{"id":5458,"date":"2024-03-13T10:01:31","date_gmt":"2024-03-13T02:01:31","guid":{"rendered":"https:\/\/www.1ai.net\/?p=5458"},"modified":"2024-03-13T10:01:31","modified_gmt":"2024-03-13T02:01:31","slug":"midjourney%e8%83%bd%e8%ae%a9%e8%a7%92%e8%89%b2%e4%bf%9d%e6%8c%81%e4%b8%80%e8%87%b4%e4%ba%86%ef%bc%8c%e7%9c%8b%e7%9c%8bmidjourney%e6%8e%a7%e5%88%b6%e8%a7%92%e8%89%b2%e4%b8%80%e8%87%b4%e6%80%a7%e7%9a%84","status":"publish","type":"post","link":"https:\/\/www.1ai.net\/en\/5458.html","title":{"rendered":"Midjourney can make the role consistent. Let&#039;s take a look at how midjourney controls the consistency of roles."},"content":{"rendered":"<p><a href=\"https:\/\/www.1ai.net\/en\/tag\/midjourney\" title=\"_Other Organiser\" target=\"_blank\" >Midjourney<\/a>An official message was released:<\/p>\n<p><img loading=\"lazy\" decoding=\"async\" class=\"alignnone size-full wp-image-5459\" title=\"640-37\" src=\"https:\/\/www.1ai.net\/wp-content\/uploads\/2024\/03\/640-37.png\" alt=\"640-37\" width=\"1080\" height=\"466\" \/><\/p>\n<p>I'll go ahead and translate MJ's official message in the screenshot above:<\/p>\n<p>Hey @y'all today we're testing a new \"Character Reference\" feature!<\/p>\n<p>This is similar to the Style Reference feature, except that instead of matching a reference style, it tries to match the character to a \"character reference\" image.<\/p>\n<p><strong>How it works:<\/strong><\/p>\n<ul class=\"list-paddingleft-1\">\n<li>Enter the --cref URL after the hint, which points to an image of a character<\/li>\n<li>you can use --cw to modify the reference strength from 100 to zero<\/li>\n<li>strength 100 (-cw 100) is the default value, using facial, hair and clothing<\/li>\n<li>the strength 0 (-cw 0) will only focus on the face (suitable to change clothes\/ hair etc.)<\/li>\n<\/ul>\n<section><strong>What it's used for:<\/strong><\/section>\n<ul class=\"list-paddingleft-1\">\n<li>\n<section>This feature works best with characters made using Midjourney images. It is not designed for real people\/photos (and may distort them like regular image prompts)<\/section>\n<\/li>\n<li>\n<section>Cref works similarly to regular image cues, except that it \"focuses\" on character traits.<\/section>\n<\/li>\n<li>\n<section>The accuracy of this technique is limited, it cannot replicate exact dimples\/freckles\/ or t-shirt logos<\/section>\n<\/li>\n<li>\n<section>Cref is available for Niji and regular MJ models, and can also be used in combination with -sref<\/section>\n<\/li>\n<\/ul>\n<p><strong>Advanced Features:<\/strong><\/p>\n<ul class=\"list-paddingleft-1\">\n<li>\n<section>You can use multiple URLs to mix information\/roles from multiple images, like this --cref URL1 URL2 (similar to multiple images or style tips)<\/section>\n<\/li>\n<\/ul>\n<section><strong>How does it work on the web test version?<\/strong><\/section>\n<ul class=\"list-paddingleft-1\">\n<li>\n<section>Drag and drop or paste the image into the imagine column and it now has three icons. Selecting these icons will determine whether it is an image tip, style reference or character reference. Hold down the Shift key and add the option to use an image in more than one category<\/section>\n<\/li>\n<\/ul>\n<section>This message on the above official website is still very informative, it doesn't matter if you didn't understand it.<strong>I'll explain it to you part by part in conjunction with the operation below.<\/strong><\/section>\n<section data-role=\"title\" data-tools=\"135\u7f16\u8f91\u5668\" data-id=\"104802\">\n<section>\n<section>\n<section data-brushtype=\"text\"><\/section>\n<section><strong>New Feature - \"Role Reference\"<\/strong><\/section>\n<\/section>\n<\/section>\n<\/section>\n<p>This time midjourney has been updated with a new feature for testing<strong>\"Character Reference\"<\/strong>You can also understand that<strong>\"Role consistency\"<\/strong>.<\/p>\n<p>It's similar to the \"Style Consistency\" update from a year ago, but<strong>This \"character consistency\" focuses on the facial features of the reference character rather than on \"stylistic characteristics\".<\/strong><\/p>\n<section data-tools=\"135\u7f16\u8f91\u5668\" data-id=\"130819\">\n<section>\n<section>\n<section data-width=\"48%\">\n<section>\n<section><strong>01<\/strong><\/section>\n<\/section>\n<section data-width=\"100%\"><img loading=\"lazy\" decoding=\"async\" class=\"alignnone size-full wp-image-5462\" title=\"640-40\" src=\"https:\/\/www.1ai.net\/wp-content\/uploads\/2024\/03\/640-40.png\" alt=\"640-40\" width=\"928\" height=\"1232\" \/><\/section>\n<section>\n<section>\n<section data-autoskip=\"1\"><strong>Original \"Reference\" image<\/strong><\/p>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n<section data-width=\"48%\">\n<section>\n<section><strong>0<\/strong><strong title=\"\" data-original-title=\"\">2<\/strong><\/section>\n<\/section>\n<section data-width=\"100%\"><img loading=\"lazy\" decoding=\"async\" class=\"alignnone size-full wp-image-5461\" title=\"640-39\" src=\"https:\/\/www.1ai.net\/wp-content\/uploads\/2024\/03\/640-39.png\" alt=\"640-39\" width=\"928\" height=\"1232\" \/><\/section>\n<section>\n<section>\n<section data-autoskip=\"1\"><strong>Using \"Character Reference\" to create diagrams<\/strong><\/p>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n<p>Above is the effect of my simple try, the left is the reference picture of the prompt word straight out, the right picture is using the<strong>Cue + \"Role Reference\" parameter<\/strong>Out of the picture.<\/p>\n<p>As you can see, the results are still good.<strong>Not just character facial features, but also clothing features are well done.<\/strong><\/p>\n<p>If you want to make a consistent picture of the characters before, you have to use the face changer robot, but then, the angle of the head down does not miss the whole face if the face changer robot is also unable to help, so this function is still very good and powerful!<\/p>\n<section data-role=\"title\" data-tools=\"135\u7f16\u8f91\u5668\" data-id=\"104802\">\n<section>\n<section>\n<section><strong>How does \"role referencing\" work?<\/strong><\/section>\n<\/section>\n<\/section>\n<\/section>\n<p>This part of the official message said a total of 4 points, I now combined with the operation of a little by little to explain to you.<\/p>\n<p><strong>1 Enter -cref URL after hint, which points to an image of a character<\/strong><\/p>\n<p>This is well understood, and is similar to the style consistency function previously described are<strong>Phrasing + Parameters (-cref+URL for reference diagrams)<\/strong><\/p>\n<p>Let's take the two pictures above as an example, the original picture I came right out with the cue word:<\/p>\n<p><img loading=\"lazy\" decoding=\"async\" class=\"alignnone size-full wp-image-5460\" title=\"640-38\" src=\"https:\/\/www.1ai.net\/wp-content\/uploads\/2024\/03\/640-38.png\" alt=\"640-38\" width=\"1080\" height=\"330\" \/><\/p>\n<p>Then I took this picture as a reference picture for character reference, but I wanted to make a picture of this girl in a red dress in a church, and I simply wrote a prompt:<\/p>\n<p><strong><em>A girl in a red dress, in church, head-on<\/em><\/strong>and then get the URL of the original image above.<strong>After the hint insert -cref URL<\/strong>, like the following:<\/p>\n<p><img loading=\"lazy\" decoding=\"async\" class=\"alignnone size-full wp-image-5463\" title=\"640-41\" src=\"https:\/\/www.1ai.net\/wp-content\/uploads\/2024\/03\/640-41.png\" alt=\"640-41\" width=\"903\" height=\"491\" \/><\/p>\n<p>Well, here's where we find a problem.<strong>I clearly put a red dress in the prompt, but the picture he gave me was still a light green dress similar to the original picture<\/strong>What's going on here?<\/p>\n<p>Don't worry, this is going to be about the last three points, in fact, the last three points can be said together:<\/p>\n<p><strong>2 you can use --cw to modify the reference strength from 100 to zero. strength 100 (-cw 100) is the default value, using facial, hair and clothing, strength 0 (-cw 0) it focuses only on the face (suitable for changing clothing\/ hair, etc.)<\/strong><\/p>\n<p>that's a good idea, so the picture that just came out was either a red dress or a red dress because i didn't add-cw, so he agreed to the-cw 100, so he looked at the face, hair and clothes, and if i wanted to change to a red dress, i would add-cw zero to the hint, as follows:<\/p>\n<p><img loading=\"lazy\" decoding=\"async\" class=\"alignnone size-full wp-image-5464\" title=\"640-42\" src=\"https:\/\/www.1ai.net\/wp-content\/uploads\/2024\/03\/640-42.png\" alt=\"640-42\" width=\"919\" height=\"490\" \/><\/p>\n<p>You see this time the dress becomes a red dress, so<strong>if you want to keep a face to change your hair or your clothes, don't forget this<\/strong>.<\/p>\n<section data-role=\"title\" data-tools=\"135\u7f16\u8f91\u5668\" data-id=\"104802\">\n<section>\n<section>\n<section><strong>Some notes on \"character referencing\"<\/strong><\/section>\n<\/section>\n<\/section>\n<\/section>\n<p>Just read the official MJ message above for that part, I'll just highlight 2 points:<\/p>\n<p><strong>\u2460 This feature works best if your reference image is generated by MJ, but if you use your own photos or non-human characters, it may not come out as well, and distortions may occur.<\/strong><\/p>\n<p><strong>\u2461 This feature works for both Niji and MJ models, but its accuracy is limited at the moment, it can't accurately replicate such details as dimples\/freckles\/ or t-shirt logos etc. of the reference image.<\/strong><\/p>\n<section data-role=\"title\" data-tools=\"135\u7f16\u8f91\u5668\" data-id=\"104802\">\n<section>\n<section>\n<section><strong>\"Advanced features of Role Reference<\/strong><\/section>\n<\/section>\n<\/section>\n<\/section>\n<p>That's the part where<strong>You can use multiple URLs to mix information\/roles from multiple images, like this --cref URL1 URL2 (similar to multiple images or style tips)<\/strong>.<\/p>\n<section><\/section>\n<section>Let me give you an example, the 2 pictures below, haha, isn't that a big contrast, I wanted to blend their features to come up with a picture to see what it looks like.<\/section>\n<section data-tools=\"135\u7f16\u8f91\u5668\" data-id=\"130819\">\n<section>\n<section>\n<section data-width=\"48%\">\n<section>\n<section><strong>01<\/strong><\/section>\n<\/section>\n<section data-width=\"100%\"><img loading=\"lazy\" decoding=\"async\" class=\"alignnone size-full wp-image-5462\" title=\"640-40\" src=\"https:\/\/www.1ai.net\/wp-content\/uploads\/2024\/03\/640-40.png\" alt=\"640-40\" width=\"928\" height=\"1232\" \/><\/section>\n<section>\n<section>\n<section data-autoskip=\"1\"><strong>Refer to Figure 1<\/strong><\/p>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n<section data-width=\"48%\">\n<section>\n<section><strong>0<\/strong><strong title=\"\" data-original-title=\"\">2<\/strong><\/section>\n<\/section>\n<section data-width=\"100%\"><img loading=\"lazy\" decoding=\"async\" class=\"alignnone size-full wp-image-5465\" title=\"640-43\" src=\"https:\/\/www.1ai.net\/wp-content\/uploads\/2024\/03\/640-43.png\" alt=\"640-43\" width=\"928\" height=\"1232\" \/><\/section>\n<section>\n<section>\n<section data-autoskip=\"1\"><strong>Refer to Figure 2<\/strong><\/p>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n<section>Then I need to use.<strong>+-cref Figure 1URL Figure 2URL<\/strong>, like the following:<\/section>\n<p><img loading=\"lazy\" decoding=\"async\" class=\"alignnone size-full wp-image-5466\" title=\"640-44\" src=\"https:\/\/www.1ai.net\/wp-content\/uploads\/2024\/03\/640-44.png\" alt=\"640-44\" width=\"1066\" height=\"490\" \/><\/p>\n<section>Then we look at the outtake effect, which indeed blends the character features of reference image 1 and reference image 2, and the result is still good.<\/section>\n<section><\/section>\n<section data-role=\"title\" data-tools=\"135\u7f16\u8f91\u5668\" data-id=\"104802\">\n<section>\n<section>\n<section><strong>How \"Role Reference\" works in the web beta<\/strong><\/section>\n<\/section>\n<\/section>\n<\/section>\n<p>This \"Character Reference\" function is also available in the web version of the current test.<strong>In the web version we have to click on the plus sign next to the prompt word input box to expand it and then upload our reference image<\/strong>, I labeled the steps in the screenshot below:<\/p>\n<p><img loading=\"lazy\" decoding=\"async\" class=\"alignnone size-full wp-image-5468\" title=\"640-46\" src=\"https:\/\/www.1ai.net\/wp-content\/uploads\/2024\/03\/640-46.png\" alt=\"640-46\" width=\"1080\" height=\"387\" \/><\/p>\n<p>Then we see that there are three small icons on this thumbnail that we can choose from, and these three icons are tabulated:<\/p>\n<p><strong>The first little man icon indicates the \"character reference\" we're talking about this time.<br \/>\n<\/strong><\/p>\n<p><strong>The second pin icon indicates a \"style reference\".<\/strong><\/p>\n<p><strong>The third picture icon indicates \"picture reference\".<\/strong><\/p>\n<p><img loading=\"lazy\" decoding=\"async\" class=\"alignnone size-full wp-image-5467\" title=\"640-45\" src=\"https:\/\/www.1ai.net\/wp-content\/uploads\/2024\/03\/640-45.png\" alt=\"640-45\" width=\"347\" height=\"152\" \/><\/p>\n<p>So let's choose the first one and then just type in the prompt word then enter and wait for the picture to come out. It feels like the web version is a little bit easier to use hahaha.<\/p>\n<section data-role=\"splitline\" data-tools=\"135\u7f16\u8f91\u5668\" data-id=\"127772\">\n<section>\n<section data-width=\"100%\"><\/section>\n<\/section>\n<\/section>\n<p>Well, that's all for this time.<\/p>\n<p>The \"Character Consistency\" feature of this update is really good, and it's the best of the best.<strong>I'm sure this feature will be the most used feature in MJ in the future, bar none!<\/strong><\/p>","protected":false},"excerpt":{"rendered":"<p>Midjourney officially released a message: I'm going to translate the MJ official message from the above screenshot: Hey, we're testing a new character reference function today, which is similar to the style reference function, except that it's not a reference style, but an attempt to match the role reference image. How it works: Enter -cref URL after the hint, which points to an image of a character, and you can use -cw to modify the reference \"strength\" from 100 to 0 (\u2013cw 100) by default, using facial, hair and clothing 0 (\u2013cw 0) which focuses only on the face (suitable for changing clothing\/ hair, etc.)<\/p>","protected":false},"author":1,"featured_media":0,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[149,144],"tags":[186,1664],"collection":[260],"class_list":{"0":"post-5458","1":"post","2":"type-post","3":"status-publish","4":"format-standard","5":"hentry","6":"category-jiaocheng","7":"category-baike","8":"tag-midjourney","10":"collection-midjourney"},"acf":[],"_links":{"self":[{"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/posts\/5458","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/comments?post=5458"}],"version-history":[{"count":0,"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/posts\/5458\/revisions"}],"wp:attachment":[{"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/media?parent=5458"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/categories?post=5458"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/tags?post=5458"},{"taxonomy":"collection","embeddable":true,"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/collection?post=5458"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}