AI art - how does it work?
Understanding how AI art works is the key to knowing what to expect from it - both its' advantages and disadvantages. AI doesn't have a thinking brain that a human does - it operates on the concept of machine learning. For example, if you mention "a corgi" to a human, they will immediately know what a corgi is. If you show a photo of one to a human, they will recognize it as a corgi regardless of the angle, pose, lighting, etc. AI doesn't have the sapience required for that approach 0 instead, when you say "corgi" to an AI, what it does is extensively search for it. It will find and look upon hundreds or thousands of pictures of a corgi. It still won't know what it is, since a machine can't really "know" in a way that a sentient being does, but it will have a very good idea on what a corgi looks like from combining those pictures.
The good and the bad of AI art - tempering your expectations:
Both advantages and the disadvantages come down to a single thing: the AI is a computer, which works in the way described above. The most obvious advantage to AI art is accessibility. Anyone with an access to a modern-day electronic device and an AI art tool can create beautiful works of art in mere minutes. Which is also the second advantage: speed. A computer thinks much faster than a human brain does, and draws much faster than a human hand - it can do in minutes what would take days or weeks to a human. However, the AI can make mistakes; even some that no human artist would do. Since AI doesn't know what a corgi is - it only has an idea about what it looks like - when it makes a mistake it can't tell that it did something wrong. The most usual mistake for example, that even the most advanced AIs struggle with, is drawing extremities. Hands, feet and paws might have weird shapes or a wrong number of fingers/toes. Sometimes it might struggle with maintaining symmetry; usually not really noticeable at a glance but becoming apparent with closer inspection of a resulting image. Sometimes it'll pick a bad batch to "learn" from and the results will reflect that. There are still limitations to what an AI can do and when approaching AI art, one must do so with these limitations in mind.
What can I do to make my AI art better?
While the AI does the heavy lifting, the user has one key part where they can exert control: the input prompt. The most important (and really only)bit of knowledge needed when using AI art tools is understanding how to give good prompts. AI works best with well defined, but clear and concise instructions. For our app specifically, make sure you choose the breed and the coat color correctly. If you're uploading a photo to be worked on it's best that the photo is sharp, with even lighting, well-defined colors and a clear difference between the object of the photo and the background (for example, uploading a photo of a black-and-white cat resting on a black-and-white carpet will likely give AI troubles in separating the two, resulting in a jumbled photo). If you choose the "mixed breed" option from the breed list, you'll be given a text prompt. While It is recommended that you only enter the breeds in here, it is still a fully functioning text prompt which can in some cases work just as well as a photo prompt. The same principle applies: be as clear and concise as possible. Define everything that you want to see in a picture, but try not to overdo past the definitions or use abstract terms. Also avoid giving conflicting prompts when possible; for example, if you tell the AI that your pet is one breed but upload the photo of another breed, the result won't be satisfying as the AI will have to attempt to draw two different things at once.
Finally, if you want a different result from what you got, you can simply run the prompt again. AI can make vastly different images from the same prompt (the magnitude of which depending on how much freedom it was given when making the input prompt) - try as many times as you'd like and you might be surprised!