Skip to the main content.
Get Started
Get Started

1 min read

Room for Improvement: Midjourney's AI Images

Room for Improvement: Midjourney's AI Images

Midjourney is a text-to-image AI tool. Like us humans, AI tools are not perfect, and there is still room for improvement, especially in generating realistic images. Here are three main challenges we faced recently:

1. Five-finger problems

Generating lifelike hand images can be complex, and several challenges often arise. Among them are issues such as the presence of extra or missing fingers, inaccuracies in hand proportions, and the need to refine details like nails and skin texture to achieve a more realistic appearance.
Surreal photography of a man in his late 30s, looking at creative marketing graphic digital designs on a large floor to ceiling monitor glass screen, creating a sense of happiness and satisfaction, wearing business casual clothes, in a modern office, blue and purple gradient lighting, in the style of biomorphic forms, cryengine, meticulous photorealistic still lifes, molecular structures
Prompt: Surreal photography of a man in his late 30s, looking at creative marketing graphic digital designs on a large floor to ceiling monitor glass screen, creating a sense of happiness and satisfaction, wearing business casual clothes, in a modern office, blue and purple gradient lighting, in the style of biomorphic forms, cryengine, meticulous photorealistic still lifes, molecular structures


2. Creating Accurate and Realistic Landscapes

Midjourney is an impressive tool that often surpasses our expectations in image generation. However, it may not excel in producing highly realistic or accurate cityscapes. One of its drawbacks is the tendency to manipulate buildings or landmarks in a way that compromises their authenticity.
aerial photography of Sydney, beautiful architecture, stunning view, city lights, blue hour
Prompt: aerial photography of Sydney, beautiful architecture, stunning view, city lights, blue hour --ar 3:2


2. Eyes on group photos

In Midjourney, images are sometimes generated with unrealistic eyes, especially on group photos. The eyes may appear filled with black, lack proper proportion, or have pupils looking in different directions.
Casual photograph of a diverse team collaborating, candid, happy, neutral tones, wearing red green and yellow clothes, warm, natural daylight, set in a bright office, vivid colours red green
Prompt: Casual photograph of a diverse team collaborating, candid, happy, neutral tones, wearing red green and yellow clothes, warm, natural daylight, set in a bright office, vivid colours red green, Canon EOS 5D Mark IV camera, lifelike, contrast, wide shot, 4k, --ar 7:4


Midjourney has undoubtedly pushed the boundaries of text-image AI generation, offering impressive capabilities and remarkable potential. However, like any cutting-edge technology, there are challenges that need to be addressed. The exciting news is that AI tools are continuously improving and advancing rapidly, so it’s possible that these challenges may not persist in the coming months.

 

Midjourney's Inpainting feature: Vary (Region)

Midjourney's Inpainting feature: Vary (Region)

In the ever-evolving realm of image editing, Midjourney leaps boldly ahead with its groundbreaking inpainting feature named "Vary (Region)". This...

Read More
Exploring Midjourney's AI-Generated Illustrations

Exploring Midjourney's AI-Generated Illustrations

Midjourney is getting a ton of love for its unique approach to making photos without models or photoshoots. But hey, there's more to it than just...

Read More
Trying Midjourney's Video Feature

Trying Midjourney's Video Feature

Have you tried the fun little feature in Midjourney that makes your image move? Try using --video parameter; it produces a 5-second video of your...

Read More