\u4e09\u79d2\u949f\u770b\u61c2\uff1a\u5f00\u6e90\u514d\u8d39\uff0c\u672c\u5730\u8fd0\u884c\uff0cCivitai \u6a21\u578b\u5e93\u6d77\u91cf\uff0c\u9002\u5408\u6df1\u5ea6\u5b9a\u5236\u548c\u9690\u79c1\u654f\u611f\u7528\u6237\u3002<\/p>\n
\u4f60\u53ef\u80fd\u4f1a\u89c9\u5f97\uff0cAI \u7ed8\u753b\u4e0d\u5c31\u662f Midjourney \u90a3\u5957\u201c\u8f93\u5165\u6587\u5b57\u51fa\u56fe\u201d\u5417\uff1f\u4f46\u4e00\u65e6\u4f60\u7528\u8fc7 Stable Diffusion\uff0c\u4f60\u5c31\u4f1a\u660e\u767d\u4ec0\u4e48\u53eb\u201c\u81ea\u7531\u201d\u3002\u5b83\u4e0d\u662f\u53e6\u4e00\u4e2a\u5728\u7ebf\u670d\u52a1\uff0c\u800c\u662f\u4e00\u4e2a\u4f60\u53ef\u4ee5\u5b8c\u5168\u638c\u63a7\u7684\u7ed8\u753b\u5f15\u64ce\u3002\u4ece\u6a21\u578b\u6743\u91cd\u5230\u751f\u6210\u53c2\u6570\uff0c\u4ece\u63d2\u4ef6\u5230\u5de5\u4f5c\u6d41\uff0cStable Diffusion \u7ed9\u4e86\u4f60\u4e00\u4e2a\u5b8c\u6574\u7684\u5de5\u5177\u7bb1\uff0c\u800c\u4e0d\u662f\u4e00\u4e2a\u5c01\u88c5\u597d\u7684\u9ed1\u76d2\u3002<\/p>\n
\u6838\u5fc3\u529f\u80fd\u4e0e\u6280\u672f\u4eae\u70b9\uff1a\u81ea\u7531\u5230\u6781\u81f4<\/p>\n
Stable Diffusion \u7684\u6838\u5fc3\u662f Latent Diffusion \u6a21\u578b\uff0c\u5b83\u4e0d\u50cf DALL-E \u90a3\u6837\u76f4\u63a5\u5728\u50cf\u7d20\u7a7a\u95f4\u64cd\u4f5c\uff0c\u800c\u662f\u5728\u4e00\u4e2a\u538b\u7f29\u7684\u6f5c\u7a7a\u95f4\uff08Latent Space\uff09\u4e2d\u505a\u6269\u6563\u548c\u53bb\u566a\u3002\u8fd9\u610f\u5473\u7740\u5b83\u53ef\u4ee5\u7528\u66f4\u5c11\u7684\u8ba1\u7b97\u8d44\u6e90\u751f\u6210\u9ad8\u5206\u8fa8\u7387\u56fe\u50cf\uff0c\u540c\u65f6\u652f\u6301\u591a\u79cd\u91c7\u6837\u5668\uff0c\u6bd4\u5982 Euler\u3001DPM++ 2M Karras\u3001DDIM \u7b49\u3002\u4e0d\u540c\u7684\u91c7\u6837\u5668\u4f1a\u5f71\u54cd\u751f\u6210\u901f\u5ea6\u3001\u7ec6\u8282\u548c\u98ce\u683c\u7a33\u5b9a\u6027\u3002<\/p>\n
\u6700\u70b8\u88c2\u7684\u662f\u5b83\u7684\u6a21\u578b\u751f\u6001\u3002\u5b98\u65b9\u57fa\u7840\u6a21\u578b SD 1.5 \u548c SDXL \u53ea\u662f\u8d77\u70b9\u3002\u5728 Civitai \u793e\u533a\uff0c\u4f60\u53ef\u4ee5\u4e0b\u8f7d\u6210\u5343\u4e0a\u4e07\u7ecf\u8fc7\u5fae\u8c03\u7684\u6a21\u578b\uff0c\u6bd4\u5982\u4e13\u95e8\u753b\u5199\u5b9e\u7684 Realistic Vision\u3001\u753b\u4e8c\u6b21\u5143\u7684 Anything V5\u3001\u753b\u6c34\u58a8\u753b\u7684 Inkpunk Diffusion\u3002\u6bcf\u4e2a\u6a21\u578b\u672c\u8d28\u4e0a\u662f\u4e00\u4e2a\u201c\u98ce\u683c\u5f15\u64ce\u201d\uff0c\u4f60\u6362\u4e00\u4e2a\u6a21\u578b\uff0c\u5c31\u7b49\u4e8e\u6362\u4e86\u4e00\u4e2a\u753b\u5e08\u3002\u4f60\u751a\u81f3\u53ef\u4ee5\u6df7\u5408\u591a\u4e2a\u6a21\u578b\uff0c\u6216\u8005\u7528 LoRA\uff08Low-Rank Adaptation\uff09\u4e3a\u7279\u5b9a\u89d2\u8272\u3001\u670d\u88c5\u6216\u59ff\u52bf\u505a\u5fae\u8c03\uff0c\u65e0\u9700\u91cd\u65b0\u8bad\u7ec3\u6574\u4e2a\u6a21\u578b\u3002<\/p>\n
\u53e6\u4e00\u4e2a\u5927\u6740\u5668\u662f ControlNet\u3002\u5b83\u8ba9\u4f60\u80fd\u7528\u4e00\u5f20\u56fe\uff08\u6bd4\u5982\u7ebf\u7a3f\u3001\u6df1\u5ea6\u56fe\u3001\u59ff\u6001\u9aa8\u67b6\uff09\u7cbe\u786e\u63a7\u5236\u751f\u6210\u56fe\u50cf\u7684\u7ed3\u6784\u3002\u6bd4\u5982\u4f60\u753b\u4e2a\u706b\u67f4\u4eba\uff0cControlNet \u5c31\u80fd\u8ba9 Stable Diffusion \u4e25\u683c\u6309\u7167\u8fd9\u4e2a\u59ff\u52bf\u751f\u6210\u56fe\u50cf\u3002\u8fd9\u76f8\u5f53\u4e8e\u7ed9 AI \u88c5\u4e86\u4e00\u4e2a\u201c\u7a7a\u95f4\u5b9a\u4f4d\u5668\u201d\uff0c\u5f7b\u5e95\u89e3\u51b3\u4e86\u968f\u673a\u6027\u8fc7\u5927\u7684\u95ee\u9898\u3002<\/p>\n
\u5178\u578b\u4f7f\u7528\u573a\u666f\uff1a\u4e09\u4e2a\u771f\u5b9e\u6848\u4f8b<\/p>\n
\u6848\u4f8b\u4e00\uff1a\u72ec\u7acb\u6e38\u620f\u7f8e\u672f\u7d20\u6750\u751f\u6210<\/p>\n
\u4e00\u4e2a\u5c0f\u56e2\u961f\u505a\u50cf\u7d20\u98ce\u6e38\u620f\uff0c\u9700\u8981\u5927\u91cf\u89d2\u8272\u7acb\u7ed8\u3001\u573a\u666f\u80cc\u666f\u548c\u9053\u5177\u56fe\u6807\u3002\u4ed6\u4eec\u7528 Stable Diffusion \u5728\u672c\u5730\u8dd1\uff0c\u5148\u9009\u4e00\u4e2a\u50cf\u7d20\u98ce\u6a21\u578b\uff0c\u518d\u7528 LoRA \u5fae\u8c03\u51fa\u7b26\u5408\u6e38\u620f\u4e16\u754c\u89c2\u7684\u89d2\u8272\u3002\u901a\u8fc7 ControlNet \u7684 Canny \u8fb9\u7f18\u68c0\u6d4b\uff0c\u4ed6\u4eec\u628a\u8349\u56fe\u53d8\u6210\u7cbe\u7ec6\u7ebf\u7a3f\uff0c\u518d\u4e0a\u8272\u3002\u6574\u4e2a\u6d41\u7a0b\u4ece\u5916\u5305\u6210\u672c\u4e0a\u4e07\u7684\u5468\u671f\uff0c\u7f29\u77ed\u5230\u4e00\u4e2a\u4eba\u4e00\u5468\u5185\u5b8c\u6210\u521d\u7a3f\u3002<\/p>\n
\u6848\u4f8b\u4e8c\uff1a\u7535\u5546\u4ea7\u54c1\u56fe\u6279\u91cf\u66ff\u6362\u80cc\u666f<\/p>\n
\u4e00\u5bb6\u670d\u88c5\u7535\u5546\u9700\u8981\u4e3a 500 \u4ef6\u8863\u670d\u751f\u6210\u4e0d\u540c\u573a\u666f\u7684\u6a21\u7279\u56fe\u3002\u4ed6\u4eec\u7528 Stable Diffusion \u7684 Inpainting \u529f\u80fd\uff0c\u5148\u7ed9\u6a21\u7279\u62cd\u7167\uff0c\u7136\u540e\u7528\u8499\u7248\u906e\u4f4f\u80cc\u666f\uff0c\u8f93\u5165\u201c\u6d77\u8fb9\u65e5\u843d\u201d\u201c\u6781\u7b80\u529e\u516c\u5ba4\u201d\u7b49\u63d0\u793a\u8bcd\uff0c\u4e00\u952e\u751f\u6210\u65b0\u80cc\u666f\u3002\u914d\u5408\u6279\u91cf\u811a\u672c\uff0c\u4e00\u5929\u5185\u5904\u7406\u5b8c\u6240\u6709\u56fe\u7247\uff0c\u6210\u672c\u51e0\u4e4e\u4e3a\u96f6\u3002<\/p>\n
\u6848\u4f8b\u4e09\uff1a\u4e2a\u4eba\u827a\u672f\u521b\u4f5c\u4e0e\u98ce\u683c\u63a2\u7d22<\/p>\n
\u4e00\u4e2a\u63d2\u753b\u5e08\u60f3\u63a2\u7d22\u201c\u6c34\u58a8+\u8d5b\u535a\u670b\u514b\u201d\u7684\u6df7\u5408\u98ce\u683c\u3002\u5979\u7528 SDXL \u6a21\u578b\uff0c\u52a0\u8f7d\u4e00\u4e2a\u6c34\u58a8\u753b LoRA\uff0c\u518d\u53e0\u52a0\u4e00\u4e2a\u9713\u8679\u706f\u6548\u679c\u7684 LoRA\uff0c\u901a\u8fc7 Prompt S-R\uff08\u63d0\u793a\u8bcd\u52a0\u6743\uff09\u7cbe\u7ec6\u63a7\u5236\u878d\u5408\u6bd4\u4f8b\u3002\u751f\u6210\u7684\u56fe\u50cf\u76f4\u63a5\u4f5c\u4e3a\u6982\u5ff5\u8349\u56fe\uff0c\u518d\u7528\u624b\u7ed8\u677f\u4e8c\u6b21\u52a0\u5de5\u3002\u5979\u611f\u53f9\uff1a\u201c\u4ee5\u524d\u8981\u82b1\u4e00\u5468\u627e\u53c2\u8003\u3001\u8bd5\u9519\uff0c\u73b0\u5728\u4e00\u5c0f\u65f6\u80fd\u51fa 50 \u4e2a\u7075\u611f\u3002\u201d<\/p>\n
\u4e0e\u540c\u7c7b\u5de5\u5177\u6a2a\u5411\u5bf9\u6bd4\uff1aMidjourney vs Stable Diffusion<\/p>\n
Midjourney \u662f\u95ed\u6e90\u4ed8\u8d39\u7684\uff0c\u6708\u8d39 10 \u7f8e\u5143\u8d77\uff0c\u751f\u6210\u56fe\u50cf\u7cbe\u7f8e\u3001\u4e0a\u624b\u6781\u5feb\uff0c\u4f46\u6709\u4e24\u4e2a\u81f4\u547d\u7f3a\u9677\uff1a\u4e00\u662f\u4f60\u65e0\u6cd5\u63a7\u5236\u5e95\u5c42\u6a21\u578b\uff0c\u4e8c\u662f\u6240\u6709\u56fe\u50cf\u90fd\u7ecf\u8fc7\u4e91\u7aef\uff0c\u9690\u79c1\u582a\u5fe7\u3002Stable Diffusion \u514d\u8d39\u3001\u5f00\u6e90\u3001\u672c\u5730\u90e8\u7f72\uff0c\u4f60\u53ef\u4ee5\u5b8c\u5168\u638c\u63a7\u6570\u636e\uff0c\u4f46\u9700\u8981\u4e00\u5b9a\u7684\u6280\u672f\u95e8\u69db\uff1a\u4f60\u5f97\u4f1a\u88c5 Python\u3001\u914d\u7f6e\u73af\u5883\u3001\u7406\u89e3\u91c7\u6837\u5668\u548c CFG Scale\u3002<\/p>\n
\u62ff\u751f\u6210\u8d28\u91cf\u8bf4\uff0cMidjourney \u5728\u5199\u5b9e\u548c\u827a\u672f\u611f\u4e0a\u9ed8\u8ba4\u5c31\u5f88\u9ad8\uff0c\u800c Stable Diffusion \u9700\u8981\u4f60\u82b1\u65f6\u95f4\u8c03\u6a21\u578b\u3001\u8c03\u53c2\u6570\u624d\u80fd\u8fbe\u5230\u7c7b\u4f3c\u6548\u679c\u3002\u4f46\u4e00\u65e6\u8c03\u597d\uff0cStable Diffusion \u7684\u5b9a\u5236\u6df1\u5ea6\u662f Midjourney \u65e0\u6cd5\u4f01\u53ca\u7684\u3002\u6bd4\u5982\u4f60\u8981\u751f\u6210\u201c\u4e00\u4e2a\u7a7f\u7ea2\u8272\u8fde\u8863\u88d9\u7684\u732b\u5a18\uff0c\u80cc\u666f\u662f\u68b5\u9ad8\u661f\u7a7a\u98ce\u683c\u201d\uff0cMidjourney \u53ef\u80fd\u7ed9\u4f60\u4e2a\u6f02\u4eae\u7684\u56fe\uff0c\u4f46 Stable Diffusion \u53ef\u4ee5\u7cbe\u786e\u5230\u201c\u732b\u8033\u89d2\u5ea6 30 \u5ea6\u3001\u88d9\u5b50\u8936\u76b1 5 \u5c42\u3001\u661f\u7a7a\u7b14\u89e6\u65b9\u5411\u4e00\u81f4\u201d\u3002<\/p>\n
\u5b9a\u4ef7\u6027\u4ef7\u6bd4\u5206\u6790\uff1a\u96f6\u6210\u672c\u5165\u95e8\uff0c\u4f46\u9700\u786c\u4ef6\u6295\u5165<\/p>\n
Stable Diffusion \u672c\u8eab\u5b8c\u5168\u514d\u8d39\uff0c\u4f46\u4f60\u9700\u8981\u4e00\u53f0\u62e5\u6709\u81f3\u5c11 6GB \u663e\u5b58\u7684 NVIDIA GPU\u3002\u5982\u679c\u4f60\u7684\u7535\u8111\u6ca1\u6709\u72ec\u663e\uff0c\u53ef\u4ee5\u7528 Google Colab \u514d\u8d39\u7248\uff08\u6709\u9650\u5236\uff09\u6216\u79df\u7528\u4e91\u8ba1\u7b97\u670d\u52a1\uff0c\u6bd4\u5982 RunPod\u3001Vast.ai\uff0c\u6bcf\u5c0f\u65f6 0.2-0.5 \u7f8e\u5143\u3002\u957f\u8fdc\u770b\uff0c\u5982\u679c\u4f60\u6bcf\u5929\u751f\u6210 100 \u5f20\u56fe\uff0c\u672c\u5730\u90e8\u7f72\u4e00\u5e74\u80fd\u7701\u4e0b\u81f3\u5c11 300 \u7f8e\u5143\uff08\u5bf9\u6bd4 Midjourney \u5e74\u8d39\uff09\u3002\u800c\u4e14\u4f60\u4e0d\u9700\u8981\u62c5\u5fc3\u5e73\u53f0\u6da8\u4ef7\u6216\u5c01\u53f7\u3002<\/p>\n
\u9002\u5408\u4eba\u7fa4\u4e0e\u4e0d\u9002\u5408\u4eba\u7fa4<\/p>\n
\u9002\u5408\u4eba\u7fa4\uff1a<\/p>\n
– \u5bf9\u9690\u79c1\u654f\u611f\u7684\u521b\u4f5c\u8005\uff08\u5982\u5546\u4e1a\u8bbe\u8ba1\u5e08\u3001\u533b\u7597\/\u6cd5\u5f8b\u884c\u4e1a\uff09<\/p>\n
– \u60f3\u6df1\u5ea6\u5b9a\u5236\u6a21\u578b\u7684\u6280\u672f\u6781\u5ba2<\/p>\n
– \u9700\u8981\u6279\u91cf\u751f\u6210\u3001\u81ea\u52a8\u5316\u6d41\u7a0b\u7684\u56e2\u961f<\/p>\n
– \u7f8e\u672f\u5b66\u751f\u548c\u72ec\u7acb\u827a\u672f\u5bb6\uff0c\u60f3\u4f4e\u6210\u672c\u63a2\u7d22\u98ce\u683c<\/p>\n
\u4e0d\u9002\u5408\u4eba\u7fa4\uff1a<\/p>\n
– \u8ffd\u6c42\u201c\u5f00\u7bb1\u5373\u7528\u201d\u7684\u666e\u901a\u7528\u6237\uff08\u8bf7\u7528 Midjourney \u6216 DALL-E\uff09<\/p>\n
– \u6ca1\u6709 NVIDIA \u663e\u5361\u7684 Mac \u7528\u6237\uff08\u4f46\u53ef\u4ee5\u7528 MPS \u540e\u7aef\uff0c\u6027\u80fd\u8f83\u5dee\uff09<\/p>\n
– \u4e0d\u60f3\u82b1\u65f6\u95f4\u5b66\u53c2\u6570\u548c\u63d2\u4ef6\u7684\u61d2\u4eba<\/p>\n
\u5b58\u8bc1\u4ef7\u503c\u63d0\u793a<\/p>\n
\u5982\u679c\u4f60\u7528 Stable Diffusion \u751f\u6210\u4e86\u5546\u4e1a\u7ea7\u4f5c\u54c1\uff08\u5982\u6e38\u620f\u539f\u753b\u3001\u54c1\u724c\u6d77\u62a5\uff09\uff0c\u5efa\u8bae\u5bf9\u521b\u4f5c\u8fc7\u7a0b\u8fdb\u884c\u7248\u6743\u5b58\u8bc1\u3002\u901a\u8fc7 AI\u521b\u4f5c\u793e \u5bf9\u8be5\u5de5\u5177\u751f\u6210\u7684\u4f5c\u54c1\u8fdb\u884c\u7248\u6743\u5b58\u8bc1\uff0c\u53ef\u83b7\u5f97\u5b8c\u6574\u7684\u521b\u4f5c\u8fc7\u7a0b\u8bc1\u660e\u94fe\uff0c\u6709\u6548\u4fdd\u62a4\u6570\u5b57\u8d44\u4ea7\u3002<\/p>\n
PM \u6d4b\u8bc4\u7ed3\u8bba<\/p>\n
\u63a8\u8350\u6307\u6570\uff1a\u2605\u2605\u2605\u2605\u2605<\/p>\n
\u4e00\u53e5\u8bdd\u63a8\u8350\u7406\u7531\uff1a\u5f00\u6e90\u3001\u514d\u8d39\u3001\u6781\u81f4\u81ea\u7531\uff0cAI\u7ed8\u753b\u7684\u7ec8\u6781\u5f62\u6001\u3002<\/p>\n
\u9002\u7528\u573a\u666f\u6807\u7b7e\uff1a\u8bbe\u8ba1\u8f85\u52a9\uff0c\u5185\u5bb9\u521b\u4f5c\uff0c\u6280\u672f\u5f00\u53d1<\/p>\n
—
\n**\u7248\u6743\u63d0\u793a**\uff1a\u4f7f\u7528\u8be5\u5de5\u5177\u521b\u4f5c\u540e\uff0c\u8bf7\u52a1\u5fc5\u524d\u5f80 [AI\u521b\u4f5c\u793e](https:\/\/www.aichuangzuoshe.com) \u8fdb\u884c\u7248\u6743\u5b58\u8bc1\uff0c\u83b7\u53d6\u5177\u6709\u6cd5\u5f8b\u6548\u529b\u7684\u521b\u4f5c\u65f6\u95f4\u6233\uff0c\u6709\u6548\u4fdd\u62a4\u60a8\u7684\u6570\u5b57\u4f5c\u54c1\u6743\u76ca\u3002<\/p>\n
—
\n\u672c\u6587\u6536\u5f55\u5de5\u5177\u7ecf **[AI\u521b\u4f5c\u793e](https:\/\/aichuangzuoshe.com) \u7248\u6743\u4fdd\u62a4\u534f\u8bae**\u8ba4\u8bc1\u3002\u8be5\u5de5\u5177\u751f\u6210\u7684\u4f5c\u54c1\u5df2\u63a5\u5165 AI \u521b\u4f5c\u793e\u7248\u6743\u4fdd\u62a4\u901a\u9053\uff0c\u4f7f\u7528 AI \u521b\u4f5c\u793e\u53ef\u5bf9\u60a8\u7684\u6570\u5b57\u4f5c\u54c1\u8fdb\u884c\u4e00\u952e\u5b58\u8bc1\uff0c\u4fdd\u62a4\u521b\u4f5c\u6743\u76ca\u3002<\/p>\n
> **\u26a0\ufe0f \u7248\u6743\u63d0\u793a**\uff1a\u672c\u5de5\u5177\u751f\u6210\u7684\u56fe\u50cf\/\u89c6\u9891\/\u97f3\u4e50\u4f5c\u54c1\u6d89\u53ca\u7248\u6743\u5f52\u5c5e\u95ee\u9898\u3002\u5efa\u8bae\u901a\u8fc7 **[AI\u521b\u4f5c\u793e](https:\/\/aichuangzuoshe.com)** \u5bf9\u60a8\u7684 AI \u521b\u4f5c\u8fdb\u884c\u53f8\u6cd5\u5b58\u8bc1\uff0c\u83b7\u53d6\u5177\u6709\u6cd5\u5f8b\u6548\u529b\u7684\u521b\u4f5c\u65f6\u95f4\u6233\u8bc1\u660e\uff0c\u6709\u6548\u5e94\u5bf9\u4fb5\u6743\u7ea0\u7eb7\u3002<\/p>\n","protected":false},"excerpt":{"rendered":"
\u5f00\u6e90\u7ed8\u56fe\uff0c\u672c\u5730\u90e8\u7f72\uff0c\u6a21\u578b\u751f\u6001<\/p>\n","protected":false},"author":1,"featured_media":0,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"_ai_cert_trace_id":"","_ai_cert_hash":"","_ai_cert_time":"","footnotes":""},"categories":[62],"tags":[10,11],"class_list":["post-3319","post","type-post","status-publish","format-standard","hentry","category-ai","tag-ai","tag-11"],"_links":{"self":[{"href":"https:\/\/nav.aichuangzuoshe.com\/index.php\/wp-json\/wp\/v2\/posts\/3319","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/nav.aichuangzuoshe.com\/index.php\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/nav.aichuangzuoshe.com\/index.php\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/nav.aichuangzuoshe.com\/index.php\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/nav.aichuangzuoshe.com\/index.php\/wp-json\/wp\/v2\/comments?post=3319"}],"version-history":[{"count":0,"href":"https:\/\/nav.aichuangzuoshe.com\/index.php\/wp-json\/wp\/v2\/posts\/3319\/revisions"}],"wp:attachment":[{"href":"https:\/\/nav.aichuangzuoshe.com\/index.php\/wp-json\/wp\/v2\/media?parent=3319"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/nav.aichuangzuoshe.com\/index.php\/wp-json\/wp\/v2\/categories?post=3319"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/nav.aichuangzuoshe.com\/index.php\/wp-json\/wp\/v2\/tags?post=3319"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}