{"id":242,"date":"2022-08-30T22:12:45","date_gmt":"2022-08-30T13:12:45","guid":{"rendered":"https:\/\/sheltie-garage.xyz\/tech\/?p=242"},"modified":"2022-08-30T22:12:49","modified_gmt":"2022-08-30T13:12:49","slug":"stable-diffusion%e3%81%a7img2img%e3%82%92%e5%8b%95%e3%81%8b%e3%81%97%e3%81%9f%e3%81%8b%e3%81%a3%e3%81%9f%e3%83%bb%e3%83%bb%e3%83%bb","status":"publish","type":"post","link":"https:\/\/sheltie-garage.xyz\/tech\/2022\/08\/stable-diffusion%e3%81%a7img2img%e3%82%92%e5%8b%95%e3%81%8b%e3%81%97%e3%81%9f%e3%81%8b%e3%81%a3%e3%81%9f%e3%83%bb%e3%83%bb%e3%83%bb\/","title":{"rendered":"Stable Diffusion\u3067img2img\u3092\u52d5\u304b\u3057\u305f\u304b\u3063\u305f\u30fb\u30fb\u30fb"},"content":{"rendered":"\n

\u524d\u56de\u306e\u8a18\u4e8b<\/a>\u306e\u7d9a\u304d\u3067\u3059\u3002Stable Diffusion\u306etxt2img\u304c\u52d5\u4f5c\u3059\u308b\u3053\u3068\u524d\u63d0\u3067\u8a71\u304c\u9032\u307f\u307e\u3059<\/p>\n\n\n\n

\u305f\u3060\u3001\u5148\u306b\u65ad\u3063\u3066\u304a\u304f\u3068\u3001\u52d5\u304d\u307e\u305b\u3093\u3067\u3057\u305f\u3002\u30b9\u30af\u30ea\u30d7\u30c8\u306f\u8d77\u52d5\u3057\u307e\u3057\u305f\u304c<\/p>\n\n\n\n

3\u6642\u9593\u304f\u3089\u3044\u653e\u7f6e\u3057\u3066\u3082\u30012%\u304f\u3089\u3044\u3057\u304b\u51e6\u7406\u304c\u9032\u307e\u306a\u304b\u3063\u305f\u306e\u3067\u8ae6\u3081\u307e\u3057\u305f<\/strong><\/p>\n\n\n\n

\u305d\u308c\u3067\u3082\u5b9f\u884c\u3057\u3066\u307f\u305f\u3044\uff01\u3068\u3044\u3046\u65b9\u306e\u305f\u3081\u306b\u8a18\u9332\u3060\u3051\u306f\u6b8b\u3057\u3066\u304a\u304d\u307e\u3059\u30fb\u30fb\u30fb<\/p>\n\n\n\n

\u30b9\u30af\u30ea\u30d7\u30c8\u3092\u4fee\u6b63\u3059\u308b<\/h2>\n\n\n\n

cpu\u30e2\u30fc\u30c9\u3067\u52d5\u4f5c\u3059\u308b\u3088\u3046\u306b\u4ee5\u4e0b\u306e\u30d5\u30a1\u30a4\u30eb\u3092\u7de8\u96c6\u3057\u307e\u3059<\/p>\n\n\n\n

scripts\/img2img.py<\/strong>
43\u884c\u76ee\u8fba\u308a\u306e\u90e8\u5206\u3092\u4ee5\u4e0b\u306e\u3088\u3046\u306b\u4fee\u6b63<\/p>\n\n\n\n

    # model.cuda()\n    model.to('cpu')\n    model.eval()\n    return model<\/code><\/pre><\/div>\n\n\n\n
203\u884c\u76ee\u8fba\u308a\u306e\u90e8\u5206\u3092\u4ee5\u4e0b\u306e\u3088\u3046\u306b\u4fee\u6b63<\/p>\n\n\n\n
    # device = torch.device("cuda") if torch.cuda.is_available() else torch.device("cpu")\n    device = torch.device("cpu")\n    model = model.to(device)<\/code><\/pre><\/div>\n\n\n\n
247\u884c\u76ee\u8fba\u308a\u306e\u90e8\u5206\u3092\u4ee5\u4e0b\u306e\u3088\u3046\u306b\u4fee\u6b63<\/p>\n\n\n\n
    with torch.no_grad():\n        # with precision_scope("cuda"):\n        with precision_scope("cpu"):\n            with model.ema_scope():<\/code><\/pre><\/div>\n\n\n\n
ldm\/models\/diffusion\/ddim.py<\/strong>
21\u884c\u76ee\u8fba\u308a\u306eif attr.device != torch.device(\u201ccuda\u201d):\u5468\u8fba\u3092\u4ee5\u4e0b\u306e\u901a\u308a\u4fee\u6b63<\/p>\n\n\n\n
#            if attr.device != torch.device("cuda"):\n#                attr = attr.to(torch.device("cuda"))\n            if attr.device != torch.device("cpu"):\n                attr = attr.to(torch.float32).to(torch.device("cpu")).contiguous()\n        setattr(self, name, attr)<\/code><\/pre><\/div>\n\n\n\n
\u4ee5\u4e0a\u306e\u4fee\u6b63\u3067\u52d5\u304f\u306f\u305a\u3067\u3059<\/p>\n\n\n\n
\u6307\u793a\u7d75\u306e\u6e96\u5099<\/h2>\n\n\n\n
\u751f\u6210\u3059\u308b\u753b\u50cf\u306e\u30d2\u30f3\u30c8\u3068\u306a\u308b\u6307\u793a\u7d75\u3092\u6e96\u5099\u3057\u307e\u3059\u3002
\u304a\u7d75\u304b\u304d\u30bd\u30d5\u30c8\u3092\u5229\u7528\u3057\u3066\u6e96\u5099\u3057\u307e\u3057\u3087\u3046\u3002\u4eca\u56de\u306f\u4ee5\u4e0b\u306e\u753b\u50cf\u3092\u6307\u793a\u7d75\u306b\u5229\u7528\u3057\u307e\u3057\u305f\u3002<\/p>\n\n\n\n
\"\"<\/figure>\n\n\n\n
\u624b\u6301\u3061\u306e\u767d\u4e0a\u30d5\u30d6\u30ad\u3061\u3083\u3093\u3092\u5199\u771f\u306b\u64ae\u308a\u3001\u30c8\u30ec\u30fc\u30b9\u3057\u305f\u3082\u306e\u306b\u306a\u308a\u307e\u3059\u3002
\u3053\u306e\u4e0b\u624b\u304f\u305d\u306a\u30a4\u30e9\u30b9\u30c8\u304c\u3001\u679c\u305f\u3057\u3066\u3069\u306e\u3088\u3046\u306b\u5909\u5316\u3059\u308b\u306e\u304b\u30fb\u30fb\u30fb<\/p>\n\n\n\n
\u30b3\u30de\u30f3\u30c9\u306f\u4ee5\u4e0b\u306e\u3068\u304a\u308a\u3067\u3059<\/p>\n\n\n\n
python scripts\/img2img.py --prompt "shirakami hubuki hololive illustlated by nagishiro mito" --init-img .\/shitae.JPG --strength 0.8 --precision full<\/code><\/pre><\/div>\n\n\n\n
\u4ee5\u4e0a\u3067\u3059<\/h2>\n\n\n\n
\u4e0a\u8a18\u3092\u5b9f\u884c\u3057\u305f\u3068\u3053\u308d\u3001\u3072\u3068\u307e\u305a\u30b9\u30af\u30ea\u30d7\u30c8\u306f\u8d77\u52d5\u3057\u307e\u3057\u305f\u304c\u3001\u5192\u982d\u306b\u8a18\u8f09\u3057\u305f\u901a\u308a\u51e6\u7406\u304c\u5168\u304f\u9032\u307e\u306a\u304b\u3063\u305f\u306e\u3067\u8ae6\u3081\u307e\u3057\u305f\u3002
\u3084\u306f\u308a\u3001CPU\u3067\u6a5f\u68b0\u5b66\u7fd2\u3092\u51e6\u7406\u3059\u308b\u306e\u306f\u7121\u7406\u304c\u3042\u308b\u3088\u3046\u306a\u306e\u3067\u3001\u9031\u672b\u306bWIndowsPC(RTX2080\u642d\u8f09\u306e\u30b2\u30fc\u30df\u30f3\u30b0\u30ce\u30fc\u30c8)\u3067\u518d\u6311\u6226\u3057\u305f\u3044\u3068\u601d\u3044\u307e\u3059<\/p>\n\n\n
\n\t
\n\t\t
<\/div>\n\t\t
\n\t\t\t
\n\t\t\t\t\t\t\t\t\u30eb\u30df\u30ca\u30b9\u30dc\u30c3\u30af\u30b9 \u767d\u4e0a\u30d5\u30d6\u30ad 1\/7\u30b9\u30b1\u30fc\u30eb PVC\u88fd \u5857\u88c5\u6e08\u307f\u5b8c\u6210\u54c1\u30d5\u30a3\u30ae\u30e5\u30a2\t\t\t\t\t\t\t<\/div>\n\n\t\t\t
\n\t\t\t\t\t\t\t\t\t\t\t<\/div>\n\t\t\t\t\t\t