighting=advanced_sigma_weighting)
advanced_sigma_weighting=advanced_sigma_weighting,
advanced_mask_weighting=advanced_mask_weighting)
p.sd_model.forge_objects.unet = unet

2
ldm_patched/modules/controlnet.py
View File

@ -87,7 +87,7 @@ def compute_controlnet_weighting(control, cnet):
final_weight = final_weight * sigma_weight * frame_weight
if isinstance(advanced_mask_weighting, torch.Tensor):
control_signal = control_signal * torch.nn.functional.interpolate(advanced_mask_weighting, size=(H, W), mode='bilinear')
control_signal = control_signal * torch.nn.functional.interpolate(advanced_mask_weighting.to(control_signal), size=(H, W), mode='bilinear')
control[k][i] = control_signal * final_weight[:, None, None, None]