Image to Image - Walkthrough

What is image to image?

“Image to Image” is a great feature if you have an image you want to use as a reference but wish to alter. As the name suggests, an image is transformed into a new image. The composition and colors of the original are carried over into the new generation.

Where can I find it?

There are various ways to access “Image to Image”. First, you can find the function on the left side, where all our functions are typically located. Once selected, you will see a blank interface where you can upload an image as a reference.

Another option is to use a generated image as a reference by clicking the “Continue Workflow” button. Here too, you can always choose one of our other functions. Once selected, your reference image will appear in the small preview box at the bottom right. You also have the option to change the function using the small brush icon. As you can see, you have a lot of flexibility.

Now you have a reference image. On the right side in the settings, you will see everything you can change about the image.

The Settings

First is the Sampler. You can play around with it as it has a big impact on the generation but if you are unsure. Just keep it on default. Then there is Image Dimension. Currently, the format here should match the format of the reference image to avoid distortion. However, we are working to eliminate this requirement in the future.

The Number of Images indicates how many images will be generated at once. This provides you with a larger selection. The Image Weight Slider is probably one of the most important as it controls how much the image should differ from your reference. The further it is set towards 100, the more the image changes. At 100, you get a completely new image that has nothing to do with your original. If the slider is set to 0, the image does not change at all, and you get your original recreated.

Time - Image Quality counts the number of steps the image will need. Like a painting, the first step is a roughly sketched composition. With each subsequent step, the image becomes more precise. Therefore, quality is also related to the time required to create the image. However, at some point, a generated image will have such a level of detail that any further steps will make no difference. We have found that a value between 20 and 40 is sufficient. Anything beyond that does not noticeably change the image anymore.

AI-Human Collaboration allows you to set how creative the AI can be or how strictly it should adhere to your prompt.

The Seed can be thought of as the origin of your image, like the seed of a tree planted in the ground. If all conditions remain the same—same light, same amount of water, wind, temperature, etc.—the same tree will grow again.

This is how you can imagine the Seed here: If this number remains the same, the origin of the image remains the same, and so does the image with the same settings. If this number changes, the origin of the generation changes, and so does the resulting image. Here is a simplified representation.

Prompt and Output

The Prompt is the field where you decide what should be generated. You write in simple words what you want to see. Don't be alarmed. In this example, the quality booster was turned on. This automatically generates many prompts to support your output. This is currently well-suited for realistic images with characters.

Finally, press Create and wait a moment for your image to be created.

If you want to learn more about our features read our tutorials for ControlNet or Inpainting. Or just go to our tutorial blog section