My expectations for Ai image generation 2025
As I sit here at my workstation, I can't help but marvel at how far AI-powered image generation has come in the year. It's almost hard to believe that I used to struggle just to create the simplest of scenes, let alone anything with real narrative depth.
But now, the possibilities feel endless. I pull up my latest project - a sprawling fantasy landscape, complete with towering castles, majestic forests, and a pair of adventurers locked in an intense battle. With just a few keystrokes, I set the scene in motion, and suddenly, the characters spring to life, their movements fluid and natural.
But as of make a story with Ai image generation tools it is hard to make the scene to look like as you like it to be. More control of the camera and character interaction would be a great improvement.
For example if you want to have two characters to interact i different ways you have to work really hard. Often you get great images but not images that can tell a story. I have made some quick tests. Take a look at these examples below.
SDXL
1woman, woman is angry and looking at man. 1man, looking at woman, scared. Side-view, view from the side.
SD 1.5
1woman, woman is angry and looking at man. 1man, looking at woman, scared. Side-view, view from the side. full body shot. city environment, in a city. Miami, USA. Beautiful day, cinematic light, shadows, raw photo,
SDXL
1woman, woman is angry and looking at man. 1man, looking at woman, scared. Side-view, view from the side. dressed in beautiful clothes, full body shot. city environment, in a city. Miami, USA. Beautiful day, cinematic light, shadows, raw photo,
The examples are simple and quick and of course with some better prompts the images could have been much better, but this is to mainly to illustrate a point and suggestions to make tools that add functions that make ai tools more easy to handle if you want to make a story with it.
It could function like you give a name:
"1woman name Carla. 1man name Alfredo. Holding hands. Side-view, view from the side. dressed in beautiful clothes, walking, in a city. Miami USA. Beautiful day, cinematic light, shadows, raw photo"
next image:
"1woman name Carla. 1man name Alfredo. Carla slips and fall face down, Alfredo tries to help, rushing to Carlas side. low angle, from below. dressed in beautiful clothes, in a city. Miami USA. Beautiful day, cinematic light, shadows, raw photo"
next image:
"1woman name Carla. 1man name Alfredo. Carla is crying, tears, Carlas clothes is dirty, Alfredo tries to comfort Carla, Alfredo is hugging Carla. view from the front, frontview. dressed in beautiful clothes, in a city. Miami USA. Beautiful day, cinematic light, shadows, raw photo"
And so on... It is of course possible to make a story with a well tuned promt but it takes a lot of time and effort to make it work. If it could improve by making key elements better it would be great.
Image prompting adjusted to make a story could be the next step from making great looking random images. Think of the possibility's to make exciting visual stories with consistent characters for games, movies, books.