Reporter tested the "Wonderful Duck Camera": Every step has a dissuasion operation... Did the hippocampus and innocent blue panic?, 9.9 yuan to generate photo numbers | Style |... Sea

21 photos, 9.9 yuan. Under normal circumstances, it can generate several high-definition photos in half an hour that "cannot even see the problem with the mother". Are offline photo studios panicking about things like hippocampus and innocent blue?

Actual measurement: "Every step has a dissuasion operation"

This application that can generate various styles of photos is called the Miao Duck Camera. On July 17th, Miaoya announced its official opening to the public. After its launch, Miaoya quickly became popular, with skyrocketing traffic and users waiting longer for images to be generated.

According to the reporter's actual measurement and the introduction of the "Miaoya Camera" WeChat official account, the use of Miaoya has three steps: making a digital avatar, generating a photo, and refining a photo. The first step is to upload a front facing photo and at least 20 upper body photos, and an additional fee of 29.9 yuan is required to truly start making a digital avatar.

After the second step of creating a digital avatar is completed, users can use the "Generate Photos" function to generate photos of their digital avatar in various styles. The third step is the refinement stage. If the user is not satisfied with the generated results, they can also click "I want to be more like me" to regenerate.

▲ Photos generated through actual testing by reporters

Based on actual testing by reporters and feedback from netizens, almost every step of the process of using Miaoyang to generate photos involves persuasion.

For example, in the first step, the user needs to upload a front facing photo and at least 20 upper body photos. "For people like us who don't like taking selfies, it's difficult to gather 21 photos and take them right away. In addition, uploading so many photos, although Miaoya promises to delete them, whether facial information will be leaked has also raised concerns for users." A user who recently experienced Miaoya said.

In addition, due to high traffic, users are currently waiting for a longer time to generate digital avatars. On the morning of the 25th, after uploading 21 photos, the reporter began the process of creating a digital avatar. The page indicated that there were still 537 people working on it, which is expected to take 3 hours and 16 minutes. According to user feedback, if the production is uploaded on the 20th, it will take at least 10 hours to wait.

However, what truly "discourages" some users is still the issue with the quality of generated photos. On social media platforms, many users share photos generated by Miao Ya and have expressed doubts about who this is. "It has to be said that the generated image is quite beautiful, but the more it looks, the less it looks like me, or maybe it looks like me 10 years from now," said user Ye after experiencing it on the 23rd.

One netizen said, "Others' wonderful ducks: the same; my wonderful ducks: it doesn't matter. I lost nine yuan and nine yuan."

20 minutes vs. next day pickup

Ye told reporters that the waiting time for generation is not regular. On the 23rd, she only waited for 20 minutes to complete it, while on the 22nd, she had to wait for more than 5 hours to experience it. "Without patience, I gave up.".

Although some users complain that the "self" generated by Miao Ya is not like themselves, there are more praises on social media platforms. "It's the most effective photo generation tool I've ever experienced. My mom asked me if I went to take photos." "Miao Duck was domesticated very well" "It's my own AI.".

Another user asked: Will the hippocampus and innocent blue really not panic when they see it?

In fact, after Miao Ya became popular among user groups, there was a discussion about whether AI photography would replace offline photo studios.

Ye, who holds the view of "substitution", said, "Offline photo studios sometimes change people's appearance. If you don't pay special attention to the appearance, then the AI photo wins, saving a lot of trouble. It can be generated as quickly as a few tens of minutes."

According to a reporter from a seahorse shop employee in Beijing, it is currently necessary to wait at least one day to take a photo of a person who has undergone personal refinement. "Usually it's picked up every other day, it can't be faster," said the salesperson. Another customer service representative from a store in Beijing said that taking a professional image photo priced at 399 yuan as an example, it usually takes about 3-4 hours to complete the first version of the photo. If there are more photos, it will take even longer.

More users who hold the view of "no replacement" are more concerned about the quality of generated photos. "In addition to beauty, photography also needs to record the state at that time, which is not done well. With the addition of templates, the generated images cannot show age," said one experiencer.

Another experiencer expressed that these generated photos are not easy to use in real life. "As ID photos, I am worried that others may see them as generated, and I am also worried that facial recognition may not pass. As a photo, I am not actually my true self. Currently, I can only post it on my social media for entertainment."

After experiencing it, a computer vision algorithm engineer told reporters that from the generated images, they do not look very similar to themselves and belong to the beauty type. It may have changed the basic structure of the face, which is the distribution of key points. It is preliminarily judged that facial recognition cannot be passed, but further testing is needed.

In the view of Tianhu, the founder of micro media, the current application scenario of Miao Ya is relatively vertical, with only image photos. Although the effect and reputation are good, and it is more convenient and efficient compared to offline shooting, it cannot completely replace the latter.

In response to the voices of netizens, Miaoya once stated on the 17th that the photo effect has not yet reached the ideal state of the team, and will continue to optimize "Titian"'s ability; Miaoya Camera is not the enemy of photographers. We will soon initiate an internal testing collaboration with the photographer invitation system. Through Miaoya, photographers can create their own style and aesthetics into templates, allowing more people to enjoy their creativity.

How long can a wonderful duck be amazing?

Public information shows that Miaoya's main company is Unordered Network Technology Co., Ltd., established in 2006 with a registered capital of 55.2 million US dollars. Its legal representative, Zhang Long, has multiple companies under his name. Zhang Long is also the legal representative, executive director, and general manager of Youku Video Media Technology Co., Ltd; Feng Yunle, the supervisor of Unordered Network Technology Co., Ltd., also serves as a supervisor for multiple Alibaba based companies.

On the 24th, a person related to Alibaba Entertainment told reporters, "Miaoya is a startup project that Alibaba Entertainment has just started investing in."

Although there is currently a high volume of visits and support from major manufacturers, Miaoya still faces the old problem of tool products: how to retain users? How to go further?

According to Tianhu's analysis of reporters, Miaoya's screen scrolling in a short period of time mainly meets the three conditions of "hitting a single explosive point, capturing interest points well, and utilizing social media for dissemination.". "As a C-end tool product, Miaoya's technology threshold is not high, and the scene is not complicated. Miaoya should have undergone a lot of image photo material training in the early stage. From the generated works, the level is very high, but just like ChatGPT's serious and chaotic speech, there may also be distortion, such as facial similarity."

Tianhu said, "How to retain users is a common problem for these types of applications. If we cannot expand the current application scenarios, increase usage frequency, or use the first mover advantage to form our own data or supply chain barriers, it is difficult to retain users. We can only make products that rely on demographic dividends as a 'flash in the pan'."