Linh's "father" is Quang Dong, 23, residing in Tay Ho ward, Hanoi.
Dong previously operated an e-commerce store specializing in fashion apparel. One of his persistent challenges was the cost of hiring models to showcase products. "Hiring a model cost several million dong for a single session, but the effectiveness was highly unpredictable. If the model didn't appeal to consumer tastes, the profits wouldn't even cover production costs," Dong explained.
From late 2025, he began experimenting with virtual models, leading to the creation of "Duong Thuy Linh." Dong designed Linh as a young woman with a distinct East Asian face, fair skin, and a "young lady" style, aligning with the fashion line he was selling. Hundreds of drafts were discarded when the AI generated characters with unusual body proportions, oversized heads, or skewed lighting.
![]() |
Mau anh do Quang Dong, 23 tuoi, o Ha Noi tao ra bang cong cu AI. Anh: Nhan vat cung cap
"The biggest challenge was to convince viewers visually, making it impossible for them to realize this was a virtual model," Dong stated. He used the Nano Banana Pro tool to design backgrounds, meticulously detailing even the shadows of objects. When "dressing" the model, Dong input actual product images into the prompt, describing the material and drape of the fabric to ensure the output image was 95% accurate compared to reality. To achieve smooth character movements, he combined two applications: Veo 3 and Kling AI.
After only one month, "Duong Thuy Linh"'s videos garnered millions of views. With an appearance that makes "anything look good," this virtual model secures 100-200 orders daily.
"In the first month, the virtual model generated over 300 million VND in revenue," Dong revealed. Currently, he spends only 5-10 minutes producing a video, a significant reduction from the many hours previously required. This efficiency allows "Duong Thuy Linh" to maintain 15-20 videos daily, resulting in a steady increase in orders.
Mau ao Duong Thuy Linh, so huu gan 50.000 nguoi theo doi do Quang Dong lap ra. Video: Duong Thuy Linh
The trend of "nurturing AI virtual models" is quietly expanding. In Ho Chi Minh City, Nguyen Thanh Nam, 31, chose to train AI to become a key opinion consumer (KOC). As a marketing professional, Nam uses technology to realize filming scenarios that traditional budgets would not permit.
Nam's distinct approach focuses on the "soul," expressions, and voice. "A common beginner's mistake is creating models with stiff faces and uncoordinated mouth movements. I had to refine each prompt for the AI to understand a wink or a smirk," Nam explained.
The realism of his creations leads many viewers to not recognize them as technological products. Numerous brands have commissioned Nam to produce advertising videos, each costing 1-5 million VND.
An entire ecosystem supporting the "virtual model" profession is also emerging. On short video platforms, thousands of virtual model accounts, ranging from children to middle-aged individuals, are appearing, selling everything from functional foods to feng shui items. Services offering model creation coaching for 500,000 VND per hour, or on-demand video production, are flourishing.
Minh Trang, 34, owner of a children's fashion chain in Ho Chi Minh City, stated that she has reduced image costs by 40% thanks to AI models. Previously, photographing child models was extremely challenging due to children easily crying and shooting schedules frequently falling apart. Now, she only needs to send fabric images and designs, and within a day, she receives a complete photo set in any desired setting. The cost for each AI video is only about 100,000 VND.
However, Trang recognized that the biggest barrier remains trust. Many discerning customers request to see regular camera videos to verify material quality, as AI images, while smooth, sometimes feel "too good to be true." AI helps attract initial attention, but to convert customers, she still relies on real staff for live stream consultations.
Nguyen Thanh Nam 31 tuoi, TP HCM chon huong huan luyen mau ao thanh KOC review san pham. Video: Nhan vat cung cap
The demand for learning the "virtual model" profession also surged in early 2026. Duy Anh, an AI trainer in Hanoi, welcomed 150 students in January, double the number from the previous month. Most students are fashion shop owners and KOCs aged 22-45.
"Individuals without graphic design knowledge can produce videos after 3-5 days," Duy Anh said. Students focus on "mapping" techniques (applying real clothing to virtual models) to prevent misrepresentation and reduce customer complaints. He advised users to clearly label AI-generated content and use real voiceovers to maintain engagement. "AI could capture 70% of the general photo model market share. However, this is only a cost-optimization tool. If real product quality is overlooked, brands will quickly be eliminated," he remarked.
From an expert's perspective, Nguyen Bao Trung, Acting Director of the Vision Center, FPT Smart Cloud, FPT Group, noted this trend is inevitable due to the synergy of massive data repositories, powerful AI models, and advancing GPU hardware.
"Vietnam is rapidly emerging in the creative application layer. Young people today do not need significant capital to invest in manufacturing or real estate. They only need to leverage creativity and generative AI models to create valuable digital assets," Trung analyzed.
However, he warned of legal and ethical risks when using others' images without permission to train AI. Technology companies are developing tools to detect AI content and establish standardized data sources.
He also proposed mandatory "AI-generated" labeling and a penalty framework for deceptive practices or image copyright infringement.
"AI will not completely replace humans. To avoid being obsolete, young people need to cultivate storytelling skills rather than just chasing tools," Trung advised.
For Quang Dong, as "Duong Thuy Linh" gained popularity, he faced criticism for deceiving customers. Dong asserted that honesty is crucial: "I always publicly state that this is an AI photo model. The goal is to illustrate the product beautifully, but the core remains that the goods customers receive must match the description."
Despite offering cost and visual advantages, content creators acknowledge AI's limitations. "AI can be more aesthetically pleasing than real people, but it cannot replace actual personal experiences and emotions," Thanh Nam shared.
Nga Thanh
