ChatGPT is causing a stir but not because of the words

Advertising images are composited from ChatGPT shared by Vietnamese users. Photo: TSCS .

On March 26, OpenAI integrated the “Images in ChatGPT” feature, which allows users to generate images directly using the GPT-4o model. 48 hours later, OpenAI CEO Sam Altman wrote on X that due to the high traffic of the image generator, the company had to suspend access to free accounts to 3 times/day.

During that time, many customers in Vietnam had the opportunity to experience it. Experts and industry insiders have also had some comments and advice around this new image generator.

"Surplus" for advertising photos

Trained on a large language model, Images in ChatGPT can support Vietnamese well. In addition, users have discovered many practical applications of the feature such as creating concepts, styles for photos, making professional product descriptions or illustrating articles and comics.

Specifically, when entering the keyword “ChatGPT create image” into Facebook or TikTok, countless articles, videos, comments or instructions on how to create images will immediately appear. “I find ChatGPT’s new image creation feature quite impressive, especially its ability to understand Vietnamese commands,” said Ms. Truong Thuy, a freelancer working in AI graphic design.

Meanwhile, the Anh Viet account posted an article commenting that ChatGPT processes images as well as Photoshop but smarter.

Many applications of Images in ChatGPT are discovered by users. Photo: Meta.

However, many other users also argue that this is just an entertaining tool, creating funny images and cannot be used in work. They also point out that compared to other models, Images in ChatGPT takes more time to process.

Commenting on Images in ChatGPT, Mr. Trung Nguyen, Founder of Advertising Vietnam and Snapcut.ai, an online video editing tool using artificial intelligence, said that this feature is not really new.

“Before OpenAI was launched, image creation specialists did very well, sometimes even better than this version of ChatGPT,” Mr. Trung shared with Tri Thuc - Znews .

However, the big problem with other AI image generation tools is that it takes time, or users have to enter long, detailed commands to get the right image.

Same command, compare image results generated by ChatGPT (left) and Midjourney. Photo: Digidop.

Meanwhile, GPT-4o, which has been integrated by OpenAI in advance, has the ability to understand language and user requests, as well as being more creative than traditional tools. To produce a satisfactory photo, regardless of quality or aesthetics, users only need to enter a concise command.

Mr. Hoang Viet, owner of a wedding photography studio in Dong Nai, said that previous tools were too dependent on powerful hardware (high-end GPUs) with high costs, only suitable for a certain customer base. The results were not as sharp as Midjourney, or as deeply customized as Stable Diffusion, but the photo creation tools on ChatGPT, Gemini or Grok have provided a cheaper, convenient option when editing right on the web or phone.

ChatGPT Limits

OpenAI’s latest update comes on the heels of Gemini Flash and xAI’s Grok Aurora models, which can generate images directly in chat boxes. TechCrunch ’s tests show that ChatGPT’s image-generating AI can handle a wide range of styles, while Gemini and Grok prioritize high-resolution, detailed object processing.

According to Mr. Viet, Images in ChatGPT is well-suited for product photography, and has the ability to restore images that many graphic designers admire. Therefore, it can be predicted that this feature can affect industries such as graphic design, marketing, content writing, and e-commerce.

The photo was created on ChatGPT with the command "Create a realistic close-up photo of a woman standing in a crowd in Times Square, looking at the camera and smiling, with DSLR-like quality. Photo: ZDNet.

The group of people who regularly work with photos are also the ones who are the fastest to catch up with the photo creation feature in ChatGPT. Mr. Hoang Viet's recent post illustrating how to combine wedding photos also attracted attention with more than 1,000 comments and hundreds of shares.

“Currently, it is not yet possible to replace and be 100% accurate, but at this rate, in the near future AI will strongly change the market,” Mr. Viet commented on the potential of the tool.

Generative intelligence is forcing people in all fields to constantly learn to escape old processes and adapt to new technologies. Sam Altman believes that in the near future, ChatGPT will develop the ability to create more content, while improving the latency of Images in ChatGPT.

Source: https://znews.vn/chatgpt-gay-sot-nhung-khong-phai-o-cau-chu-post1541683.html