Hello everyone, I am Doufu, a post-80s IT professional, former HR at a state-owned bank, and currently an AIGC practitioner, exploring AI video and music.
By the end of 2024, major video models are rushing to release their updates. Sora kicked off a series of release conferences, and other platforms followed suit with new versions and products one after another. Just half a month into 2025, the competition has begun again. A few days ago, Haicui AI launched the ‘Character Reference’ feature, and Vidu has entered the 2.0 era.
In addition to announcing the new Vidu 2.0 model, it also officially launched the Vidu China site[1].
Although Vidu is a product developed in our country, it was initially born at the end of April 2024, just two months after the release of Sora’s ‘teaser’. The domestic video large model market was still blank. In order to align with international standards and establish an international image, the Vidu International site[2] was officially launched in July 2024.
Vidu Update Log
-
• Vidu 1.0: Initial version, supports two core functions: text-to-video and image-to-video, offering two duration choices of 4s and 8s, with a maximum resolution of 1080P. -
• Vidu 1.5: Added capabilities for character consistency, anime style, and text and special effects generation. -
• Vidu 2.0: Generates 4 seconds of 512p resolution (fast mode) in as little as 10 seconds, providing more stable visuals and achieving scene consistency.
The Impossible Triangle of Vidu
The impossible triangle—cheap, fast, high quality. Can we have all three?
Let’s look at the data:
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
Generate in as little as 10 seconds, costing less than 0.3 Yuan per second, and unlimited off-peak generation (flagship version). Isn’t that tempting? Is it possible?
Scene Consistency Testing
Recently, before the release of Vidu 2.0, Doufu encountered a technical bottleneck in AI video production—scene consistency, which means: adjacent shots are images of the same scene but with different angles.
Solution:
-
• Unify the scene description and seed value in the output prompt,disadvantage: infinite drawing attempts may not succeed. -
• Determine the image for a specific scene and angle, and use Runway’s camera techniques to solve the problem of different angles in the same scene; transitioning from close-up to wide shot may require reversal.(Recommended).
From the release of Vidu 2.0 onwards, the above solutions have improved.
Using Vidu 2.0’s Reference Video Generation feature, Doufu provided one image each for the subject and the scene, generating both panoramic and close-up videos (skipping the output image step), which solved both character consistency and scene consistency. One word: Awesome!
Vidu Membership Package Benefits
Is it the lowest cost? Fast and cheap!
Vidu Latest Events
To celebrate the official launch of Vidu AI Vidu 2.0 version and the simultaneous release of the Chinese version, Vidu has initiated a video creation event themed around ‘speed’, inviting users to participate through social media. The event includes multiple reward categories, such as like awards, evaluation awards, creative awards, and newcomer awards, with rewards ranging from points to JD gift cards. Participants need to post their works on social media using specific topics and @ the official account. The event runs from January 15 to January 27, 2025, with results announced on February 11.
For more event details, see the official event page[3]
More Excellent Vidu Material Works
Conclusion
Major AI video platforms have entered a competitive landscape, allowing users to enjoy better products at the same or even lower prices, which is fantastic.
Every time Doufu uses Keling to generate a video, it consumes 35 points, and with the uncertainty of drawing attempts, it feels painful. Now with Vidu 2.0, I can consider giving it a try.
Thank you for reading this far. If you have further discussions or inquiries, feel free to scan the QR code below to add Doufu, and I will give you a free AI gift package, available on a first-come, first-served basis until supplies last.
Recommended Reading
-
• Unveiling the Three Key Prompt Tools for AI Painting -
• Midjourney Treasure Plugins, I Won’t Tell Just Anyone! -
• Over 100,000 Likes! AI Creates Ancient Statue Videos, Teaching You for 0 Cost in 1 Minute! -
• AI Makes Educational Short Videos, 10,000 Likes is No Dream, It’s Hard Not to Go Viral! -
• After the Transformation of Mythical Beasts from the Classic Texts, Everything Can Become a Spirit, This Flood of Traffic Must Be Caught! -
• Using AI to Compose Music After Work, Revealing the Secrets of Earning Extra Income, Learn in 10 Minutes! -
• Jimeng AI: AI Creation for Painting, Video, and Music? It’s All Here! -
• Jay Chou Won’t Release a New Single This Year, But I Premiered My Music Single and MV with AI, Even Beginners Can Learn!
Feel free to click below to follow my public account. Thank you for patiently reading. Remember to like and see!
Reference Links
<span>[1]</span> Vidu China Site: https://www.vidu.cn<span>[2]</span> Vidu International Site: https://vidu.studio<span>[3]</span> Official Event Page: https://www.vidu.cn/activity?key=vidu2.0