Get the Same Character Every Time in Midjourney: Use the New CF and CW

midjourney blog post image

Why Consistent Characters Matter

Imagine creating a series of images featuring the same person or animated character. Maybe they are in different locations or wearing different clothes. Without a way to reference the original look, your character might change wildly from one image to the next. That’s where consistency tools come in handy.

Introducing –cref: Character Reference

The main tool for keeping a character the same is the –cref parameter. This tells Midjourney to look at an image you provide and use its character features as a base for generating new images.

Using –cref with an Uploaded Image

You can start with any image, even one you didn’t create in Midjourney. Here’s how:

  • Upload the image (e.g., a photo of a person) to Discord in your Midjourney thread or server. Make sure it’s saved in Discord.

  • Upload the image (e.g., a photo of a person) to Discord in your Midjourney thread or server. Make sure it’s saved in Discord.

  • Click on the image to open it fully, then right-click (or long-press on mobile) and select “Copy Image Address” or “Copy Media Link”.

  • Click on the image to open it fully, then right-click (or long-press on mobile) and select “Copy Image Address” or “Copy Media Link”.

  • Start your Midjourney prompt with `/imagine`.

  • Start your Midjourney prompt with `/imagine`.

  • Paste the image link you copied. This goes first.

  • Paste the image link you copied. This goes first.

  • Add your text prompt describing the scene, action, or clothing (e.g., “wearing a dark suit”).

  • Add your text prompt describing the scene, action, or clothing (e.g., “wearing a dark suit”).

  • Add `–cref` followed by a space.

  • Add `–cref` followed by a space.

  • Paste the *same* image link again immediately after `–cref` (no space needed between `–cref` and the link).

  • Paste the *same* image link again immediately after `–cref` (no space needed between `–cref` and the link).

  • Your full prompt will look something like: `/imagine [image link] wearing a dark suit –cref [image link]`

  • Your full prompt will look something like: `/imagine [image link] wearing a dark suit –cref [image link]`

    Midjourney will then generate images attempting to match the character from your link while following your text prompt.

    Using –cref with a Midjourney-Generated Character

    What if you created a character you like using Midjourney and want to use *that* character consistently? The process is similar:

  • Generate your initial character using a prompt like `/imagine English bulldog in a Pixar style`.

  • Generate your initial character using a prompt like `/imagine English bulldog in a Pixar style`.

  • Find the image you like the best and upscale it (U button). This makes it a standalone image Midjourney can reference clearly.

  • Find the image you like the best and upscale it (U button). This makes it a standalone image Midjourney can reference clearly.

  • Click on the upscaled image to open it fully.

  • Click on the upscaled image to open it fully.

  • Right-click on the image and select “Copy Image Address” or “Copy Media Link”.

  • Right-click on the image and select “Copy Image Address” or “Copy Media Link”.

  • Start a new prompt: `/imagine`

  • Start a new prompt: `/imagine`

  • Paste the image link you copied (the link to the upscaled Midjourney image).

  • Paste the image link you copied (the link to the upscaled Midjourney image).

  • Add your new text prompt (e.g., “reading a book on the beach”).

  • Add your new text prompt (e.g., “reading a book on the beach”).

  • Add `–cref` followed by a space.

  • Add `–cref` followed by a space.

  • Paste the *same* image link again (the link to the upscaled Midjourney image).

  • Paste the *same* image link again (the link to the upscaled Midjourney image).

  • Your prompt will look like: `/imagine [upscaled MJ image link] reading a book on the beach –cref [upscaled MJ image link]`

  • Your prompt will look like: `/imagine [upscaled MJ image link] reading a book on the beach –cref [upscaled MJ image link]`

    Midjourney will now generate your character in the new scene, aiming for consistency based on the reference image.

    Automate your consistent character workflows! Explore the

    Controlling Character Details with –cw

    Sometimes you want to keep the character’s face and body shape the same, but you want to change their clothing, styling, or the setting’s influence on their look. This is where the –cw parameter comes in. CW stands for “Character Weight.”

  • `–cw 100` (Default): This is the standard setting if you don’t specify –cw. It puts a lot of weight on keeping the *entire* character consistent, including clothing, hairstyle, and general appearance details from the reference image.

  • `–cw 100` (Default): This is the standard setting if you don’t specify –cw. It puts a lot of weight on keeping the *entire* character consistent, including clothing, hairstyle, and general appearance details from the reference image.

  • `–cw 0`: This setting focuses mainly on the character’s face and body proportions. It gives Midjourney more freedom to change clothing, accessories, or express different emotions influenced by the prompt, while keeping the core character recognizable.

  • `–cw 0`: This setting focuses mainly on the character’s face and body proportions. It gives Midjourney more freedom to change clothing, accessories, or express different emotions influenced by the prompt, while keeping the core character recognizable.

  • You can use values between 0 and 100 for a balance.

  • You can use values between 0 and 100 for a balance.

    Example: Changing Clothing with –cw 0

    Let’s say you have a reference image of a character wearing a specific jersey, but you want to generate them in a different colored jersey. Here’s how:

  • Get the image link for your character (either uploaded or Midjourney-generated).

  • Get the image link for your character (either uploaded or Midjourney-generated).

  • Start your prompt: `/imagine`

  • Start your prompt: `/imagine`

  • Paste the image link.

  • Paste the image link.

  • Add your text prompt describing the new clothing (e.g., “wearing a pink and blue jersey”).

  • Add your text prompt describing the new clothing (e.g., “wearing a pink and blue jersey”).

  • Add `–cref` followed by a space.

  • Add `–cref` followed by a space.

  • Paste the *same* image link again.

  • Paste the *same* image link again.

  • Add `–cw 0` at the end.

  • Add `–cw 0` at the end.

  • Your prompt is: `/imagine [image link] wearing a pink and blue jersey –cref [image link] –cw 0`

  • Your prompt is: `/imagine [image link] wearing a pink and blue jersey –cref [image link] –cw 0`

    This tells Midjourney: “Use this character as a guide, but only keep the face and structure (cw 0). Feel free to change the clothing based on the text prompt.”

    Repetitive Midjourney tasks got you down? Consider the

    Putting It All Together

    By combining –cref and –cw, you gain much better control over character consistency in Midjourney. Use –cref to point Midjourney to who the character is, and use –cw to adjust how strictly it follows the reference image’s style and clothing versus changing them based on your prompt.

    Final Thoughts

    These new tools, –cref and –cw, are powerful additions for creators who need to generate related images featuring the same character. Practice using them with different reference images and text prompts to see how creative you can get while keeping your characters recognizable.

    Looking to scale your Midjourney creations or manage complex projects with consistent elements? The

    🖥

    Do This Locally with Logen

    Skip the subscription. Generate images like these on your own GPU — private, unlimited, no cloud dependency.

    Try Logen Free →

    Midjourney vs Logen

    FeatureMidjourneyLogen
    Cost$10-60/moFree
    PrivacyCloud (images stored)100% local
    GenerationsLimited by planUnlimited
    Internet RequiredAlwaysNo
    Custom ModelsNoYes (any ComfyUI model)
    GPU RequiredNo (cloud)Yes (local)
    Ease of UseDiscord botDesktop app
    Try Logen Free →
    Scroll to Top