{
  "id": 196339,
  "title": "How to remove pixelated grid effect from the generated Image?",
  "url": "/competitions/gan-getting-started/discussion/196339",
  "author_name": "",
  "post_date": "2020-11-10T18:22:09.770318700Z",
  "votes": null,
  "comment_count": 2,
  "views": 0,
  "content": "<p>Every time I train my model, I get this pixelated grid-like noise in my generated image. Any ideas to remove this noise and make the generated image more clear?<br>\nPS: the output generated image is attached below</p>",
  "messages": [
    {
      "id": "1074503",
      "postDate": "11/10/2020 18:22:09",
      "content": "<p>Every time I train my model, I get this pixelated grid-like noise in my generated image. Any ideas to remove this noise and make the generated image more clear?<br>\nPS: the output generated image is attached below</p>",
      "rawMarkdown": "Every time I train my model, I get this pixelated grid-like noise in my generated image. Any ideas to remove this noise and make the generated image more clear?\nPS: the output generated image is attached below",
      "votes": null
    },
    {
      "id": "1076477",
      "postDate": "11/12/2020 15:13:49",
      "content": "<p>In my (limited) experience, these artifacts usually depends on the choice mainly of the loss function, and the other hyperparameters. You might try MSE if you are not already using it, as it usually produces more \"blurred\" results</p>",
      "rawMarkdown": "In my (limited) experience, these artifacts usually depends on the choice mainly of the loss function, and the other hyperparameters. You might try MSE if you are not already using it, as it usually produces more \"blurred\" results",
      "votes": null
    },
    {
      "id": "1077361",
      "postDate": "11/13/2020 14:12:44",
      "content": "<p><a href=\"https://distill.pub/2016/deconv-checkerboard/\" target=\"_blank\">https://distill.pub/2016/deconv-checkerboard/</a></p>\n<p>This article tackles to solve this problem. </p>\n<p>\"Upsampling + Convolution\" is recommended.</p>\n<p>FYI.</p>",
      "rawMarkdown": "https://distill.pub/2016/deconv-checkerboard/\n\nThis article tackles to solve this problem. \n\n\"Upsampling + Convolution\" is recommended.\n\nFYI.",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 1076477,
      "author_name": "carloalbertobarbano",
      "author_url": "",
      "post_date": "11/12/2020 15:13:49",
      "content": "<p>In my (limited) experience, these artifacts usually depends on the choice mainly of the loss function, and the other hyperparameters. You might try MSE if you are not already using it, as it usually produces more \"blurred\" results</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 1077361,
      "author_name": "stealthflow",
      "author_url": "",
      "post_date": "11/13/2020 14:12:44",
      "content": "<p><a href=\"https://distill.pub/2016/deconv-checkerboard/\" target=\"_blank\">https://distill.pub/2016/deconv-checkerboard/</a></p>\n<p>This article tackles to solve this problem. </p>\n<p>\"Upsampling + Convolution\" is recommended.</p>\n<p>FYI.</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "1074503": "Every time I train my model, I get this pixelated grid-like noise in my generated image. Any ideas to remove this noise and make the generated image more clear?\nPS: the output generated image is attached below",
    "1076477": "In my (limited) experience, these artifacts usually depends on the choice mainly of the loss function, and the other hyperparameters. You might try MSE if you are not already using it, as it usually produces more \"blurred\" results",
    "1077361": "https://distill.pub/2016/deconv-checkerboard/\n\nThis article tackles to solve this problem. \n\n\"Upsampling + Convolution\" is recommended.\n\nFYI."
  },
  "source": "meta"
}