Jakarta, INTI – Recently, social media has been buzzing with the surge of images generated by text-to-image AI technology. From Instagram to X (formerly Twitter), hyper-realistic images have become increasingly common across timelines. Among the most widely used platforms Among the most widely used platforms for creating these images are ChatGPT and Google Gemini
Image Results from ChatGPT: Realistic but Lacking in Variety
In a test using five sample images generated by ChatGPT, the results were technically impressive. Lighting, color contrast, and composition such as blurred backgrounds (bokeh) and subject positioning closely resembled professionally taken photos.
Additional details like shadows, skin pores, and even facial oil reflections enhanced the sense of realism in the images. However, ChatGPT also displayed several limitations.
Some facial expressions appeared stiff, body poses tended to repeat, and the colors were often overly vibrant. In many cases, body proportions seemed unbalanced with heads that were too large or too small. Eye contact with the camera also felt lifeless, lacking emotional depth. Even when using the same prompt in a new session, the output showed little to no creative variation. ChatGPT seemed stuck in a repetitive visual style and structure.
Image Results from Google Gemini: More Expressive and Varied
Google Gemini, on the other hand, demonstrated a different set of strengths. While its images may not be as detailed or high in contrast as ChatGPT’s, they appeared more natural and human like. Lighting was more balanced, colors were softer, and objects blended more harmoniously with the background.
Notably, Gemini was able to produce five image samples with varied poses, expressions, and settings. From camera angles to character gestures, the outputs looked more creative and emotionally expressive. The characters' faces conveyed more lifelike emotions, with relaxed smiles and engaging eye contact.
However, not all results were perfect. One sample looked overly "smooth" as if heavily edited with digital tools. In real-world scenarios, such as at a concert with limited lighting, perfectly clear and noise-free images can feel inauthentic and overly artificial.
The Key to Generating Realistic AI Images
When creating AI-generated images, prompts or textual descriptions play a critical role. The more detailed the prompt, the more accurate and realistic the resulting image. Both ChatGPT and Gemini interpret prompts differently, even with the same content, so users can tailor the descriptions based on their goals.
For example, a prompt describing a young woman attending a nighttime K-pop concert can be used to test how well AI captures emotional and vivid image atmospheres.
Use AI Responsibly and Ethically
It is important to to use AI-generated images responsibly. Avoid using them to spread hoaxes, impersonate others, or commit any form of fraud. Most AI platforms automatically include watermarks or tags to indicate the image was machine-generated, as part of transparency efforts.
When sharing AI-generated images on social media, it is advisable to include a note clarifying that the image was created using AI. This promotes honesty, digital ethics, and helps prevent potential misuse in public spaces.
Conclusion:
The comparison between ChatGPT and Google Gemini in generating AI images reveals that each has its own strengths and weaknesses. ChatGPT excels in technical details and lighting, but lacks variation and emotional expression. Meanwhile, Gemini produces images that are more expressive, creative, and natural looking, even if they are not as visually sharp. With the right prompt, both platforms can be powerful tools for visual exploration using AI as long as they are used wisely and responsibly.
Read More:OpenAI Brings AI Technology to Barbie and Hot Wheels