Midjourney: AI image generator and why I use it

I have discovered this AI text-to-image generation tool that pushes the boundaries of how we define art. I graduated with a fine art degree. I am used to spending days and months developing the art direction of my art projects. Then I discover Midjourney, an AI bot that uses Discord commands along with users' design ideas to create artistic images. You will write a prompt describing your imagined image to the Midjourney bot. The result is instant! The prompt will be organized into several commands to depict the AI vision of your image. The AI generated images are incomparable with fine art photographs because humans and AI are creative very differently. AI creativity is very random and the image quality lacks details and resolutions. Therefore, I think AI-generated images work fine in storyboards and sketches which need to be further developed and edited in other design software. However, Midjourney gives me art direction that I would have never thought of.

Architecture Story Telling: SkyCity

I use Midjourney to test and develop ideas quickly. Inspired by Miyazaki’s animation: Howl Moving Castles, and Tomás Saraceno’s art installations about civilization in the sky. So I have developed my first architecture idea: SkyCity, depicting a future city built in the sky. SkyCity is about the ultimately optimized interconnectivity of civil infrastructure in the sky. I am sharing three of my favorites in the gallery below.

Concrete Sandwich

The lower cantilevers support the higher cantilevers. Each cantilever has an open roof contains several apartments or commercial spaces. Cantilever rooftops are private to their residents, but you can view others’ cantilever rooftops from higher up. The rich enjoy more privacy and sunshine, living in their higher cantilevers, which are truly superior than the others.

Air Express

Bridge houses are built into mountains. SkyCity citizens can enjoy the convenience in the city while being in nature. Overpass bridges have multiple layers with apartments and shops built on top, and trains in the bottom. Image a super long corridor with apartments on both sides. The train beneath will drop you right at your destination.

Cloud Candy

We dream of living on clouds. In SkyCity there are uncountable “clouds”, scaffolding platforms everywhere in the sky. People call them stairways to the heaven. These scaffolding platforms provide solitary space to the people who need it.

This is the general prompt that I develop upon for SkyCity, lots of reediting to discover how AI depicts the same vision differently.

/imagine overlapping bridge layers of maze multi-level highway overpass derived from scaffoldings in clouds, monoliths divides the photo, 3d unreal engine 5, bladerunner in detailed Hong Kong --ar 9:16

syntax: /imagine command prompt describing the visual composition -- aspect ratio

Midjourney offers a free trial of approximately 25 free uses of their /imagine command which will be equal to approximately 25 images.

Upscale and Variations.

After a 2x2 grid of images has finished generating two rows of buttons will appear:

• In the top row: U1, U2, U3, and U4 are buttons for Upscaling your chosen image. Upscaling an image generates a larger roughly 1024x1024 pixel version of the selected image. By default, this will also generate additional details.
• In the bottom row: V1, V2, V3, and V4 are buttons for creating Variations of your chosen image. Creating variations will generate four new images, similar in overall style and composition to the image you selected.

Upscale to Max

After using U button to upscale, you can choose to:

• Make Variations: generates new alternatives in a grid.
• Upscale to Max: upscales the image to an larger resolution of roughly 1664x1664.
• Light Upscale Redo: removes some existing details. Sometimes AI creates details that look unnatural for portraits.