|
from ..utils import common_annotator_call, create_node_input_types
|
|
import comfy.model_management as model_management
|
|
|
|
class DSINE_Normal_Map_Preprocessor:
|
|
@classmethod
|
|
def INPUT_TYPES(s):
|
|
return create_node_input_types(
|
|
fov=("FLOAT", {"min": 0.0, "max": 365.0, "step": 0.05, "default": 60.0}),
|
|
iterations=("INT", {"min": 1, "max": 20, "step": 1, "default": 5})
|
|
)
|
|
|
|
RETURN_TYPES = ("IMAGE",)
|
|
FUNCTION = "execute"
|
|
|
|
CATEGORY = "ControlNet Preprocessors/Normal and Depth Estimators"
|
|
|
|
def execute(self, image, fov, iterations, resolution=512, **kwargs):
|
|
from controlnet_aux.dsine import DsineDetector
|
|
|
|
model = DsineDetector.from_pretrained().to(model_management.get_torch_device())
|
|
out = common_annotator_call(model, image, fov=fov, iterations=iterations, resolution=resolution)
|
|
del model
|
|
return (out,)
|
|
|
|
NODE_CLASS_MAPPINGS = {
|
|
"DSINE-NormalMapPreprocessor": DSINE_Normal_Map_Preprocessor
|
|
}
|
|
NODE_DISPLAY_NAME_MAPPINGS = {
|
|
"DSINE-NormalMapPreprocessor": "DSINE Normal Map"
|
|
} |