A question about the DA3Metric-Large model

When I adopt this model, the output intrinsics are empty, which results in the following error:
"AssertionError: Export to GLB: prediction.intrinsics is required but not available"

the code is as follows:
from depth_anything_3.api import DepthAnything3
import torch, glob, os
model_path = "model/DA3METRIC-LARGE"
model = DepthAnything3.from_pretrained(model_path).cuda()
image_path = "/home/images_test"
images = sorted(glob.glob(os.path.join(image_path, "images/*.png")))
pred = model.inference(images,
export_dir = image_path + "/output",
export_format="glb"
)