selimaj-devandclaude
df9a2587e6
tools/capture drives a store page's skin viewer in headless Firefox, stepping its clock one tick at a time and taking every tick from ten fixed cameras. It then fits our own rig to the frames by rendering the model (a Python port of LimbBend and PoseApplier) and matching outlines and colours, and writes the result as Emotecraft JSON. It measures pixels only and never reads the page's animation data. The README covers the steps, the checks and the limits. Co-Authored-By: Claude Opus 5.5 <[email protected]>
64 lines
2.3 KiB
Python
64 lines
2.3 KiB
Python
"""Draws the fitted pose's outline over the captured frames, to see where the fit is off.
|
|
|
|
.venv/bin/python overlay.py <name> [--ticks 0:13:3] [--views y000,y090,y045,top000]
|
|
|
|
Writes frames/<name>/overlay.png: rows are views, columns ticks. Red is where only our model is, blue
|
|
where only theirs is, and grey where both are.
|
|
"""
|
|
import argparse
|
|
import json
|
|
from pathlib import Path
|
|
|
|
import numpy as np
|
|
from PIL import Image
|
|
|
|
from model import Model, rig_to_pose
|
|
from render import Camera, render_masks
|
|
|
|
|
|
def main():
|
|
parser = argparse.ArgumentParser()
|
|
parser.add_argument("name")
|
|
parser.add_argument("--ticks", default="0:13:3")
|
|
parser.add_argument("--views", default="y000,y090,y045,top000")
|
|
args = parser.parse_args()
|
|
|
|
base = Path("frames") / args.name
|
|
meta = json.loads((base / "views.json").read_text())
|
|
data = json.loads((base / "rig.json").read_text())
|
|
by_tick = {t["tick"]: np.array(t["rig"]) for t in data["ticks"]}
|
|
ticks = [t for t in range(*map(int, args.ticks.split(":"))) if t in by_tick]
|
|
views = args.views.split(",")
|
|
scale = 2
|
|
cams = {v: Camera(meta["views"][v], meta["size"], meta["aspect"], scale) for v in views}
|
|
model = Model(spacing=0.25)
|
|
|
|
tiles = []
|
|
for v in views:
|
|
row = []
|
|
for t in ticks:
|
|
pts = np.concatenate(list(model.pose(rig_to_pose(by_tick[t])).values()))
|
|
ours = render_masks(pts, data["calib"], {v: cams[v]})[v]
|
|
frame = Image.open(base / v / f"{t:04d}.png").reduce(scale)
|
|
theirs = np.asarray(frame)[:, :, 3] > 127
|
|
img = np.full(ours.shape + (3,), 30, np.uint8)
|
|
img[theirs & ours] = (150, 150, 150)
|
|
img[ours & ~theirs] = (230, 60, 60)
|
|
img[theirs & ~ours] = (60, 110, 240)
|
|
row.append(img)
|
|
tiles.append(row)
|
|
|
|
both = np.zeros(tiles[0][0].shape[:2], bool)
|
|
for row in tiles:
|
|
for img in row:
|
|
both |= img.sum(axis=2) > 90
|
|
ys, xs = np.nonzero(both)
|
|
y0, y1, x0, x1 = max(ys.min() - 4, 0), ys.max() + 4, max(xs.min() - 4, 0), xs.max() + 4
|
|
sheet = np.concatenate([np.concatenate([img[y0:y1, x0:x1] for img in row], axis=1) for row in tiles], axis=0)
|
|
Image.fromarray(sheet).save(base / "overlay.png")
|
|
print("wrote", base / "overlay.png", "ticks", ticks)
|
|
|
|
|
|
if __name__ == "__main__":
|
|
main()
|