| 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" |
| } |