In just few clicks, you can generate on-model photos of your clothes from only a single flat-lay photo, for almost all clothing categories!

In this guide, we will explain how the generative AI works on Uwear.ai and how you can get the most of it, from the photos you input to the best prompts to use.

Our goal is to build a tool that uses AI to generate diverse photoshoot from one flat may photo. At the core of our generative engine is a text to image model like Stable Diffusion, MidJourney or Dall-e. So any prompts you find elsewhere should also work on Uwear.ai, albeit some tweaking that we tackle later on in the guide. Let’s get into it!

1. Prepare and upload your clothing items

In this section, we will go over each of the required inputs when uploading a clothing item on uwear.ai.

1.1 What input photos should I use?

Uwear.ai is currently optimized to take flat lay photos as an input. Anything besides that is not optimal, although it might still work!

• Flat-lays: This is the best type of photos you can use. By using a flat lay photo, as opposed to a on-mannequin or ghost mannequin photo, you do not bias the AI into generating a particular body shape, which will give you better control.

  

  Input Photo

  
  

  

  Generated photos with two different body shapes
  
  
  

• Ghost mannequin: Uwear.ai also works with ghost mannequins and will achieve almost the same results as with a pure flat lay photo. The only inconvenience is that the AI will try to follow the body shape of the ghost when generating images. As you can see in the example below, the AI has a hard time diverting away from the breast form.

  

  Input photo
  
  

  

  Generated photos on two different AI women models
  
  
  

• On-model or on mannequin: On-model and on-mannequin images of clothes will work the same way as ghost mannequin. The only requirement is that you have to crop your image to the specific clothing your want to upload, otherwise, the AI will consider all clothes in the image as a single input.

  

  Croping the input photo to the clothing of interest
  
  

  

  Generated photos of two AI men wearing the same green shorts
  
  
  

1.2 What other labels should I use when uploading my clothes?

EDIT: since Sept. 19th, we do not require adding a gender, category or subcategory of the clothing.

When uploading your items to Uwear.ai, you will be asked to define some parameters to help the AI understands what visuals you are looking to generate:

• Gender: This label defines the gender of the models you would like to see wearing your clothes. If your item is unisex, and you would like photos of men and women wearing the same item, simply upload your items twice with the different target genders.
  
  

  

  A woman and a man wearing the same linen shirt
  
  

• Description of the item: Your clothing description is what helps the AI understand the main features of the item as well as things that are not visible on a flat-lay. You should try to at least include the category, the color, the fit and the length.
  
  

• Category and Sub-category: Picking a category contributes to framing the camera zoom and adjusting to your item. For instance, a human model will pose differently to showcase a t-shirt as opposed to a blouse. This means that it is not crucial to perfectly categorize your item as long as you pick something close enough. For example, categorizing knee-length bath robes as “midi dresses” will work great!
  
  
  

2. Describe your AI clothing photos

Now that our item is ready let's start generating some photos. In this process, the prompt is the most powerful lever. A simple way to think about it: the shorter the prompt, the more freedom you give to the AI which increases the overall quality. The longer the prompt, the more constraints you impose on the generation. Which will make the AI struggle sometimes but when successful your photos will have a unique feeling! So what should you include?

2.1 The Prompt structure

We like to think of the prompt as having three main components:

• The Human Model: you should always start by mentioning the gender of your model, otherwise the AI might generate images with the wrong gender. Optionally, you can mention their ethnicity, as well as any particular feature that is important for your shoppers such as tattoos, piercings, curly or short hair, etc.

• Accessories and other clothing: you can mention other clothing if you envision your item in a specific outfit.

• Background: finally, you can specify the context, but more on this below.
  
  

  

  
  

  

  
  
  

  
  

2.2. Tips and tricks for describing models in AI clothes photoshoots

To describe your models, you can really write whatever comes to mind, as long as it's realistic and coherent. But here’s a list of points to consider:

• Mention a specific country or city to have more variety of models: german woman

• Give your models an age: 20 years old german woman

• Describe the color and style of the hair:  20 years old german woman, bob cut black hair

• Decide on the body weight of your model: 20 years old german woman, bob cut black hair, chubby

• Pick an emotion: 20 years old german woman, bob cut black hair, chubby, glaring
  

  

  Prompt: "20 years old german woman, bob cut black hair, chubby, glaring"

  
  
  

2.3. Studio and outdoor backgrounds

And finally, don’t forget to describe the context of the photoshoot! For studio photoshoots, you will have to mention “studio” or “plain background”. If you do not mention it, the AI will sometimes place your models in a background that may have something to do with your prompt or clothing description.

Two female models wearing the same high-rise wide legs jeans

That’s it for now! You are ready to start generating on uwear.ai.

We will be updating this guide as we work on our AI models so come back from to times to check out new tricks. And as always, please feel free to write to us if you have any questions: hello@uwear.ai