Pika 1.0 has just come out, and it’s a serious competitor to Runway ML, another text-to-video and image-to-video platform with so-so quality. I have tested the Pika AI video to its maximum so that you don’t have to, and I want to tell you that it could be a great tool in the future. But it’s not the one yet.

When you go to their website, you can use either text, image, video, or a combination of text and an image or video as your input. But the fantastic thing is that you can use a negative prompt to exclude specific elements from your generation. You can also use a “seed” value for consistent characters or settings in subsequent generations.

Before moving on to the verdict, I want to share how my experimentation with this tool went. (You can check and edit the demo prompts by clicking the “Explore’ button on the website.)

Text-to-Video

Prompt: Franz Kafka listening to a little girl in a park

Output:

Prompt: Franz Kafka talking to a little girl in a park.

Output:

Prompt (using the same seed as in the first video): Franz Kafka talking to a little girl in a park.

Output:

I tried multiple times with images and videos accompanied by text, but the results weren’t up to par.

Verdict

So, is the Pika AI Video any good?

The short answer is: not yet.

However, this is their first version and has only recently come out of beta. So who knows if it will become the number one AI video tool soon. However, this is not the case so far, and it’d be silly to use it for anything except fun purposes.

Alternatives?

Many people will tell you to use Runway ML instead. But it also sucks.

The only powerful alternative I can think of is Assistive Video.

Why am I saying it?

The answer is that it’s currently the only text-to-video tool with results up to par with Google’s VideoPoet (which hasn’t been made public yet.) Check it for yourself.