May 2023 – present
Project Profile
This project has been cooperated with Prof. Lili Chen who is the chairman of the Department of Stomatology at Tongji Medical College, and this project aims to realize remote non-invasive medical healing of injured rats based on electromagnetic induction. In this project, I was mainly responsible for the magnetical part, which included the determination and optimization of electromagnetic-related parameters. Besides, I have also done some work on the initial schematic diagram, for which I used emerging AI drawing tools.
AI drawing and Midjourney
As a native of the AI era, one should have a certain AI mindset to understand new ways of thinking and solving.
AI drawing is an indispensable tool in the academic realm. In today’s landscape, AI, exemplified by ChatGPT, has made a significant impact across diverse industries. AI drawing, as an integral component of this technology, holds immense potential in various domains, as well as in the realm of scientific writing. With the aid of AI drawing tools, all it takes is providing a precise descriptive phrase (prompt) and/or an image, and AI swiftly generates the desired illustration based on its algorithm.
Midjourney is a very popular and simple one among them. Operating within the Discord platform, it can generate four similar images through your input (like image prompts, text prompts, and parameters). The fundamental command, as illustrated in Figure 1, involves simply typing ‘/imagina prompt’ in the channel and incorporating your customized instructions (Discord channels give very user-friendly prompts). If you find a result that meets your satisfaction among the four options, you can use it as a starting point to generate four different variations, allowing for greater flexibility and creative exploration.
As you can see, I provided simple scientific schematics of a device used for human spine detection as image prompts, along with text prompts such as ‘scientific drawing,’ ‘detailed information,’ and ‘clarity.’ However, the results generated by Midjourney are not entirely satisfactory. They contain various errors and lack the necessary level of detail and precision. These shortcomings can be attributed to a combination of factors, including the prompts you provided and the inherent limitations of existing AI drawing technology.
Despite these challenges, there is still potential to achieve the desired schematic by combining the output from Midjourney with other software tools. This collaborative approach can help address the deficiencies and ensure the final result meets the required standards of detail and rigor.
Midjourney in this project
Here, I would like to diagram an injured rat being treated inside a closed medical box. Ideally, I’d like Midjourney to generate the diagram I need directly, so I fed these prompts (shown in the caption) into Midjourney and tried a few different prompts. By the way
Indeed, while the images generated by Midjourney are aesthetically pleasing, they may not always align perfectly with your specific requirements, and occasional anomalies, such as the rats having two tails, can occur. Achieving ideal images often depends on crafting precise and effective prompts. While leveraging ChatGPT for prompt refinement is a viable strategy, it’s important to note that it may not guarantee quick and consistent results.
AI-generated images can be highly context-dependent, and fine-tuning prompts to obtain desired outcomes can require experimentation and iteration. It’s also important to keep in mind the inherent limitations of current AI drawing technology, which may not always produce perfectly tailored results.
Midjourney and PS
Since Midjourney cannot now generate detailed and accurate images based on your limited prompts, combining it with other software like Adobe Photoshop (PS) is an excellent approach. This way, you can use Midjourney to generate individual objects separately, such as a mouse and a medical kit, and then integrate them in PS or other software. Figures 5 and 6 showcase Midjourney’s drawings of an injuried mouse, while Figures 7 and 8 depict the medical box, respectively.
In this case, I simply selected separate images of a mouse and a medical kit and then stitched the two together in PS to get the image I wanted.
