EN EL CENTRO DE LA SEGUNDA IMAGEN HAY UNA VIRGEN Y...

python
from PIL import Image, ImageEnhance, ImageFilter
import numpy as np, os, matplotlib.pyplot as plt, math, cv2, PIL
img1_path="/mnt/data/Captura de pantalla 2025-12-14 143524.png"
img2_path="/mnt/data/Captura de pantalla 2025-12-07 202059vr.png"
img1=Image.open(img1_path).convert("RGBA")
img2=Image.open(img2_path).convert("RGBA")
img1.size, img2.size


((676, 1104), (930, 1658))

python
import matplotlib.pyplot as plt
plt.figure(figsize=(6,10))
plt.imshow(img1)
plt.axis('off')
plt.show()

python
plt.figure(figsize=(6,10))
plt.imshow(img2)
plt.axis('off')
plt.show()

python
import numpy as np
def show_crop(im, box, figsize=(5,5)):
    crop=im.crop(box)
    plt.figure(figsize=figsize)
    plt.imshow(crop)
    plt.axis('off')
    plt.show()
    return crop

# rough box for medallion in img2
box_med=(250, 650, 690, 1240)  # (left, top, right, bottom)
crop_med = show_crop(img2, box_med, figsize=(5,7))

python
# show potential oval area
box1=(200, 360, 480, 690)
crop1 = show_crop(img1, box1, figsize=(4,5))

python
box_oval=(235, 420, 445, 670)
crop_oval = show_crop(img1, box_oval, figsize=(4,5))

python
box_oval=(225, 415, 455, 710)
crop_oval = show_crop(img1, box_oval, figsize=(4,6))

python
import numpy as np, cv2, matplotlib.pyplot as plt
crop_med_np = np.array(crop_med.convert("RGB"))
h,w = crop_med_np.shape[:2]
h,w


(590, 440)

python
plt.figure(figsize=(6,8))
plt.imshow(crop_med_np)
plt.axis('on')
plt.show()

python
import numpy as np, cv2
img_gc = cv2.cvtColor(crop_med_np, cv2.COLOR_RGB2BGR)
mask = np.zeros(img_gc.shape[:2], np.uint8)
bgdModel = np.zeros((1,65), np.float64)
fgdModel = np.zeros((1,65), np.float64)
rect = (115, 195, 215, 360)  # x,y,w,h
cv2.grabCut(img_gc, mask, rect, bgdModel, fgdModel, 5, cv2.GC_INIT_WITH_RECT)
mask2 = np.where((mask==2)|(mask==0), 0, 1).astype('uint8')
# refine with morphological operations
kernel = np.ones((3,3),np.uint8)
mask2 = cv2.morphologyEx(mask2, cv2.MORPH_OPEN, kernel, iterations=1)
mask2 = cv2.morphologyEx(mask2, cv2.MORPH_DILATE, kernel, iterations=1)
# create rgba
rgba = np.dstack([crop_med_np, mask2*255]).astype(np.uint8)
statue = Image.fromarray(rgba, 'RGBA')
# crop to bounding box of mask
coords = np.argwhere(mask2==1)
y0,x0 = coords.min(axis=0)
y1,x1 = coords.max(axis=0)+1
statue_cropped = statue.crop((x0,y0,x1,y1))
statue_cropped.size


(213, 352)

python
plt.figure(figsize=(4,6))
plt.imshow(statue_cropped)
plt.axis('off')
plt.show()

python
crown_crop = crop_med.crop((60, 0, 380, 210))
plt.figure(figsize=(6,4))
plt.imshow(crown_crop)
plt.axis('off')
plt.show()

python
crown_np = np.array(crown_crop.convert("RGB"))
img_c = cv2.cvtColor(crown_np, cv2.COLOR_RGB2BGR)
mask = np.zeros(img_c.shape[:2], np.uint8)
bgdModel = np.zeros((1,65), np.float64)
fgdModel = np.zeros((1,65), np.float64)
rect = (15, 20, crown_np.shape[1]-30, crown_np.shape[0]-40)
cv2.grabCut(img_c, mask, rect, bgdModel, fgdModel, 5, cv2.GC_INIT_WITH_RECT)
mask2 = np.where((mask==2)|(mask==0), 0, 1).astype('uint8')
# refine: remove small background by keeping largest component
num_labels, labels, stats, centroids = cv2.connectedComponentsWithStats(mask2, connectivity=8)
# ignore background label 0
areas = stats[1:, cv2.CC_STAT_AREA]
if len(areas)>0:
    largest = 1 + areas.argmax()
    mask2 = (labels==largest).astype('uint8')
mask2 = cv2.morphologyEx(mask2, cv2.MORPH_CLOSE, np.ones((5,5),np.uint8), iterations=1)
rgba = np.dstack([crown_np, mask2*255]).astype(np.uint8)
crown_rgba = Image.fromarray(rgba, 'RGBA')
# crop to bbox
coords = np.argwhere(mask2==1)
y0,x0 = coords.min(axis=0)
y1,x1 = coords.max(axis=0)+1
crown_ex = crown_rgba.crop((x0,y0,x1,y1))
crown_ex.size


(258, 146)

python
plt.figure(figsize=(6,3))
plt.imshow(crown_ex)
plt.axis('off')
plt.show()

python
# visualize full img1 with coordinate grid for approximate
import matplotlib.pyplot as plt
fig = plt.figure(figsize=(6,10))
ax = plt.gca()
ax.imshow(img1)
ax.set_xlim(150, 530)
ax.set_ylim(760, 300)  # zoom on center top area with oval
ax.axis('on')
plt.show()

python
pix = img1.getpixel((265, 600))
pix


(235, 203, 90, 255)

python
from PIL import ImageDraw

# create crown with cross
crown = crown_ex.copy()
cw,ch = crown.size
cross_w = int(cw*0.09)  # width of cross stem
cross_h = int(ch*0.35)
arm_w = int(cw*0.22)
arm_h = int(ch*0.08)
# create transparent layer for cross
cross_layer = Image.new("RGBA", (cw, ch), (0,0,0,0))
draw = ImageDraw.Draw(cross_layer)
# position at top center
cx = cw//2
top_y = int(ch*0.02)
# vertical stem rectangle
stem_x0 = cx - cross_w//2
stem_x1 = cx + cross_w//2
stem_y0 = top_y
stem_y1 = top_y + cross_h
# horizontal arms
arm_y0 = top_y + int(cross_h*0.35) - arm_h//2
arm_y1 = arm_y0 + arm_h
arm_x0 = cx - arm_w//2
arm_x

python
extra = int(ch*0.35)
new_canvas = Image.new("RGBA", (cw, ch+extra), (0,0,0,0))
new_canvas.paste(crown, (0, extra), crown)
draw = ImageDraw.Draw(new_canvas)
# cross dimensions relative to cw
cross_h = int(extra*0.9)
cross_w = int(cw*0.06)
arm_w = int(cw*0.18)
arm_h = int(extra*0.18)
cx = cw//2
top_y = int(extra*0.05)
stem_x0 = cx - cross_w//2
stem_x1 = cx + cross_w//2
stem_y0 = top_y
stem_y1 = top_y + cross_h
arm_y0 = top_y + int(cross_h*0.40) - arm_h//2
arm_y1 = arm_y0 + arm_h
arm_x0 = cx - arm_w//2
arm_x1 = cx + arm_w//2
gold = (240, 210, 130, 255)
shadow = (120, 90, 40, 140)
off=2
# shadow
draw.rectangle([stem_x0+off, stem_y0+off, stem_x1+off, stem_y1+off], fill=shadow)
draw.rectangle([arm_x0+off, arm_y0+off, arm_x1+off, arm_y1+off], fill=shadow)
# main
draw.rectangle([stem_x0, stem_y0, stem_x1, stem_y1], fill=gold)
draw.rectangle([arm_x0, arm_y0, arm_x1, arm_y1], fill=gold)
crown_with_cross2=new_canvas
plt.figure(figsize=(6,4))
plt.imshow(crown_with_cross2)
plt.axis('off')
plt.show()
crown_with_cross2.size


(258, 197)

python
from PIL import ImageChops
base = img1.copy()

# Fill inner ellipse with sampled gold to remove old silhouette (but keep border)
inner_bbox = (250, 520, 428, 716)
gold_color = base.getpixel((265, 600))
fill_layer = Image.new("RGBA", base.size, (0,0,0,0))
draw = ImageDraw.Draw(fill_layer)
draw.ellipse(inner_bbox, fill=gold_color)
base = Image.alpha_composite(base, fill_layer)

# Scale and paste statue
ellipse_w = inner_bbox[2]-inner_bbox[0]
ellipse_h = inner_bbox[3]-inner_bbox[1]
target_w = int(ellipse_w*0.70)
target_h = int(ellipse_h*0.92)
s_w,s_h = statue_cropped.size
scale = min(target_w/s_w, target_h/s_h)
new_size = (max(1,int(s_w*scale)), max(1,int(s_h*scale)))
statue_resized = statue_cropped.resize(new_size, resample=Image.Resampling.LANCZOS)

# paste centered in ellipse
cx = (inner_bbox[0]+inner_bbox[2])//2
cy = (inner_bbox[1]+inner_bbox[3])//2 + int(ellipse_h*0.05)  # slightly down
pos = (cx - new_size[0]//2, cy - new_size[1]//2)
base.paste(statue_resized, pos, statue_resized)

# Add cross above existing gold crown
cross_layer = Image.new("RGBA", base.size, (0,0,0,0))
draw = ImageDraw.Draw(cross_layer)

# cross dimensions relative to image
cross_h = 45
cross_w = 14
arm_w = 40
arm_h = 12
cx = base.size[0]//2  # 338
top_y = 420
stem = [cx - cross_w//2, top_y, cx + cross_w//2, top_y + cross_h]
arm_y = top_y + int(cross_h*0.35)
arm = [cx - arm_w//2, arm_y - arm_h//2, cx + arm_w//2, arm_y + arm_h//2]
gold = (240, 210, 130, 255)
shadow = (90, 60, 20, 120)
off=2
draw.rectangle([stem[0]+off, stem[1]+off, stem[2]+off, stem[3]+off], fill=shadow)
draw.rectangle([arm[0]+off, arm[1]+off, arm[2]+off, arm[3]+off], fill=shadow)
draw.rectangle(stem, fill=gold)
draw.rectangle(arm, fill=gold)

base = Image.alpha_composite(base, cross_layer)

# Refine and mark lines: unsharp mask and slight contrast
enhanced = base.filter(ImageFilter.UnsharpMask(radius=2, percent=180, threshold=3))
enhanced = ImageEnhance.Contrast(enhanced).enhance(1.08)
enhanced = ImageEnhance.Sharpness(enhanced).enhance(1.15)

# show result
plt.figure(figsize=(6,10))
plt.imshow(enhanced)
plt.axis('off')
plt.show()

python
def add_drop_shadow(rgba_img, offset=(3,3), shadow_color=(0,0,0,140), blur_radius=4):
    # rgba_img: PIL RGBA
    w,h = rgba_img.size
    shadow = Image.new("RGBA", (w+abs(offset[0])*2, h+abs(offset[1])*2), (0,0,0,0))
    # Create shadow from alpha
    alpha = rgba_img.split()[-1]
    shadow_alpha = alpha.copy().filter(ImageFilter.GaussianBlur(blur_radius))
    shadow_layer = Image.new("RGBA", rgba_img.size, shadow_color)
    shadow_layer.putalpha(shadow_alpha)
    sx = abs(offset[0]) + max(offset[0],0)
    sy = abs(offset[1]) + max(offset[1],0)
    shadow.paste(shadow_layer, (sx, sy), shadow_layer)
    # Paste original

python
out_path="/mnt/data/estandarte_editado.png"
enhanced2.save(out_path)
out_path, os.path.getsize(out_path)


('/mnt/data/estandarte_editado.png', 1107155)