AB
AiBoss
Tutorials

How to create AI-powered product promotion videos using Sora: Two steps to solve the issue of consistency between the performer and the product.

Tutorials

How to create AI-powered product promotion videos using Sora: Two steps to solve the issue of consistency between the performer and the product.

I recently discovered an amazing AI-powered approach to creating creative ads. I've been using Sora 2 for the past couple of days and I'm blown away by its ad creativity! Just input an ad copy or a rough idea, and in just a few minutes, it can automatically generate ads with voiceovers...

I recently discovered something amazingAICreative advertising production ideas.

Experience these two days Sora 2. I was amazed by its advertising creativity! Just input an advertising slogan or a general idea, and it can generate an ad in just a few minutes.automaticGenerate a complete video with voice-over, camera movement, and transitions.

However, there is still a small problem—Sora 2. Uploading real photos of people is not supported. If you want the video to be a little longer, the consistency of the characters cannot be guaranteed.

The method K-sister shared today can perfectly solve the problem of "role inconsistency":

The whole process only requires 2 steps:

1. Send product photos and a prompt to... Sora 2. Generate video;

2. Replace the people in the video with Wan2.2 Animate to ensure consistency throughout;

Let's try it out together.

We upload product images, using... Sora 2. Generate some ad clips.

Prompt wordsUGC ad video. No text overlay. A girl holding the iron pot shown in the picture talks to the audience. The entire video is narrated in Chinese.

The content can be segmented with different focuses. For example, the first segment can focus on the materials, the second on the experience, and the third on the perks. Combining these segments in this way will make the content more comprehensive.

We import the video into CapCut, select the first frame where the person appears, and export the still frame.

When replacing people in a video, it will be more effective to upload a model image with proportions similar to the person in the first frame of the video.

Every time a character changes to a different scene, we need to export the first still frame and replace the character in the image.

Upload the video stills and model photos together to Jimeng to get the replaced scene image.

Prompt wordsThe girl in Figure 1 has her face, hairstyle, and clothes changed to those in Figure 2 (air bangs, short sleeves), but the girl's pose in Figure 1 remains the same.

Open the Tongyi Wanxiang official website, select 【Digital Human】【Character Replacement】. Upload the original video and the image with the replaced character, then click Generate.

This resulted in the replaced video clip, with only minor changes to the characters' appearances. The scene and characters' movements were essentially the same as the original video.

After replacing all the video footage with the characters, you can stitch them together and export them in CapCut. Let's take a look at the final result:

Sora 2. Although the realism is very strong, the script design, camera movement and transitions are very good. Wan 2.2 Animate was used to make up for the lack of consistency in the characters. One person can make a professional advertising video with almost zero cost.

Sora The application scenarios for 2 will also be greatly expanded, from abstract and light entertainment to real advertising, promotional videos and even short drama creation.

This is alsoAIThis is a key step for video tools to move from toys to productivity. In the future, advertising companies, MCN agencies, and even film and television production will increasingly rely on this technology to improve efficiency and reduce costs, rather than remaining at the proof-of-concept stage.