Are you wondering how to outperform the competition with your online store and convince internet users to buy from you? We recommend you focus on creating personalised and engaging content in product images.

Are you wondering how to outperform the competition with your online store and convince internet users to buy from you? We recommend you focus on creating personalized and engaging content in product images. To create them, it is best to use artificial intelligence nowadays. Using it will make it easier to engage customers and create shopping experiences that are not only interactive and varied, but also tailored to each individual’s preferences.

Where AI makes it easier for online stores to edit product images

1. Generating background

  • This is the simplest use of generative image AI.
  • Using simple prompts, you can generate a variety of backgrounds that you can immediately use in your graphic creations.
  • Before generating a new background, we recommend removing the original one first.
  • Also, check its output quality. Midjourney can currently generate it up to 4096 x 4096 px.
Source: Midjourney

2. Label and sticker generation

  • Very similar to the previous case. You can generate all kinds of label or sticker versions.
  • Note that the available generative AIs can’t yet completely generate specific text in the image. Currently, it can handle 4 — 5 specific letters. If there is more text, there is a high chance that it will be unintelligible.
  • AI tools cannot yet generate objects on transparent backgrounds. Therefore, for generated labels or stickers, you still need to remove the background to make them as usable as possible.
  • Also, keep in mind that the output is a .jpeg format and not a vector graphic. Therefore, with different sizes of creatives and stickers, these objects may be blurry.
Source: Midjourney

3. Creating whole creative with input image

  • The two previous cases show how to generate a picture output using text input (prompts). Many tools allow you to use a combination of a prompt and an input image as input.
  • This is ideal if you have a product image for which you want to generate a custom graphic template.
  • There can be a certain problem that the AI modifies the input image and the resulting creative is not the same product (but only very similar). This is unacceptable in e‑commerce because advertising platforms require that the image must have a specific product. So beware of what some AI tools promise you.
Source: Midjourney

4. Dress the product as a model

  • If we have an input image from the fashion segment, we can “dress” it up as a virtual model. Data analysis shows that images that depict models are more converting.
  • Generating a model with AI is currently quite simple. You can even simply select their age, appearance and many other attributes.
  • However, you have to take into account that the input image with the product will be altered as a result. It will look very similar in type, but it will not be a specific product. Therefore, such images are not yet usable for advertising, but this is subject to change.
Source: Internal source of MERGADO and Midjourney

5. Generating image from description

  • You can use this option when you don’t have a product image but you have a product description.
  • However, even here you will find that generative AI cannot yet generate a specific product image.
Source: Midjourney

6. Generating description from image

  • You can also use AI in the opposite way to the previous case, i.e. if you have an image and want to generate a description of it.
  • This way you get a textual description of the product, which you can then process as an intermediate product in ChatGPT.
  • Beware that the generated text may not be factually accurate. Therefore, check it before using it.
  • You can use the finished text, for example, as a product description, to create advertisements or keywords, etc.
Source: Midjourney

7. Outpainting

  • Artificial intelligence is also a great help in the case of images that are taken in the context of the environment (the product is a couch, for example, and the photo is set in the entire living room) either in portrait or landscape and you want to flip them into a square format.
  • To avoid white bars in the edited image, outpainting will “illustrate” the missing scene. The AI doesn’t even disturb the main product by doing this, it just guesses the rest of the scene. Therefore, it’s okay to use such an image in an advertisement.
Source: Internal source of MERGADO

8. Increase resolution of small images

  • A few clicks will help you adjust the resolution of small, pixelated product images.
  • By getting them into the form that advertising platforms require, you can advertise them well.
  • This feature works automatically and in bulk on all images in the template you create.
Source: Feed Image Editor

9. Background removal

  • You solve one of the standard requirements of advertising channels for product images — a white background without distracting elements.
  • Use this feature to easily replace unacceptable background colors with white. Edit product photos embedded in the environment or remove graphic or text elements.
Source: Feed Image Editor

10. Removing watermarks

  • No more manually removing watermarks image by image. This feature will in bulk and automatically handle their removal for all your product photos.
  • Advertising platforms will no longer reject your images in ads for this reason.
  • Just tag an identical watermark and the AI will take care of the rest, finding it in the rest of your images and removing it.
Source: Feed Image Editor

Don’t limit yourself to just transactions when trading online. Offer your customers a personalised, engaging and interactive experience. Innovative solutions using AI can help you do just that.

Read more:

Veronika Kusalíková

A copywriter and content specialist Veronika is in charge of the English content and communication. She spreads the word about Mergado and e‑commerce on our blog and social media. After she shuts her computer down, you can catch her knitting a sweater, strolling around second-hand shops or traveling.