face-swap-docker / plugins /plugin_codeformer.py
pknez's picture
Upload 913 files
0c87db7
raw
history blame
No virus
2.92 kB
# Codeformer enchance plugin
# author: Vladislav Janvarev
# CountFloyd 20230717, extended to blend original/destination images
from chain_img_processor import ChainImgProcessor, ChainImgPlugin
import os
from PIL import Image
from numpy import asarray
modname = os.path.basename(__file__)[:-3] # calculating modname
# start function
def start(core:ChainImgProcessor):
manifest = { # plugin settings
"name": "Codeformer", # name
"version": "3.0", # version
"default_options": {
"background_enhance": True, #
"face_upsample": True, #
"upscale": 2, #
"codeformer_fidelity": 0.8,
"skip_if_no_face":False,
},
"img_processor": {
"codeformer": PluginCodeformer # 1 function - init, 2 - process
}
}
return manifest
def start_with_options(core:ChainImgProcessor, manifest:dict):
pass
class PluginCodeformer(ChainImgPlugin):
def init_plugin(self):
import plugins.codeformer_app_cv2
pass
def process(self, img, params:dict):
import copy
# params can be used to transfer some img info to next processors
from plugins.codeformer_app_cv2 import inference_app
options = self.core.plugin_options(modname)
if "face_detected" in params:
if not params["face_detected"]:
return img
# don't touch original
temp_frame = copy.copy(img)
if "processed_faces" in params:
for face in params["processed_faces"]:
start_x, start_y, end_x, end_y = map(int, face['bbox'])
padding_x = int((end_x - start_x) * 0.5)
padding_y = int((end_y - start_y) * 0.5)
start_x = max(0, start_x - padding_x)
start_y = max(0, start_y - padding_y)
end_x = max(0, end_x + padding_x)
end_y = max(0, end_y + padding_y)
temp_face = temp_frame[start_y:end_y, start_x:end_x]
if temp_face.size:
temp_face = inference_app(temp_face, options.get("background_enhance"), options.get("face_upsample"),
options.get("upscale"), options.get("codeformer_fidelity"),
options.get("skip_if_no_face"))
temp_frame[start_y:end_y, start_x:end_x] = temp_face
else:
temp_frame = inference_app(temp_frame, options.get("background_enhance"), options.get("face_upsample"),
options.get("upscale"), options.get("codeformer_fidelity"),
options.get("skip_if_no_face"))
if not "blend_ratio" in params:
return temp_frame
temp_frame = Image.blend(Image.fromarray(img), Image.fromarray(temp_frame), params["blend_ratio"])
return asarray(temp_frame)