May 26, 2023 15:00:00

I tried using the AI tool `` Reimagine XL '' that can quickly generate `` similar images '' from one image and accelerate material collection

When creating materials and works, you may find yourself in a situation where you want to prepare multiple images similar to this image. If you use `` Reimagine XL '' developed by AI company Stability AI known for image generation AI `` Stable Diffusion '', you can quickly generate similar images simply by dragging and dropping images. It was available for free in Japan, so I actually tried using it.

Reimagine XL

https://clipdrop.co/stable-diffusion-reimagine

When you access Reimagine XL, the following screen will be displayed.

Using Reimagine XL is very easy. Just drag and drop the image you want to generate a similar image to the area enclosed by the dotted line.

Three similar images were generated in tens of seconds after dragging and dropping. You can check each image by clicking the thumbnails at the bottom of the screen.

The original image I entered this time is below.

And the three similar images generated are as follows. We were able to generate an image that kept the features such as 'closed eyes', 'smiling with teeth', and 'wearing a straw hat and yellow clothes'.

If you don't like the generated image, you can regenerate it by clicking the refresh button on the top right of the thumbnail.

Also, by clicking the share button at the bottom right of the image, you can download an image that combines the original image and 3 similar images.

The downloaded image is below. There is a watermark 'Clipdop' and 'stability.ai' in the lower right.

Reimagine XL is an enhanced version of ``

Stable diffusion reimagine '' released in March 2023. Looking at the following image created with Stable diffusion reimagine, you can see that Reimagine XL has significantly improved image generation performance.

Since I knew the performance of Reimagine XL, I input various images and generated similar images. First, let's generate a similar image of the following photos taken during mountain climbing.

The generated result looks like this.

Next, I will try it with a picture taken of the shopping street at night.

We were able to generate a similar image while maintaining the atmosphere of a 'shopping street lined with shops after closing'. However, when enlarged, 'a mysterious structure similar to a bicycle' is reflected.

Finally, I will try it with pictures of Nepali food.

The generated result is below. Although there are problems such as 'the grain of rice is too big' and 'the shape of the vessel is unnatural,' the atmosphere has been reproduced.