Learn how to solve problems with complex prompts.
#Tip 5 - Get the main subject first, then add details
dense hound
The more elements you request, the more difficult it is for the AI to get a coherent solution. Try simplifying to get the most important element first. If that works, add the next most important thing. You may notice that coherence falls apart when you add some specific goal. When prompts fail in ways you don't understand, try doing the prompt in stages, specifying the most important element first. If that gens, add details in stages. Here is an example in which we start with our main subject and a very flexible art style: “a robot surfing. digital art.”
These are all pretty good, because we kept clear focus on the main subject. If we want more details around the scene we could get them by outpainting one of these images. But let’s try adding more detail about the scene in the prompt itself. Let’s try to get a huge wave, and the specific trick called tube riding: “a robot surfing a huge wave. digital art. tube riding.”
We still have a coherent image, and the second one in particular is well-composed. But do you notice that the robots are now smaller? That is because we put more emphasis on the wave, which now must take up a larger portion of the image, reducing focus on the robot. When you add more requests for a setting, the main subject may start to break down.
Let’s imagine our robot is in Hawaii, wearing a lei while surfing: “a robot wearing a flower lei surfing a huge wave. digital art. tube riding.”
