Unifying characters and changing scenes, PixVerse, a video generation artifact, has been played out by netizens, and its super consistency has become a "killer move"

Another double click is the debut of a new feature.

Have you ever encountered someone who wants to change the background of a character in a picture, but the AI ​​always comes up with "the object is neither the person nor the object" Effect. Even in mature generation tools such as Midjourney and DALL・E, some prompt techniques are required to maintain character consistency, otherwise the characters will change here and there and you will not be able to achieve the results you want.

However, this time it’s your chance.

AIGC Tools PixVerse's new "Character - Video" feature can help you achieve all this. Not only that, it can generate dynamic videos to make your characters more vivid.

Input a picture, and you will be able to get the corresponding dynamic video results. On the basis of maintaining the consistency of the character, the rich background elements and character dynamics make the generated results easy to see. It looks very agile.

Let’s take a look at the effect first.

We can see from these examples that the faces of the characters generated by this tool are basically consistent, accompanied by vivid expressions, and Their backgrounds are richer. This helps users to resolve the complicated steps of re-creation to a great extent. Now, you can update it with one click and complete new works effortlessly.

Netizen @heyronir showed off his generation effect. He first entered the following picture:

Here are the results he got:

                                                                                                  ‐   ‐ ‐ ‐ Result 2Comparing the input and generated results, we can see that the character in Result 1 has face shape, facial proportions, The bone appearance is consistent with the original material, but the eye sockets are slightly deeper and the hair length is slightly shorter. Overall, it appears to be little different from the material originally provided. However, the difference between the generated result 2 and the original material is more obvious. The character's face shape, facial bone direction, pupil color, and hairstyle are all very different.

However, this has a certain correlation with the prompt entered by the user. We can continue to debug in the tool to achieve the desired effect. If you are interested, go and try it.

Experience address: https://app.pixverse.ai/login

This tool is also very easy to use. Simple, specifically it only takes three steps to complete.

Step 1: Find the "Character - Video" function

From the app or Start with the home page of the website. Look for the option labeled "Character," click on it to continue, then select Create Character.

Step 2: Create your character

Now you can upload the image you want to use as your character's face. Give your character a name, create it, and wait about 20 seconds for PixVerse to finish training.

Step 3: Make a Video

Click "Create" and select from the menu Choose a role. Enter the required prompt, select the aspect ratio of the generated video, click Create, and in a few minutes, the character video will be created.

Let’s take a look at other results generated by netizens to see if they will bring you some new inspiration:

#In fact, this is not the first time that PixVerse has been exposed to everyone’s vision and attracted widespread attention. It is a video generation tool launched by Aishi Technology. Just enter text and PixVerse will generate a video based on it.

In the PixVerse Discord channel, it will also give four options for each text. You can slowly select among the options and constantly adjust the generation direction. You can even choose the weather based on its screen, with sunny, rainy, day, and night options available.

In February of this year, Aishi Technology王长虎 was interviewed by this site. He said that PixVerse's performance has reached the level of Pika in some aspects, and even surpassed them in many reviews.

In the comparative evaluation of PixVerse, Pika and Runway, self-media KOL Guizang said: PixVerse’s model is the most balanced among the three and can have a relatively strong range of motion. , while maintaining good consistency.

https://weibo.com/6182606334/499416940252

This competes for the generation AI track. , each tool is seeking its own new way out. Apparently the skill used by PixVerse this time is "Consistency". ###

The above is the detailed content of Unifying characters and changing scenes, PixVerse, a video generation artifact, has been played out by netizens, and its super consistency has become a "killer move". For more information, please follow other related articles on the PHP Chinese website!