hwan's honest review, 2026 AI Image Generation Core with ChatGPT IMAGES 2.0 (Duct Tape) course
Reviews 4
Average rating 3
This course is essentially paying 138,600 KRW to purchase the 'prompt templates' for image generation introduced in the materials. The course content follows this pattern: 1) Enter the prompt attached in the course materials into GPT. 2) Then, this image is actually generated. At the very least, there should be a review and introduction of the complex prompts the instructor created (which span about two A4 pages), but there is none. The course content simply consists of demonstrating that images are generated when you paste the prompts from a notepad into GPT. The prompt lecture chapter only covers very general information and basic guides, such as the need for clear requirements and specifying desired camera angles. There is no actual instruction regarding the long, detailed prompts the instructor personally crafted for POP and wedding photo printing. It truly only demonstrates that images are generated when you plug them in. The only "application" mentioned is changing minor variations like titles and colors within those templates. I wondered why the course was so short, and now I see why. It feels like going to a math academy to learn the subject, but coming out with nothing but a formula sheet and no explanation of theory or fundamentals. This course might be useful for those who don't need detailed explanations and just want image generation prompts to plug into GPT. However, considering the intent is solely to provide templates, the tuition fee of nearly 140,000 KRW feels quite expensive.
2
Hello hwan, thank you for completing the course and leaving your honest feedback. I fully understand the disappointment you mentioned. You may have felt that the process of designing prompts from scratch was lacking, and you might have felt that more explanation was needed regarding why each element of a prompt is necessary. However, one of the key components we prepared for this course was verifying the actual implementation scope of image generation models. We directly experimented with resolution, aspect ratio, fonts, colors, positioning, spacing, and background removal. We identified what works well. We also summarized what can only be implemented as an approximation. We distinguished tasks that are more efficient to handle using editing tools. Without these established criteria, one can end up spending a lot of time during actual work. This is because it is easy to mistake a task that the model inherently struggles with for a problem with the prompt. This leads to constantly changing sentences and repeatedly generating images. Typical examples include tasks involving precise color values, coordinates, kerning, or transparent backgrounds. Conversely, if you know the potential range and limitations of the model, you can make quick judgments. You can divide the work between what to solve with AI and what to finish with editing tools. This also reduces unnecessary trial and error. The purpose of the experimental section was not just to introduce features, but to provide the criteria for judgment to reduce actual working time. The 'Stepping Stone Reference Method' was covered for the same reason. It is a method of not creating complex results all at once. You utilize the image created in the previous step for the next task. By breaking it down into steps like this, it is easier to maintain the consistency of the image and reduce errors. The prompt templates were also not prepared simply as sentences to be copied and used. They were tools to show you what needs to remain fixed during repetitive tasks. I also wanted to convey what needs to be changed for each task. There was also an aspect of providing examples of how much "fixing power" a prompt alone can have. We prepared this course believing that these contents could provide clear value in actual work. However, I feel a great sense of regret that this value did not seem to resonate sufficiently. That is why I have explained the intentions and the parts I consider important at such length. Thank you once again for leaving such specific feedback. I will ensure that your opinions are reflected in future improvements.
0