The Art Button Manual

Prompt

The text prompt is how you tell Stable Diffusion what you'd like to have generated. Working to improve prompts is its own skill and we suggest reading some articles and watching some YouTube videos that dive into creating great prompts.

To get you started, here is a nice introduction: https://www.howtogeek.com/833169/how-to-write-an-awesome-stable-diffusion-prompt/

Input image

An even stronger way to guide Stable Diffusion is to supply it with an image that is similar to what you want generated.

You can click to upload and select an image or drag-and-drop images onto TheArtButton.ai to use it as a prompt input image.

We even support pasting an image from image editing programs such as Photoshop or Gimp. Just copy the image in your editor and then use the keyboard shortcut to paste it into TheArtButton.ai.

Input image strength

This determines how closely you want Stable Diffusion to stick to the input image. A high strength indicates that you want something very similar, with perhaps minor changes. A low strength indicates you want something perhaps stylistically similar, or sharing colors and shapes but not necessarily the same specific image content.

Lower strength will result in increased image diversity. Great for exploring possibilities. Higher image strength is great for taking a good image and making it great.

Image size

Stable Diffusion is trained on 512 x 512 images. While generating larger images is fully supported, it can be harder to get what you want depending on what you're trying to produce.

Often using an input image can help strongly guide things when generating larger images.

Larger images take longer and as a result, cost more tokens.

Generation

Increasing the iterations will generally produce higher quality images. However, these images will take longer and can cost more tokens. Learning the relationship between samplers and iterations will enable you to generate high-quality images while being efficient with your tokens.

To learn more about samplers and iterations, here is a very deep but wonderful resource: https://invoke-ai.github.io/InvokeAI/help/SAMPLER_CONVERGENCE/

Features

Seamless tiling is great at producing textures or pattern backgrounds. These can be useful for games, websites, and other digital media.

We also support entering a specific seed. Using the same settings and seed will produce an identical image every time. If you enter a specific seed and only adjust one setting, you can learn what that specific change did to the generated image. This is great to learn more about prompt improvement and different samplers.

Costs

Generating images on TheArtButton.ai costs tokens. Generally, the longer an image takes to generate, the more tokens it will cost.

We top-up accounts to 100 tokens every 24 hours. This gives everyone a chance to experiment, learn, and improve at generating art with Stable Diffusion.

For those interested in unlocking much more image generating power, we also sell token packages to enable passionate individuals and professionals to make even more extensive use of our generating capacity.

Output images

The images that have been generated by TheArtButton.ai will display below the "Art" button that was clicked to start generating images.

These images can be shared, downloaded, or even used as input images for further tweaking to get an image that is just right.