dalle-2 is a pioneering image-generative model developed by openai that boasts a higher level of descriptive input compared to stable diffusion. this allows for the creation of abstract and imaginative outputs, such as a fire breathing dragon against a backdrop of clouds in a digital art style. dalle-2 is well-suited for generating concept art, pixel art, illustrations, abstract prompts, and 3d renders. however, it may not be the best option for producing photorealistic images.

best for:

• concept art
• pixel art
• illustrations
• abstract prompts
• 3D renders
• imagination

stable diffusion 2.1 / xl employs a more straightforward promotion method that is less reliant on descriptive inputs and more reliant on entered keywords. this approach results in better photorealistic outputs or less abstract digital art, such as a bear in front of a cave. stable diffusion is ideal for producing photo style images, photorealistic graphics, traditional and digital art. however, it may not be as effective for generating pixel art, conceptual art representations, and faces and hands. stable diffusion is based on an open-source dataset. results may appear slightly inferior in some areas, but excels in producing photo-realistic landscapes and building scenes.

best for:

• photo style generation
• photo-realistic images
• traditional and digital art
• real life

dalle mini is similar to dalle-2 but runs on unfiltered training data from the internet and uses zero-shot text-to-image technology, making it more computationally efficient than diffusion models.

best for:

• idealism art
• fun art
• illustrative art
• abstract art
• fine art
• landscape art
• composition art
• distorted art

deepai is an uniquely trained version of stable diffusion meant for creativity and eccentric art.

best for:

• unique art
• contemporary art
• old art
• abstract art
• fantasy art
• 3d art
• pixel art
• variety art

open journey v4 is a stable diffusion model trained on 124,000 Midjourney v4 images.

best for:

• high quality art
• aesthetic art
• avatars
• face art
• fantasy art
• 3d art
• comic book art
• video game art
• illusionary art

midjourney v4 is a stable diffusion model trained on hundreds of thousands of Midjourney v4 images.

best for:

• high quality art
• aesthetic art
• avatars
• face art
• fantasy art
• 3d art
• comic book art
• video game art
• illusionary art

prompt keyword suggestions for visual style effects:

• digital art
• digitial illustration
• traditional art
• oil painting of
• low-poly render
• close-up
• in style of picasso
• similar to rembrandt
• trending in artstation
• in steampunk style
• in style of murakami
• sketch drawing
• pixel art
• colorful
• black & white
• grayscale
• placid colors
• vivid colors
• water colors
• warm colors
• organic
• abstract art of
• fantasy
• detailed
• mesmerizing
• captivating
• anime
• comic-style
• high quality
• high resolution
• realistic
• 3d render
• pixar
• 4k
• texture
• vintage
• modern
• bent

In the command terminal at the top, make sure to pass the captcha check, then choose any of the eight engine models which can be found under commands to draw your image.

It usually takes anywhere from 30 seconds and up to 5 minutes or longer to generate an image.

There are multiple theories on why AI has a hard time generating faces including not wanting to "steal an identity" of a person who may own that face or may distort the face to give the appearance of a person, focusing on other aspects of the image, rather than the facial features.

There are millions of images that inspire the generation of the image of your request based on your command. Unfortunately, there are times where the AI engine is unable to understand your query and may produce undesirable results or gets straight to the point of being unable to produce any image at all or the response time of the server was taking too long and timed out, all which may result in an "image error". If you receive this error, the best thing to do is try once or twice more or try a different engine model.

All images have a copyright license which can be viewed above by clicking "view" for any. Most are free to distribute for personal or commercial license.

The design was heavily inspired with influence from ChatGPT which included the layout and colors, but the javascript and processing functionality came later.

Of course, you can click the download button to save your image or you may use our third-party service to purchase posters, canvases, or framed photos. Here is a lovely sample:

It is free to use artsy.sh. If you are a developer and wish to redistribute or create your own service, we offer the API for all of our engine models which costs $5 per month and $0.02 per generated image.

It is free to use artsy.sh to generate images, but if you are using the API, simply stop the service and you will no longer be charged the following month. We do not provide any refunds for signing up to use the API. If you happened to purchase a poster, canvas, or framed photo of your art through our third-party provider, CanvasPop, you will need to request a refund through them.

  generate image:
  curl -F "key=00000000" -F ai="dalle" -F "resolution=1024" -F "prompt=smiley face" https://artsy.sh/api

  display generated images:
  curl -F "key=00000000" -F ai="dalle" -F "directory=true" https://artsy.sh/api

Convert this code to your own language: https://curlconverter.com/