File size: 2,674 Bytes
1b140f4 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 | """Render a FLAME mesh overlay with a separately obtained FLAME model's topology."""
import argparse
import pickle
from pathlib import Path
import numpy as np
from PIL import Image, ImageDraw
from load_sample import load_sample
def visualize(sample, faces, output):
camera = sample["camera"]
vertices = sample["vertices_world"] @ np.asarray(camera["R"]).T + np.asarray(camera["t"])
K = np.asarray(camera["K"])
projection = vertices @ K.T
assert (vertices[:, 2] > 0).all()
uv = projection[:, :2] / projection[:, 2:]
image = Image.fromarray(sample["image"])
width, height = image.size
assert [width, height] == [camera["width"], camera["height"]]
# Painter's algorithm, back to front; opaque triangles hide rear surfaces.
mesh = Image.new("RGB", image.size, (0, 0, 0))
mask = Image.new("L", image.size, 0)
draw, draw_mask = ImageDraw.Draw(mesh), ImageDraw.Draw(mask)
triangles = vertices[faces]
normals = np.cross(triangles[:, 1] - triangles[:, 0], triangles[:, 2] - triangles[:, 0])
normals /= np.maximum(np.linalg.norm(normals, axis=1, keepdims=True), 1e-12)
colors = ((normals * 0.5 + 0.5) * 255).astype(np.uint8)
for index in np.argsort(triangles[..., 2].mean(axis=1))[::-1]:
points = [tuple(p) for p in uv[faces[index]].tolist()]
draw.polygon(points, fill=tuple(colors[index].tolist()))
draw_mask.polygon(points, fill=255)
overlay = Image.composite(Image.blend(image, mesh, 0.55), image, mask)
x0, y0, x1, y1 = sample["metadata"]["bbox_xyxy"]
ImageDraw.Draw(overlay).rectangle((x0, y0, x1, y1), outline=(255, 180, 0), width=2)
segmentation = Image.fromarray(np.repeat((sample["foreground"] * 255).astype(np.uint8)[..., None], 3, axis=2))
result = Image.new("RGB", (width * 3, height))
for index, panel in enumerate((image, overlay, segmentation)):
result.paste(panel, (index * width, 0))
result.thumbnail((1500, 900))
result.save(output)
if __name__ == "__main__":
parser = argparse.ArgumentParser(description=__doc__)
parser.add_argument("--root", type=Path, required=True)
parser.add_argument("--subject", required=True)
parser.add_argument("--frame", required=True)
parser.add_argument("--camera", required=True)
parser.add_argument("--flame-model", type=Path, required=True)
parser.add_argument("--output", type=Path, required=True)
args = parser.parse_args()
with args.flame_model.open("rb") as stream:
faces = np.asarray(pickle.load(stream, encoding="latin1")["f"], dtype=np.int32)
visualize(load_sample(args.root, args.subject, args.frame, args.camera), faces, args.output)
|