Creating images with Midjourney can open a world of possibilities, but you can take your designs even further with these tips and tricks.
Upscaling
Midjourney images come out at a default resolution of 1024 x 1024 pixels in a 1:1 aspect ratio. This might be fine for some uses, but if you need a higher resolution, you'll need to upscale the image. Here are some free online tools you can use to achieve this:
- https://imgupscaler.media.io/app/upscaler
- https://www.topazlabs.com/gigapixel-ai
- https://letsenhance.io/
- https://ai.nero.com/enhance/
- https://imglarger.com/
Vectorise
Vectorising is the process of transforming your image into a vector format, made up of geometric shapes and objects. This allows the image to be scaled up infinitely without losing quality.
Keep in mind that this works best with images that have clearly defined lines and shapes, and it's not as effective for photos or highly detailed images. For most cases, you probably don’t need to think about this, but if you are looking to create assets like logos to sell, it would be beneficial for you to look into this.
Removing Background
The images you get from Midjourney are in JPG format, which means they don't come with a transparent background. If you need to overlay your image on a POD product like t-shirt, or need to combine different elements together, working with images with transparent backgrounds will make your job easier.
To have a better chance of generating images that are suitable for transparent background. Use prompts like:
- Plain background
- White background
- No background
- Vector
- Flat
- Minimal
- Sticker design
After generating the image, here are some tools you can use to remove backgrounds:
- Canva background remover (pro feature, click here for a free 30 day trial)
- https://www.remove.bg/
- https://www.photoroom.com/background-remover
- https://www.adobe.com/express/feature/image/remove-background
Removing Text & Watermark
Midjourney doesn't handle text very well and sometimes, you might want to remove text from your image. Or maybe the image has a watermark from the training data. Here's how you can remove them:
- If you have Adobe Photoshop or Affinity Photos, you should be able to select the area with the text or watermark. Then use content-aware fill or generative fill for Photoshop, or inpainting for Affinity Photos.
- Online tools like Canva’s Magic Eraser tool (if you have Pro subscription) or Photopea’s Spot Healing Brush can also be used to remove text and watermarks.
- To reduce the chance of watermark appearing, you can also regenerate the image with this added to the end of the prompt: --no text watermark letters words signature
Fixing Midjourney Mistakes
Sometimes, Midjourney can make mistakes like adding extra arms or producing a weird face. You can fix these mistakes with tools like Photoshop, but that’s a big topic and outside the scope of this guide.
To minimise mistakes, ask Midjourney to generate simpler images with fewer elements. For example, instead of generating a sticker sheet, ask Midjourney to generate individual sticker and then combine them.