File size: 3,364 Bytes
ef24f03
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
from __future__ import annotations
from pathlib import Path
from typing import List, Tuple
import time, random
import numpy as np
from PIL import Image, ImageFilter, ImageOps

TMP_DIR = Path("/tmp/bgfx"); TMP_DIR.mkdir(parents=True, exist_ok=True)

_PALETTES = {
    "office": [(240,245,250),(210,220,230),(180,190,200)],
    "studio": [(18,18,20),(32,32,36),(58,60,64)],
    "sunset": [(255,183,77),(255,138,101),(244,143,177)],
    "forest": [(46,125,50),(102,187,106),(165,214,167)],
    "ocean": [(33,150,243),(3,169,244),(0,188,212)],
    "minimal": [(245,246,248),(230,232,236),(214,218,224)],
    "warm": [(255,224,178),(255,204,128),(255,171,145)],
    "cool": [(197,202,233),(179,229,252),(178,235,242)],
    "royal": [(63,81,181),(121,134,203),(159,168,218)],
}

def _save_pil(img: Image.Image, stem: str = "ai_bg", ext: str = "png") -> str:
    ts = int(time.time() * 1000)
    p = TMP_DIR / f"{stem}_{ts}.{ext}"
    img.save(p)
    return str(p)

def _palette_from_prompt(prompt: str) -> List[tuple]:
    p = (prompt or "").lower()
    for key, pal in _PALETTES.items():
        if key in p:
            return pal
    random.seed(hash(p) % (2**32 - 1))
    return [tuple(random.randint(90, 200) for _ in range(3)) for _ in range(3)]

def _perlin_like_noise(h: int, w: int, octaves: int = 4) -> np.ndarray:
    acc = np.zeros((h, w), dtype=np.float32)
    for o in range(octaves):
        scale = 2 ** o
        small = np.random.rand(h // scale + 1, w // scale + 1).astype(np.float32)
        small = Image.fromarray((small * 255).astype(np.uint8)).resize((w, h), Image.BILINEAR)
        acc += np.array(small, dtype=np.float32) / 255.0 / (o + 1)
    acc /= max(1e-6, acc.max())
    return acc

def _blend_palette(noise: np.ndarray, palette: List[tuple]) -> Image.Image:
    h, w = noise.shape
    img = np.zeros((h, w, 3), dtype=np.float32)
    t1, t2 = 0.33, 0.66
    c0, c1, c2 = [np.array(c, dtype=np.float32) for c in palette]
    m0, m1, m2 = noise < t1, (noise >= t1) & (noise < t2), noise >= t2
    img[m0], img[m1], img[m2] = c0, c1, c2
    return Image.fromarray(np.clip(img, 0, 255).astype(np.uint8))

def generate_ai_background(
    prompt: str, width: int = 1280, height: int = 720,
    bokeh: float = 0.0, vignette: float = 0.15, contrast: float = 1.05
) -> Tuple[Image.Image, str]:
    palette = _palette_from_prompt(prompt)
    noise = _perlin_like_noise(height, width, octaves=4)
    img = _blend_palette(noise, palette)

    if bokeh > 0:
        img = img.filter(ImageFilter.GaussianBlur(radius=max(0, min(50, bokeh))))

    if vignette > 0:
        import numpy as np
        base = np.array(img).astype(np.float32) / 255.0
        y, x = np.ogrid[:height, :width]
        cx, cy = width / 2, height / 2
        r = np.sqrt((x - cx) ** 2 + (y - cy) ** 2)
        mask = 1 - np.clip(r / (max(width, height) / 1.2), 0, 1)
        mask = (mask ** 2) * (1 - vignette) + vignette
        out = base * mask[..., None]
        img = Image.fromarray(np.clip(out * 255, 0, 255).astype(np.uint8))

    if contrast != 1.0:
        img = ImageOps.autocontrast(img, cutoff=1)
        arr = np.array(img).astype(np.float32)
        mean = arr.mean(axis=(0, 1), keepdims=True)
        arr = (arr - mean) * float(contrast) + mean
        img = Image.fromarray(np.clip(arr, 0, 255).astype(np.uint8))

    path = _save_pil(img)
    return img, path