Other Tips & Tricks
Use the below galleries by clicking into them and then clicking through the images.
The best tip? Keep trying. Try different words, and if you get close, just generate more images. Some people will even generate dozens of images in one prompt, so they can get exactly what they're looking for.
More tips and tricks coming soon.
Photography
Using photography keywords can help you push the AI to do exactly what you want. There are quite a number of aspects of photography that can be implemented in your prompts, with varying degrees of success.
Composition is an important aspect of photography, regarding the placement of subjects in the image. "Center Composition" can sometimes help make a scattered Wombo more clearly grouped.
Focus is similar. Using "focus/bottom" with prompts like "shoreline" can help stabilize the image into two distinct parts of land and water.
"Leading lines" can add one or more long lines in your images.
Depth of field can be used in either shallow or deep, but shallow seems to make a more consistently dramatic difference.
Resolution can make a slight difference, so try adding 4k to a prompt. In tests, 4K Wombos seemed to have sharper, more detailed aspects, even though their actual resolution had not changed. Similar phrases include RTX, ray tracing, and HDR.
Styles of photography, such as macro or long exposure, can also work well, but take some experimenting. Some prompts work better than others.
Certain artists have a photo-realistic style, and some people find better results by using their names. Richard Estes and Joongwon Jeon are two specific examples, but you can easily find more by searching "photo-realistic artists."
Grammar
Oft discussed in the WOMBOVERSE Discord, grammar is a controversial issue.
First, there is the ordering of the words in your prompt. Is "Forest by Van Gogh" better than "Van Gogh forest?" Is the word "by" necessary?
Then, there is the issue of punctuation. Do brackets help, such as in [Style: XYZ], or are they merely stylistic?
Unfortunately, there is not a clear-cut answer. Running multiple tests indicates little difference in most of these questions. For example, "Haystacks Van Gogh" and "Haystacks by Van Gogh" and "Haystacks [Style: Van Gogh]" are all unmistakably in Van Gogh's art style, and the subject is obviously haystacks. Brackets may help to separate concepts, but are still not a guarantee. "[red sky] [blue ground]" is just as likely to yield a jumbled up mess of red and blue as the same prompt with no brackets.
[Style: XYZ] is another grammatical separator which is often unnecessary. "Rose watercolor" looks just as realistically watercolored as "Rose [Style: Watercolor]." If you notice that a style doesn't seem to be working without the formatting, it doesn't hurt to try it.
Some people use pipes "|" to separate concepts, but this seems to have little or no effect on the images.
Saying Mass in the Space Vatican
People and Animals
Getting the AI to create landscapes is one thing. But getting it to create a non-demonic face of a human or animal is something entirely different.
Getting a consistently high-quality, front-on portrait is near impossible. You may, however, be able to get a good one with a specific enough prompt and enough attempts. Repeated attempts are key in creating faces. It may take dozens of times before you get anything good, even with a great prompt.
If you're not interested in trying again and again for one image, but you're still interested in having some life in your Wombos, there is one more tip: Use groups. There are lots and lots of words for the groups you might find humans in, such as club, mass, riot, etc. Most animals have one or more names for their groups, such as a flock of geese or murder of crows. Groups tend to work better because the distance from the subjects means less detail is necessary. Even still, it can take a lot of trial and error with your prompts to get a good image, especially with more complex forms like humans. Simpler forms like birds are easier.