Maze Guru Gallery has opened from now! Enter in to view Maze AI images!

Maze-AI

Welcome to Join Maze Guru Discord Community!

Join the Maze Guru Discord Server!

1. Start From Simple Meme

It’s hard for new bird to be skilled with complex prompt words and elegant images, you may feel confused with chaotic and poorly colored “defective goods”. Try to start from meme for fun! For exmaple fill in “Musk” or “Iron Man”. Maintaining interest in AI art is the hey to keeping yourself exploring ai art.

2. Perfer Short Phrases And Pay Attention to Logic

As AI is still hard to understand the overly complex human mind, it’s better for users to choose short phrases and sentence and pay attention to logic.Also, too many word combinations can overwhelm the AI.

Although advocate comprehensive phrases, too many combinations may can’t be reflected in the images, but will produce diametrically opposite artistic effects. Therefore, users need to find the balance between “too little” and “too much” in the process of trying.

What’s more, grammatical skillsv are helpful. If you want to emphasize a certain word, try to add some brackets or exclamation marks. For example, in the “Desert!!, Smog, River, (((sunset)))”, Desert and Sunset will get the priority. This trick is especially useful when generate portraits. Add a few more exclamation marks with symmetry if you want to generate a symmetry face.

Case:

[Symmetry!! portrait of a robot astronaut, floral! horizon zero dawn machine, intricate, elegant, highly detailed, digital painting, artstation, concept art, smooth, sharp focus, illustration, art by artgerm and greg rutkowski and alphonse mucha, 8k]

3. Core Descriptive Words + Artistry Modifier

[Symmetry!! portrait of a robot astronaut, floral! horizon zero dawn machine, intricate, elegant, highly detailed, digital painting, artstation, concept art, smooth, sharp focus, illustration, art by artgerm and greg rutkowski and alphonse mucha, 8k]

The green part is the subject.

The yellow part is the modifications to the subject.

The blue part is some professional artists and art studio styles.

Tips: When you see someone share prompt text you can take their artistic modification part as reference for your image’s subject. By suck simply grafting, you will easily get the desired artistic effect.

4. Choose The Right Image Size

1) For portrait images:1:1 square recommended first.

You may get two or more faces when choose horizontal or vertical, which is often dubbed “buy one get one free” by users.

2) For spatial structure images such as landscape: horizontal or vertical recommended first.

5. Work A Miracle

Even a group of wonderful prompt text may output inferior images under the randomness of AI (such as double-headed, chaotic structure, illogical, distorted and exaggerated).There is only one solution — a miracle in large quantities.

Mint up and find your own unique wealth! One piece exists!

6. About Maze AI Model

Maze AI recommends tochoose common model.

The common model has perfect logical and aesthetic understanding of characters, animals, landscapes, scenes, etc. That’s the first choice for new users and other professional art workers such as anime or design. Maze AI will contiune to increase resolution (including a lot of noise reduction, sharpening, optimization, etc.), the rendering effect of common models will get more and more consummate.

Using color model often results in a long wait due to low speed. Many new users use simple prompt phrases in the color model and find it to be very ineffective. So it still needs more attempts.

But the power of the color model lies in the “Magnificent Beauty”. When you don’t talk about logic but just want to feel the visual impact of color. We recommend you to choose color model, which is very suitable for wallpapers, backgrounds, etc., but the premise is to find suitable prompt phrases and sentences.