{"id":15265,"date":"2024-07-10T10:37:24","date_gmt":"2024-07-10T02:37:24","guid":{"rendered":"https:\/\/www.1ai.net\/?p=15265"},"modified":"2024-07-10T10:37:24","modified_gmt":"2024-07-10T02:37:24","slug":"midjourney%e6%95%99%e7%a8%8b%ef%bc%8c%e5%a6%82%e4%bd%95%e7%94%a8midjourney%e7%94%9f%e6%88%90%e4%b8%80%e8%87%b4%e6%80%a7%e7%9a%84%e4%ba%ba%e7%89%a9%e8%a7%92%e8%89%b2%ef%bc%9f","status":"publish","type":"post","link":"https:\/\/www.1ai.net\/en\/15265.html","title":{"rendered":"Midjourney tutorial, how to use Midjourney to generate consistent characters?"},"content":{"rendered":"

Midjourney<\/a>It\u2019s been a long time coming for the release of a much-anticipated new feature called \u201cCharacter Reference.\u201d It allows you to specify reference images that define the character you want to keep consistent across your images.<\/p>\n

Character Reference works great for duplicating a single character, but what about multiple characters? \ud83c\udf8e<\/p>\n

In this article, I\u2019ll share 3 advanced techniques to master character consistency, including generating multiple consistent characters like the image below.<\/p>\n

Today's content will be divided into three parts. First, a brief description of the new features<\/p>\n

Part 1: What is a "Character Reference"<\/strong><\/p>\n

Part 2: How to use the Character Reference<\/strong><\/p>\n

Part 3: 3 Advanced Tips for Using Character Reference<\/strong><\/p>\n

Part 1: What is a "Character Reference" \ud83c\udfa8<\/h1>\n

"Actor References" allow us to create characters based on existing character images provided via URLs.<\/p>\n

It aims to make the character match the reference image, allowing us to create consistent characters across different images and in different styles.<\/p>\n

Highlight Features<\/h1>\n

\u2022\u2692\ufe0f Functionality: This function is similar to the regular image prompt, but it is "focused" on character features<\/p>\n

\u2022 Use: enter \u2013cref URL after hint and add URL for role images<\/p>\n

\u2022 Format: \/imagine [text tips]\u2014cref URL<\/p>\n

\u2022\ud83d\uddbc\ufe0f Compatibility: For Midjourney V6 and Niji 6<\/p>\n

modify with -cw<\/strong><\/p>\n

you can use --cw to modify the reference strength from 0 to 100<\/p>\n

\u2013cw 100 (default) uses facial, hair and clothing<\/p>\n

\u2013cw 0 focuses on facial<\/p>\n

Part 2: How to use the Character Reference<\/h1>\n

Step 1: Choose your reference image<\/strong><\/p>\n

Use an image you generated with Midjourney or a local file to select one.<\/p>\n

\"get-339\"<\/div>\n

Step 2: Upload the image copy the URL<\/strong><\/p>\n

Upload an image. Copy the image URL.<\/p>\n

\"get-340\"<\/div>\n
\"get-341\"<\/div>\n

The following is the introduction of the "mat map" expansion<\/p>\n

\"get-342\"<\/div>\n
\"\"<\/div>\n
\"get-343\"<\/div>\n

Step 3: Enter the prompt<\/strong><\/p>\n

Enter\/imagine as usual. Enter your hint, describe the image you want to generate and add -cref URL<\/p>\n

\"get-344\"<\/div>\n

A digital papering of a girl-cref URL<\/em><\/p>\n

Digital painting of a girl \u2013cref URL<\/em><\/p>\n

\"get-347\"<\/div>\n

A photo taken in blue hours in the street-cliff URL<\/em><\/p>\n

Photo of Street Blue Time \u2013cref URL<\/em><\/p>\n

\"get-345\"<\/div>\n

Part 3: 3 Advanced Techniques for Mastering Character Consistency<\/h1>\n

You can do more customization via parameters and hint weights. In this section, I will describe some techniques you can use to adjust the style of the generated images:<\/p>\n

skills 1: use a \u2013cw change style<\/h1>\n

Tip 2: Use Pan to create multiple consistent characters<\/h1>\n

Tip 3: Create multiple consistent characters in different styles<\/h1>\n

advanced skills i: use \u2013cw change style<\/h1>\n

you can convert a virtual role to a photo, and vice versa. you can use \u2013cw to modify the reference strength from 100 to 0\u3002<\/p>\n

First, let's look at how the weights affect the results.<\/p>\n

here's the role of-cw:<\/p>\n

    \n
  • \u2013cw 100 is the default value, using facial, hair and clothing<\/li>\n
  • \u2013cw 99 to \u2013cw 1 will start to reduce the pick-up of the entire role<\/li>\n
  • \u2013cw 0 when strength is 0<\/li>\n<\/ul>\n
    \"get-346\"<\/div>\n

    use-cw0 to change style<\/p>\n

    using \u2013cw0, the system will focus on facial, which is good for changing clothing\/hairs, etc. this is similar to \"face exchange\". it applies to a complete change of art style\u3002<\/p>\n

    Just add the style to your prompt<\/p>\n

    Cyberpunk style \u2013cref URL \u2013cw 0 Cyberpunk style \u2013cref URL \u2013cw 0<\/p>\n

    \"get-348\"<\/div>\n

    The number of girls drinking coffee -cref URL -cw 0
    \nDigital patiing of the girl drining note-cref URL-cw 0<\/p>\n

    \"get-351\"<\/div>\n

    -cref URL \u2013cw 0A girl reading in the library \u2013cw 0<\/p>\n

    \"get-353\"<\/div>\n

    -cref URL \u2013cw 0A girl standing behind the neon light \u2013cw 0<\/p>\n

    \"get-352\"<\/div>\n

    use \u2013cw 0, the system gives priority to the face, making it easier for you to adjust your hair, clothing and overall art style\u3002<\/p>\n

    \u2b50\ufe0f Advanced Tip 2: Use Pan to create multiple consistent characters<\/h1>\n

    The best way to put multiple \"-cref\" characters into images is to use \"PAN\" - a function to expand the image canvas in the selected direction<\/p>\n

    Let\u2019s get started<\/p>\n

    Step 1: Turn on high variance mode<\/strong><\/p>\n

    First, use \/settings to turn on "Remix Mode" and select "High Variation Mode".<\/p>\n

    \"get-349\"<\/div>\n

    Confirm that the two buttons in the blue box are green, indicating that they are turned on<\/p>\n

    \ud83d\udccc This allows you to change the prompt while panning.<\/p>\n

    Step 2: Choose an image of a character<\/strong><\/p>\n

    Find the image and use the "U" button to zoom in on the selected image<\/p>\n

    \"get-350\"<\/div>\n

    Step 3: Use Pan<\/strong><\/p>\n

    Use "Pan" and choose the direction the new character should go \u27a1\ufe0f\u2b05\ufe0f\u2b06\ufe0f\u2b07\ufe0f<\/p>\n

    \"get-354\"<\/div>\n

    Step 4: Add a second character by changing the prompt<\/strong><\/p>\n

    Describe the details of how you want the second character to appear in the image<\/p>\n

    \"get-355\"<\/div>\n

    result<\/strong><\/p>\n

    This will stitch together a completely new image containing the first character with the new one.<\/p>\n

    Use the same --cref URL to repeat the same steps. You can constantly create different scenarios\u3002<\/p>\n

    \"get-357\"<\/div>\n

    \ud83d\udccc Tips:<\/p>\n

    Sometimes the image may look a little awkward, try:<\/p>\n

    \u2022 If you are looking for more precise image generation, add more hint word details<\/p>\n

    \u2022 Use Vary Region to make adjustments<\/p>\n

    \u2b50\ufe0f Advanced Tip 3: Create Multiple Consistent Characters in Different Styles<\/h1>\n

    This is a more advanced step that can take your creation to the next level!<\/p>\n

    We were able to change the overall style of both characters while still maintaining consistency.<\/p>\n

    Let's create a new style for each character<\/p>\n

    step 1: change style using \u2013cw0 (see technique 1)<\/strong><\/p>\n

    remember what we did with the --cw 0. we can change the style of each character and use the --cw 0 to keep the face<\/p>\n

    \"get-356\"<\/div>\n

    Original Image<\/p>\n

    Tip: Sabpunk style \u2013cref URL \u2013cw 0<\/strong><\/p>\n

    In cyberpunk style <\/strong>\u2013cref URL -cw 0<\/strong><\/em><\/p>\n

    \"get-358\"<\/div>\n

    Cyberpunk style<\/p>\n

    Step 2: Save the New URL<\/strong><\/p>\n

    Save new URLs for male roles to be used with \u2013cref<\/p>\n

    Step 3: Translate the image of the girl character<\/strong><\/p>\n

    Magnify images of the role of girls. Moves the step of the image and enters a hint with a new \u2013cref URL\u3002<\/p>\n

    You can also use Zoom and Vary Region to add more settings and dynamics<\/p>\n

    \"get-359\"<\/div>\n
    \"get-360\"<\/div>\n

    \"-cw 0\" only retains facial features<\/p>\n

    More "Character Reference" Notes<\/h1>\n

    \u2022\u2705 This feature works best when using characters generated from Midjourney images. It will not work with real people\/photos (might distort them like regular image prompts)<\/p>\n

    \u2022\u2705 Cref works similar to regular image hints, but it is "focused" on the visual features of the character<\/p>\n

    \u2022\u2705 Character attributes work well with large attributes (e.g. curly hair, pink sunglasses)<\/p>\n

    \u2022\u274c It will not replicate small details like exact dimples, freckles\/or t-shirt logos.<\/p>\n

     <\/p>","protected":false},"excerpt":{"rendered":"

    It's been a long time since Midjourney released its much-anticipated new feature called \"Character References\". It allows you to specify a reference image that defines the character you want to be consistent in your image. \"Character Reference\" works great for replicating a single character, but what about multiple characters? \ud83c\udf8e In this article, I'll share 3 advanced skills to master character consistency, including generating multiple consistent characters as shown in the image below. Today's content will be divided into three parts, starting with a brief description of the new feature Part 1: What is Role Reference Part 2: How to Use Role Reference Part 3: 3 Advanced Skills for Using Role Reference \"Part 1: What are Role References \ud83c\udfa8 Role References allow us to create roles based on existing role images provided through URLs.<\/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":[491,186,1664],"collection":[260],"class_list":{"0":"post-15265","1":"post","2":"type-post","3":"status-publish","4":"format-standard","5":"hentry","6":"category-jiaocheng","7":"category-baike","8":"tag-ai","9":"tag-midjourney","11":"collection-midjourney"},"acf":[],"_links":{"self":[{"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/posts\/15265","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=15265"}],"version-history":[{"count":0,"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/posts\/15265\/revisions"}],"wp:attachment":[{"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/media?parent=15265"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/categories?post=15265"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/tags?post=15265"},{"taxonomy":"collection","embeddable":true,"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/collection?post=15265"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}