Prompts

Creating AI art starting with a blank prompt can be a confusing and intimidating experience. It's not easy to create great looking art right away using Stable Diffusion!

That's why diffuse.store gives you a starting prompt to get you going with the process of creating interesting AI art.

Starting Out

When first working with Stable Diffusion it's easiest to experiment with the given prompt and ask for things you think would be fun and interesting. Want a cheeseburger in your Van Gogh painting? Or maybe you want an iPhone instead? Try it out!

In the beginning don't change any of the other descriptive adjectives in the prompt so that the style stays the same. That will make it more likely that the image you get looks great.

We've done the hard work for you by create starting prompt descriptions that are likely to look good.

Getting Variations

When you click 'REGENERATE IMAGE' Stable Diffusion creates a random variation of the image based on the prompt. One image can look completely different from another! 

Going Further

You've seen what Stable Diffusion is capable of and you'd like to branch out. What else can can the AI create?

Artist Collaborations

Try adding another different or related artist to the prompt. What would it look like if Picasso and Van Gogh collaborated on a painting?

Blend Styles

What would happen if a graffiti artist painted a still life painting? Ask the AI for something like "still life painting of fruit in graffiti style, street art, spray cans, tagging".

Compositions

Ask the AI to compose the image in the way you want. You can ask it for a "closeup" if you want the image zoomed-in. Or you can ask it for a "wide shot" if you want the image zoomed-out. You can also say things like "overhead shot" to say you want a perspective from above. 

Solving Issues

What happens when you don't get what you want? The hardest part of creating AI art is that the results can be uneven.

Deformed Images

Sometimes the AI simply creates images that aren't correct. Cats are missing an eye. A house has no front door. Just try to regenerate the image and see if the new version is more correct.

Missing Elements

Sometimes you ask the AI for a cat and a cheeseburger but the image only includes a cat and leaves out the cheeseburger. The AI is trying to interpret what your prompt means and can sometimes emphasize one part of the prompt over another. Adding modifying adjectives to the prompt can help emphasize that element. Example: 'a cat and a cheeseburger on a plate, photorealistic' can be changed to emphasize the cat more: 'a very cute cat and a cheeseburger on a plate, photorealistic'.

More Resources

Asking the AI for the right thing is a complex topic and can be referred to as 'prompt engineering'.

Take a look at these other resources to learn more!

https://lablab.ai/t/stable-diffusion-prompt-engineering

https://prompthero.com/stable-diffusion-prompt-guide

https://github.com/Maks-s/sd-akashic