Advanced Image Editing

Learn how to switch between different AI image models to find the perfect visuals for your project.

Musho uses Lummi for images by default, to make sure your designs always look stunning and polished. But, with advanced image editing, you can use Musho with other AI image models, or add doodles with Blush.

About Advanced Image Editing

In Musho, we offer you the power to take control of your AI-generated designs by choosing between our different AI image models. Whether you’re generating stunning illustrations using DALLE or selecting custom images from Lummi, these parameters empower you to fine-tune your creations.

Before we start:

A layer refers to an individual element within your design project. Layers can include shapes, text, images, or any other visual component that makes up your overall design.

A parameter, in the context of Musho, is a setting that can be added to a layer to influence how an image is generated. Parameters allow users to fine-tune various aspects of the AI-generated image, tailoring it to specific requirements or preferences.

An AI Image model refers to a specific type of artificial intelligence platform used to generate images. Each model has its unique capabilities and is trained on different data sets to produce various styles and types of images.

Image Editing in Layers

Musho allows you to choose between 3 different AI image models to remix images - Lummi, DALL·E 3 and SDXL. In addition, Musho also offers doodle options with our Blush integration.

In order to choose a different model, first you need to choose your layer, then add the specific model tag (e.g., --dalle) to the layer.

You can seamlessly switch between --sd, --lummi, --dalle and --blushto achieve varied visual results.

After you've set your model tag, it's time to remix! Open the Musho plugin and select your layers. Add a prompt to define what you want your images to contain, and let the magic happen.

To learn more about image editing in layers, check out the quick video tutorial below. 👇

Model Editing in Prompt

If you want to edit images across all of your layers, you can also specify the image generation model (—-sdxl —-lummi —-dalle —-blush ) directly in the prompt. This prompt tag overrides any image model tags set in the layer names, ensuring that the user's specific choice in the prompt takes priority for more tailored and intuitive image generation.

Image editing in prompt only works for remixes

Advanced Features for Lummi

Opting for Lummi images, as opposed to using DALLE or SDXL, offers a unique approach since these images have already been generated and curated using AI to form an extensive image library.

Read on to learn how to use Lummi's advanced features in Musho.

Select Custom Images

This feature enables you to precisely search for images by adding specific categories on top of the image source. To use Lummi, first add the parameter --lummi --category to your layer to specify it as your image source. Then, you can select a specific category by appending :category-name to the layer.

Complete List of Lummi Categories 📋👇
  • --lummi --caregory:avatars

  • --lummi --category:animals

  • --lummi --category:architecture

  • --lummi --category:art

  • --lummi --category:beauty

  • --lummi --category:cats

  • --lummi --category:christmas

  • --lummi --category:designers

  • --lummi --category:devices

  • --lummi --category:documentary

  • --lummi --category:family

  • --lummi --category:fashion

  • --lummi --category:flowers

  • --lummi --category:food

  • --lummi --category:hobbies

  • --lummi --category:editorial

  • --lummi --category:holidays

  • --lummi --category:kids

  • --lummi --category:medical

  • --lummi --category:nature

  • --lummi --category:lgbtiqaa+

  • --lummi --category:people-at-home

  • --lummi --category:people-at-work

  • --lummi --category:people-devices

  • --lummi --category:realness

  • --lummi --category:sports

  • --lummi --category:still-life

  • --lummi --category:street-photography

  • --lummi --category:stuff

  • --lummi --category:tacos

  • --lummi --category:travel

  • --lummi --category:vacation

  • --lummi --category:wallpaper

  • --lummi --category:wellness

Image Filtering with Lummi

Musho’s integration with Lummi brings an added dimension of precision to your AI-generated designs. By choosing Lummi as your image model, you can access an extensive, curated library of images and fine-tune your creations with advanced filtering options.

Using Lummi Filters

To utilize Lummi’s filtering capabilities, you can easily apply them to your designs within Musho. Here's how:

  1. Initiate with the Remix Function: Start by selecting the Remix option in Musho. This function is where you can apply Lummi filters to refine your image search.

  2. Specify Filters: Once you’ve chosen an image layer to edit, you can specify Lummi filters on your prompt to narrow down your search for the perfect image. These filters include Luminance, Orientation, Number of People, Focal Point, and Color.

  3. Combine Filters for Specific Results: Combine various filters to achieve more specific and tailored results. This powerful feature allows you to be as precise as needed in your image selection.

Detailed Guide on Lummi Filters


  • Usage: Adjusts the brightness level of images.

  • Options: dark, light, neutral

  • Example: -lummi --luminance=light


  • Usage: Defines the orientation of the image.

  • Options: square, vertical, horizontal.

  • Example: -lummi --orientation=vertical


  • Usage: Filters images by the number of people.

  • Example: -lummi --people=4


  • Usage: Filters images by specific color.

  • Example: -lummi --color=#FF0000

Focal Point (Coming Soon)

  • Usage: Filters images based on focused areas.

  • Options: tl, tc, tr, cl, c, cr, bl, bc, br.

  • Example: -lummi --focus=tr

Application in Design Projects

For instance, to find a bright, horizontally oriented image featuring food, the command might be:

--lummi --luminance=gt0.5 --orientation=horizontal --category=food

Important Usage Notes

Musho will always return an image result, even if it’s not an exact match for all filter criteria.

Last updated