Review of the Coloso course 'Super Introductory Course: Starting Image Generation AI with Stable Diffusion' provided by AICU

Hello everyone. I recently took the course 'Super Introductory Course: Starting Image Generation AI with Stable Diffusion' provided by AICU on the education platform Coloso, and I would like to share my thoughts. Note that the title image for this article is an image I previously generated, re-output using Stable Diffusion v3 after taking the course. I utilized the tools used in the course for the image generation.

*This post was created as part of the Coloso partner program, and
this may provide a certain amount of compensation.

'Super Introductory Course: Starting Image Generation AI with Stable Diffusion' - Coloso course provided by AICU-

1. Motivation for taking the course

I regularly check AICU's note, and it was there that I saw the Coloso course 'Super Introductory Course: Starting Image Generation AI with Stable Diffusion' introduced, which prompted me to take it. I have been generating images using Automatic1111, but I had started hearing about the latest Stable Diffusion v3 and ComfyUI, and since I wanted to deepen my understanding of new technologies, encountering this course at this time was very welcome.

2. New discoveries in technology

Using Stable Diffusion v3 via API was an experience beyond my imagination. I felt great potential in the image generation quality, which is a cut above the tools I have used until now.

What was particularly impressive was the usage method via the Stability AI site and API. For me, who had only thought about using cloud services or local environments, it was a moment where a new world opened up.

3. Insights gained from goal setting

The course included a worksheet called 'Confirming Your Goals.' The experience of gradually clarifying my goals as I filled it out was impressive.

State before taking the course

Although I had an interest in image generation AI, my specific usage methods and goals were vague.

What became clear after taking the course

• Things I want to do

  • Catching up on new models and prompt techniques

  • Improving illustration quality

  • Exploring new methods of expression

• Skills I can utilize

  • API development utilizing programming knowledge

  • System construction with ComfyUI

  • Development of new methods of expression

• Future prospects

  • Developing business utilizing AI

  • Pursuing possibilities as a hobby

  • Expanding the range of creative expression

Through the course, I was able to see concrete goals and the path to achieving them.

4. What I learned through practice

Evolving prompt technology

I learned that the way of thinking about prompts has changed significantly between the SDXL era and SD3. In the course, I was able to experience actually outputting images using samples from the Stability AI developer platform, which was a very beneficial learning experience.

What particularly impressed me was the significant improvement in controllability in Stable Diffusion v3. For example, with the following simple prompt:

"young woman is wearing kimono"

Although some parameters are set separately, even with such simple instructions, surprisingly high-quality images were generated. Note that this image was generated using the highest quality setting, Stable Image Ultra.

Efficient Development Environment

The convenient toolset provided by AICU was extremely helpful for practical API utilization. These tools are provided as Windows batch files, and at the time of the course, they were not directly usable by Mac users. However, after the course, I attempted to convert them into shell files for Mac using ChatGPT and GitHub Copilot, and I confirmed that they work without any issues. This shows that Mac users can also utilize the provided materials, such as batch files, by converting them for their own environments.

Practical Production

Through the production of lookbooks and advertising flyers, I was able to hone my practical skills. What left a strong impression was the way of thinking regarding judgment criteria—what to prioritize and what to select or discard. I was able to learn a production process with clear intent, from overall layout to fine-tuning details.

5. Future Goals

Technical Challenges

• Development of new services utilizing APIs

  • Systems that automate prompt optimization

  • Tools for efficiently managing image generation results

  • Integrating AI image generation features into web applications

System Construction

• Building original systems using ComfyUI

  • Establishing efficient image generation workflows

  • Exploring combinations of nodes that suit my preferences

  • Improving work efficiency through automated batch processing

Regarding ComfyUI, I have also taken AICU's other course, 'ComfyUI Master: Custom Nodes and App Development,' and I plan to share the detailed content in my next review.

Creative Expansion

• Pursuing deeper expressive power

  • Establishing a unique style and taste

  • Exploring effective ways to use it in lookbooks and advertisements

  • Integrating it with creative hobbies

Business Development

• Service development utilizing image generation AI

  • Customization according to client needs

  • Establishment of efficient production workflows

  • Exploration of new business models

6. Good points of the course

The PDF materials for each session are comprehensive, making the structure easy to review. Being able to check parts I missed later was extremely effective for deepening my learning.

7. Mindset as a creator

What left an impression on me throughout the course was the perspective on the responsibilities we have as users of generative AI. I was able to deepen my thinking not only on acquiring technical skills but also on how to use them appropriately.

This became a very important realization for my future activities as a creator. I have come to realize the necessity for us to act with appropriate judgment as technology advances.

Finally, I would like to thank Coloso for providing such a wonderful course, and AICU for sharing practical tools and valuable insights. I would like to take on the challenge of new possibilities by applying what I have learned in this course.

If you have the opportunity, I highly recommend you consider taking this course as well.