Exploring AI Image Generation Tools

Today i explored various image generation tools and wanted to share my experience using OpenAI's ChatGPT, Microsoft Designer, and Together AI's models like Flux1-pro, Flux1.1-pro, and Stable Diffusion XL 1.0.

I started my adventure with OpenAI's ChatGPT to help me create an effective prompt for image generation. After some back-and-forth, we settled on the following prompt:

```

A detailed wimmelbuch-style cover page illustration set in the city of Berlin with an extremely small number of people, each carefully designed with identifiable features and recognizable faces for potential reuse in future illustrations. The scene should showcase iconic Berlin landmarks like the Brandenburg Gate, Berlin TV Tower, and the Reichstag, with minimal street activity, including only a few pedestrians, cyclists, or tourists. The focus is on creating distinct characters, with visible facial expressions, detailed clothing, and accessories. The environment remains lively but is calm and spacious, with iconic yellow trams, market stalls, cafes, and open spaces that highlight the individuals present in the scene.```

## The Generation Process

With this prompt in hand, I set out to test it across different platforms. Here's a breakdown of my experience:

1. **ChatGPT (DALL-E)**:

- Result:

2. **Microsoft Designer**:

- Result:

3. **Together AI - Flux1-pro**:

- Result

:

4. **Together AI - Flux1.1-pro**:

- Result:

5. **Together AI - Stable Diffusion XL 1.0**:

- Result:

## Comparing the Results

All of them took similar time to generate an image. Each tool produced unique interpretations of the prompt, showcasing the diverse capabilities of AI image generation. Interestingly, my personal favorite emerged from the Flux 1.1-pro model. The image it generated seemed to capture my imagination.

However, it's worth noting that despite explicitly mentioning "small number of people, each carefully designed with identifiable features and recognizable faces for potential reuse in future illustrations" in the prompt, some of the models still included human figures without recognizable faces. In th end, all the tools demonstrated impressive capabilities and I will give more of them a try in the future.

Have you experimented with AI image generation? Which tools are your favorites? Share your experiences in the comments below!