Let’s take a look at the three major AI video projects that have hit the screen recently - the future is beginning to accelerate.

Original text: Let’s take a look at the three major AI video projects that have hit the screen recently - the future that is beginning to accelerate|crazy|animation|byte|ai|static_NetEase subscription< /span>

After PIKA became popular, AI video projects began to explode.

Whether in academia or in the product world, they all want to launch their own things when people focus on AI videos.

There are also those who cut leeks, but there are more things that are truly awesome academically.

There are three projects that have impressed me deeply and are also popular in my group and circle of friends:

Ali's Animate Anyone

ByteDance’s MagicAnimate

Microsoft GAIA

Both Byte and Alibaba are AI animation projects, purely competing products, which will be discussed in detail later. Microsoft's GAIA is where photos speak.

One by one they said:

1. Ari Animate Anyone

Alibaba Research Institute launched an AI animation project last Thursday. A static picture of a character + a skeletal animation can generate a character animation.

I'll just put a picture of this thing, it's more intuitive than text.

I don’t even need to describe too much in words. You should be able to see how much impact it has on the fields of animation, film and television, and games, right?

Especially the last picture. Normally, if you want to make a three-dimensional character's movements, you need to tie the bones and brush the weights, which is a lot of disgusting work. Now you just pull a skeleton for K frames, render a T-pose static image of the character, use AI, and bang, it’s done. . .

Can you imagine how much cost reduction and efficiency improvement it can bring to these industries?

Moreover, Animate Anyone performs extremely well in terms of character consistency and picture stability.

It should be obvious that this stability is quite outrageous.

but. Attention, but coming.

Ali him. . . No project demo or code is provided. . .

He always does this, throwing out publicity. . . Then there is no more. . .

So even though the screen is full, you can just take a look. If you are interested, you can bookmark their Github and keep track of it. What if this product is released?

Address: https://github.com/HumanAIGC/AnimateAnyone

2. DictionaryMagicAnimate

Just by looking at the name, you can tell that it is a competing product of Animate Anyone, right?

And the most exciting thing about this thing is:

He released the demo!

You read that right, damn, he released the demo and code!

Byte is beating Alibaba, hey, you put out a promotional video, right? Then I will let everyone start using it directly, hey, I will disgust you.

Of course, those who enjoy the benefits must be the users.

To be honest, the effect of this thing is quite outrageous.

It supports multiple people, but what can I say? There are big problems with faces and hands, and the effect is not as good as Ali’s AnimateAnyone.

Here’s a little more about netizens’ tidbits:

The advantage is that the code is released, and there may be experts who can modify it and you can experience it at the same time.

MagicAnimate的地址:https://github.com/magic-research/magic-animate

Online test address on Huggingface: https://huggingface.co/spaces/zcxu-eric/magicanimate

3. 小软GAIA

A quite surprising project last Friday, something similar to Heygen, Heygen, and D-ID, that lets "photos speak".

But the overall effect is stunning. Post the picture directly.

For example, direct photo + voice drive.

Then there's the direct photo + video driver.

They also support controlling head posture with hand-crafted poses or poses extracted from another video, while lip movements are synchronized with the speech content.

Next is the coolest part, he actually supports text to generate facial emotions!

Text prompts such as "sad", "open mouth" or "surprised" guide video generation.

sad:

surprise:

Pretty outrageous effect. You can imagine how outrageous this technology will be used in future application scenarios in movies, TV shows, short plays, etc.

Of course, there are scams.

This is the most awesome photo-talking project I have seen so far. But like Ali, the current drawback is: non-publicity.

But it also says on the project's homepage: Code (Coming Soon), which means the code will be released soon. It is expected to be available for trial soon.

Project address: https://microsoft.github.io/GAIA/

As an AI enthusiast and creator.

This should be the first time we see an extremely dense spout of AI video projects, and the approaching future is accelerating.

One day changes, this is really not empty talk.

When you look at AI, you can feel how much this industry has changed.

You can already see some prototypes of the connected future, that crazy future.

Don't be left behind by the times.

Then go back to the past.

There's no harm in looking around.

Now that you’ve read this, if you think it’s good, feel free to give it a like, watch it, and retweet it three times. If you want to receive notifications as soon as possible, you can also give me a star⭐~ Thank you for reading my article.

Guess you like

Origin blog.csdn.net/javastart/article/details/134835958