Nikras Space

Get In Touch

WRITTEN AND CREATED BY OLEKSANDR KRASNYK

This was done
by AI, right?

This collection showcases the results of my generations in Midjourney v6.1. The model
was trained based on my personalization settings*, using dataset of 390 selected images
by me. My goal was to generate images that feel as close as possible to those created
by humans. To achieve this, I conducted series of test generations.

This collection showcases the results of my generations in Midjourney v6.1. The model was trained based on my personalise settings, using dataset of 390 selected images by me. My goal was to generate images that feel as close as possible to those created by humans. To achieve this, I conducted series of test generations.

In my opinion, AI-generated images have the following typical shortcomings:

Incorrect anatomy (e.g. hands and fingers, facial asymmetry) distortions in forms

Inconsistent details that interact illogically with the surrounding environment

Repetitive patterns, which reduce the uniqueness of the image

Creating a perfect, visually appealing composition makes the image unnatural

01 Old broken electronics in a landfill

01 Old broken electronics

in a landfill

GenAI add inconsistent objects that don't fit your expectations?
Let's generate garbage!

GenAI add inconsistent objects that don't fit your expectations? Let's generate garbage!

prompt: “old computers with broken screen under the office porch near the trash bin, graffiti on the background, 4k, high-quality, --personalize p2dzw5h”

You cannot notice incongruous objects if you don’t know what they are supposed to look like—can you imagine trash?😄

Most of the photos in this collection feature a cold white balance typical of 35mm Fuji film. Interestingly, this wasn’t intentional—I didn’t include it in the prompt or aim for a film-like aesthetic. It’s purely a result of personalization in the process.

Most of the photos in this collection feature a cold white balance typical of 35mm Fuji film.

Interestingly, this wasn’t intentional—I didn’t include it in the prompt or aim for a film-like aesthetic. It’s purely a result of personalization in the process.

*Use my personalization token to generate something similar
in your Midjourney account!

*Use my personalization token to generate something similar in your Midjourney account!

p2dzw5h

02 Indoor party

02 indoor party

Can you understand what’s happening in the background if the cameraman’s hands were shaking during a drunken spree? It’s the perfect time for us to create this!

prompt: “Dynamic warm light cinematic shot, fisheye photo of apartment, girl in black dress and boy in white shirt dancing in the front of camera, happy faces, crowd on the background, chaotic, 4k, high-quality, party, --personalize p2dzw5h”

I like how each generation maintains consistency with the previous one. This is especially noticeable in the white balance, shooting techniques, and color combinations. Thanks to this, the video creates the effect of experiencing a single story throughout its entire duration.

A dynamic image with radial blur, motion blur, or the use of a long exposure can sometimes smooth out the flaws of an unsuccessful generation and “save” the image. In my opinion, this slight imperfection adds a sense of naturalness and brings a lively touch to the photographs.

03 CRIMINAL MOVIE frames

I couldn't see Godzilla in the movie because of the darkness—will the audience be able to notice flaws in our 'dark' crime story?

prompt: “Wide shot of the white van pulling up near the secluded suburban house. The scene is quiet and suspenseful, emphasizing the deserted street. Muted blues enhance the tension as a shadowy side entrance comes into view, with a stray dog watching. --personalize p2dzw5h”

I faced a truly challenging task here: to create a storytelling piece about robbers where each frame would maintain a consistent color mood. However, I couldn’t fully achieve this effect—it will likely require detailed manual work during post-processing.