Spaces:
Running
Running
File size: 13,963 Bytes
5a5f252 1baefa4 5a5f252 1baefa4 5a5f252 1baefa4 5a5f252 1baefa4 5a5f252 1baefa4 5a5f252 1baefa4 5a5f252 1baefa4 5a5f252 1baefa4 5a5f252 1baefa4 5a5f252 1baefa4 5a5f252 1baefa4 5a5f252 1baefa4 5a5f252 1baefa4 5a5f252 1baefa4 5a5f252 1baefa4 5a5f252 1baefa4 58b1e11 | 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 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 | """
relief_workshop.py
Image → 3D relief tile — versione workshop Fabrication Topologies
"""
import argparse
import struct
import sys
from pathlib import Path
import cv2
import numpy as np
from PIL import Image
def load_and_prepare(image_path: str, max_px: int) -> np.ndarray:
img = Image.open(image_path).convert("RGB")
ratio = min(max_px / img.width, max_px / img.height)
new_w = max(2, int(img.width * ratio))
new_h = max(2, int(img.height * ratio))
img = img.resize((new_w, new_h), Image.LANCZOS)
return np.array(img.convert("L"), dtype=np.float32)
def smooth_heightmap(gray: np.ndarray, blur: float) -> np.ndarray:
if blur <= 0:
return gray
result = cv2.bilateralFilter(
gray.astype(np.uint8), d=9,
sigmaColor=min(60, blur * 15),
sigmaSpace=min(10, blur * 3)
).astype(np.float32)
return result
def quantize_heightmap(gray: np.ndarray, levels: int) -> np.ndarray:
if levels < 2:
return gray
normalized = gray / 255.0
quantized = np.floor(normalized * levels) / levels
return (quantized * 255.0).astype(np.float32)
def make_tileable(gray: np.ndarray, fade_px: int) -> np.ndarray:
if fade_px <= 0:
return gray
h, w = gray.shape
mean_val = float(np.mean(gray))
result = gray.copy()
fade = min(fade_px, w // 4, h // 4)
for i in range(fade):
alpha = i / fade
result[i, :] = alpha * gray[i, :] + (1 - alpha) * mean_val
result[h - 1 - i, :] = alpha * gray[h - 1 - i, :] + (1 - alpha) * mean_val
result[:, i] = alpha * result[:, i] + (1 - alpha) * mean_val
result[:, w - 1 - i] = alpha * result[:, w - 1 - i] + (1 - alpha) * mean_val
return result
def build_heightmap(
image_path: str,
max_px: int = 500,
invert: bool = False,
blur: float = 0.5,
gamma: float = 1.1,
quantize: int = 0,
tileable: bool = False,
fade_px: int = 8,
):
gray = load_and_prepare(image_path, max_px)
h, w = gray.shape
gray = smooth_heightmap(gray, blur)
gray = np.power(gray / 255.0, gamma) * 255.0
if quantize >= 2:
gray = quantize_heightmap(gray, quantize)
if tileable:
gray = make_tileable(gray, fade_px)
if invert:
gray = 255.0 - gray
gray = np.clip(gray, 0, 255)
return gray / 255.0, w, h
def _pack_triangle(normal, v1, v2, v3) -> bytes:
return struct.pack("<ffffffffffffH", *normal, *v1, *v2, *v3, 0)
def heightmap_to_stl(
height_map: np.ndarray,
px_w: int,
px_h: int,
tile_w_mm: float,
tile_h_mm: float,
relief_mm: float,
base_mm: float,
) -> bytes:
"""
Genera STL da heightmap [0,1].
Se base_mm == 0: mesh aperta sotto (solo superficie + pareti che scendono
fino al punto più basso del rilievo, nessun piano inferiore).
Se base_mm > 0: solido chiuso con base piatta.
"""
pixel_x = tile_w_mm / (px_w - 1)
pixel_y = tile_h_mm / (px_h - 1)
no_base = (base_mm <= 0)
def z_surf(row, col):
# z della superficie di rilievo
return base_mm + height_map[row, col] * relief_mm
def z_bottom(row, col):
# z del bordo inferiore: 0 se c'è la base, minimo locale se no
if no_base:
return height_map[row, col] * relief_mm # inizia dal rilievo stesso (min=0)
return 0.0
triangles = []
# ── Superficie superiore ──
for row in range(px_h - 1):
for col in range(px_w - 1):
x0, x1 = col * pixel_x, (col + 1) * pixel_x
y0 = (px_h - 1 - row) * pixel_y
y1 = (px_h - 2 - row) * pixel_y
z00 = z_surf(row, col)
z10 = z_surf(row, col + 1)
z01 = z_surf(row + 1, col)
z11 = z_surf(row + 1, col + 1)
v00, v10 = (x0, y0, z00), (x1, y0, z10)
v01, v11 = (x0, y1, z01), (x1, y1, z11)
e1 = np.subtract(v10, v00); e2 = np.subtract(v01, v00)
n = np.cross(e1, e2); nn = n / (np.linalg.norm(n) + 1e-12)
triangles.append(_pack_triangle(nn, v00, v10, v01))
e1 = np.subtract(v01, v11); e2 = np.subtract(v10, v11)
n = np.cross(e1, e2); nn = n / (np.linalg.norm(n) + 1e-12)
triangles.append(_pack_triangle(nn, v11, v01, v10))
# ── Base inferiore (solo se base_mm > 0) ──
if not no_base:
for row in range(px_h - 1):
for col in range(px_w - 1):
x0, x1 = col * pixel_x, (col + 1) * pixel_x
y0 = (px_h - 1 - row) * pixel_y
y1 = (px_h - 2 - row) * pixel_y
n = (0.0, 0.0, -1.0)
triangles.append(_pack_triangle(n, (x0,y0,0),(x0,y1,0),(x1,y0,0)))
triangles.append(_pack_triangle(n, (x1,y1,0),(x1,y0,0),(x0,y1,0)))
W, H = tile_w_mm, tile_h_mm
if no_base:
# Senza base: le pareti laterali collegano la superficie al suo stesso
# bordo inferiore (z = height_map * relief_mm, cioè parte da 0 al minimo)
# Fronte (y=0, row = px_h-1)
for col in range(px_w - 1):
x0, x1 = col * pixel_x, (col + 1) * pixel_x
zt0 = z_surf(px_h - 1, col); zt1 = z_surf(px_h - 1, col + 1)
zb0 = height_map[px_h-1, col] * relief_mm
zb1 = height_map[px_h-1, col+1] * relief_mm
n = (0.0, -1.0, 0.0)
triangles.append(_pack_triangle(n, (x0,0,zb0),(x1,0,zt1),(x0,0,zt0)))
triangles.append(_pack_triangle(n, (x0,0,zb0),(x1,0,zb1),(x1,0,zt1)))
# Retro (y=H, row = 0)
for col in range(px_w - 1):
x0, x1 = col * pixel_x, (col + 1) * pixel_x
zt0 = z_surf(0, col); zt1 = z_surf(0, col + 1)
zb0 = height_map[0, col] * relief_mm
zb1 = height_map[0, col+1] * relief_mm
n = (0.0, 1.0, 0.0)
triangles.append(_pack_triangle(n, (x0,H,zt0),(x1,H,zt1),(x0,H,zb0)))
triangles.append(_pack_triangle(n, (x1,H,zt1),(x1,H,zb1),(x0,H,zb0)))
# Sinistra (x=0, col=0)
for row in range(px_h - 1):
y0 = (px_h - 1 - row) * pixel_y
y1 = (px_h - 2 - row) * pixel_y
zt0 = z_surf(row, 0); zt1 = z_surf(row + 1, 0)
zb0 = height_map[row, 0] * relief_mm
zb1 = height_map[row+1, 0] * relief_mm
n = (-1.0, 0.0, 0.0)
triangles.append(_pack_triangle(n, (0,y0,zt0),(0,y1,zt1),(0,y0,zb0)))
triangles.append(_pack_triangle(n, (0,y1,zt1),(0,y1,zb1),(0,y0,zb0)))
# Destra (x=W, col=px_w-1)
for row in range(px_h - 1):
y0 = (px_h - 1 - row) * pixel_y
y1 = (px_h - 2 - row) * pixel_y
zt0 = z_surf(row, px_w-1); zt1 = z_surf(row+1, px_w-1)
zb0 = height_map[row, px_w-1] * relief_mm
zb1 = height_map[row+1, px_w-1] * relief_mm
n = (1.0, 0.0, 0.0)
triangles.append(_pack_triangle(n, (W,y0,zb0),(W,y1,zt1),(W,y0,zt0)))
triangles.append(_pack_triangle(n, (W,y0,zb0),(W,y1,zb1),(W,y1,zt1)))
else:
# Con base: pareti vanno da z=0 alla superficie
# Fronte
for col in range(px_w - 1):
x0, x1 = col * pixel_x, (col + 1) * pixel_x
zt0 = z_surf(px_h-1, col); zt1 = z_surf(px_h-1, col+1)
n = (0.0, -1.0, 0.0)
triangles.append(_pack_triangle(n,(x0,0,0),(x1,0,zt1),(x0,0,zt0)))
triangles.append(_pack_triangle(n,(x0,0,0),(x1,0,0),(x1,0,zt1)))
# Retro
for col in range(px_w - 1):
x0, x1 = col * pixel_x, (col + 1) * pixel_x
zt0 = z_surf(0, col); zt1 = z_surf(0, col+1)
n = (0.0, 1.0, 0.0)
triangles.append(_pack_triangle(n,(x0,H,zt0),(x1,H,zt1),(x0,H,0)))
triangles.append(_pack_triangle(n,(x1,H,zt1),(x1,H,0),(x0,H,0)))
# Sinistra
for row in range(px_h - 1):
y0 = (px_h-1-row)*pixel_y; y1 = (px_h-2-row)*pixel_y
zt0 = z_surf(row,0); zt1 = z_surf(row+1,0)
n = (-1.0, 0.0, 0.0)
triangles.append(_pack_triangle(n,(0,y0,zt0),(0,y1,zt1),(0,y0,0)))
triangles.append(_pack_triangle(n,(0,y1,0),(0,y0,0),(0,y1,zt1)))
# Destra
for row in range(px_h - 1):
y0 = (px_h-1-row)*pixel_y; y1 = (px_h-2-row)*pixel_y
zt0 = z_surf(row,px_w-1); zt1 = z_surf(row+1,px_w-1)
n = (1.0, 0.0, 0.0)
triangles.append(_pack_triangle(n,(W,y0,0),(W,y1,zt1),(W,y0,zt0)))
triangles.append(_pack_triangle(n,(W,y0,0),(W,y1,0),(W,y1,zt1)))
header = b"\x00" * 80
count = struct.pack("<I", len(triangles))
return header + count + b"".join(triangles)
def save_depth_preview(height_map: np.ndarray, out_path: str):
Image.fromarray((height_map * 255).astype(np.uint8)).save(out_path)
# ─────────────────────────────────────────────
# BLOB PIPELINE (pezzi separati)
# ─────────────────────────────────────────────
def _contour_to_stl(contour, height_mm: float) -> bytes:
"""
Converte un singolo contorno OpenCV in un solido STL chiuso
(cilindretto: cappello superiore + base piatta + pareti laterali).
Il contorno è in coordinate pixel; scaliamo dopo.
"""
import struct
pts = contour.reshape(-1, 2).astype(float)
n = len(pts)
if n < 3:
return b""
# Centra e normalizza — scala 1 px = 1 unità, poi il chiamante ri-scala
# Triangolazione fan dalla centroide per cappello e base
cx, cy = pts[:, 0].mean(), pts[:, 1].mean()
def pack_tri(normal, v1, v2, v3):
return struct.pack("<ffffffffffffH", *normal, *v1, *v2, *v3, 0)
tris = []
# Cappello superiore (z = height_mm) — fan dalla centroide
for i in range(n):
a = pts[i]
b = pts[(i + 1) % n]
v0 = (cx, cy, height_mm)
v1 = (a[0], a[1], height_mm)
v2 = (b[0], b[1], height_mm)
e1 = (v1[0]-v0[0], v1[1]-v0[1], 0)
e2 = (v2[0]-v0[0], v2[1]-v0[1], 0)
nx = e1[1]*e2[2] - e1[2]*e2[1]
ny = e1[2]*e2[0] - e1[0]*e2[2]
nz = e1[0]*e2[1] - e1[1]*e2[0]
ln = (nx**2+ny**2+nz**2)**0.5 + 1e-12
tris.append(pack_tri((nx/ln, ny/ln, nz/ln), v0, v1, v2))
# Base inferiore (z = 0) — fan inverso
for i in range(n):
a = pts[i]
b = pts[(i + 1) % n]
v0 = (cx, cy, 0)
v1 = (b[0], b[1], 0)
v2 = (a[0], a[1], 0)
tris.append(pack_tri((0, 0, -1), v0, v1, v2))
# Pareti laterali
for i in range(n):
a = pts[i]
b = pts[(i + 1) % n]
# due triangoli per ogni segmento
v00 = (a[0], a[1], 0)
v10 = (b[0], b[1], 0)
v01 = (a[0], a[1], height_mm)
v11 = (b[0], b[1], height_mm)
# normale laterale approssimata
dx, dy = b[0]-a[0], b[1]-a[1]
ln = (dx**2+dy**2)**0.5 + 1e-12
nx, ny = dy/ln, -dx/ln
tris.append(pack_tri((nx, ny, 0), v00, v10, v01))
tris.append(pack_tri((nx, ny, 0), v10, v11, v01))
header = b"\x00" * 80
count = struct.pack("<I", len(tris))
return header + count + b"".join(tris)
def blobs_to_stl(
image_path: str,
tile_w_mm: float,
tile_h_mm: float,
height_mm: float,
threshold: int = 127,
min_area_px: int = 20,
) -> tuple[bytes, int]:
"""
Pipeline blob: soglia binaria → findContours → un cilindretto per blob.
Restituisce (stl_bytes, n_blobs).
Tutti i cilindretti sono nella stessa mesh (un solo file STL).
"""
img = Image.open(image_path).convert("L")
arr = np.array(img, dtype=np.uint8)
# Scala pixel → mm
scale_x = tile_w_mm / arr.shape[1]
scale_y = tile_h_mm / arr.shape[0]
# Soglia binaria
_, binary = cv2.threshold(arr, threshold, 255, cv2.THRESH_BINARY)
# Trova contorni esterni
contours, _ = cv2.findContours(binary, cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_SIMPLE)
all_tris = []
n_blobs = 0
for cnt in contours:
area = cv2.contourArea(cnt)
if area < min_area_px:
continue
# Semplifica il contorno per ridurre la mesh
eps = 0.01 * cv2.arcLength(cnt, True)
cnt = cv2.approxPolyDP(cnt, eps, True)
pts = cnt.reshape(-1, 2).astype(float)
if len(pts) < 3:
continue
# Converti pixel → mm (y invertita: OpenCV ha y verso il basso)
pts[:, 0] *= scale_x
pts[:, 1] = tile_h_mm - pts[:, 1] * scale_y
# Genera STL del singolo blob
blob_bytes = _contour_to_stl(pts.reshape(-1, 1, 2), height_mm)
if blob_bytes:
# Estrai solo i triangoli (salta header 84 byte)
n_tri = (len(blob_bytes) - 84) // 50
all_tris.append(blob_bytes[84:])
n_blobs += 1
if not all_tris:
raise ValueError("Nessun blob trovato. Prova ad abbassare la soglia o caricare un'immagine con forme bianche su sfondo nero.")
import struct
total_tris = sum(len(t) // 50 for t in all_tris)
header = b"\x00" * 80
count = struct.pack("<I", total_tris)
return header + count + b"".join(all_tris), n_blobs
def blob_preview(image_path: str, threshold: int = 127, min_area_px: int = 20) -> np.ndarray:
"""Restituisce un'immagine con i contorni trovati disegnati sopra l'originale."""
img = Image.open(image_path).convert("L")
arr = np.array(img, dtype=np.uint8)
_, binary = cv2.threshold(arr, threshold, 255, cv2.THRESH_BINARY)
contours, _ = cv2.findContours(binary, cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_SIMPLE)
preview = cv2.cvtColor(arr, cv2.COLOR_GRAY2RGB)
valid = [c for c in contours if cv2.contourArea(c) >= min_area_px]
cv2.drawContours(preview, valid, -1, (212, 169, 18), 2)
return preview, len(valid)
|