Over the weekend, Elon Musk's artificial intelligence company Xai released Grok Imagine, a new generation AI tool for generating images and videos. Grok Imagine now has paid subscribers available to Xai subscribers for Grok iOS and Android apps.
Musk exaggerates the project with X and shares photos and videos from Grok users. This includes mild NSFW content that the Grok app labels as “spicy.”
This tweet is currently unavailable. It may have been loaded or deleted.
AI Video is an exciting and frankly awful new frontier for the AI industry. For advocates, this technology can give artists a new medium for creativity, reducing the costs of animation and filmmaking. For critics, AI videos pose serious risks to sexual deepfakes and misinformation.
That discussion aside for now, I wanted to see how well Groke is imagination compared to Zai's biggest rival. As I wrote before, Google's VEO 3 AI video model is currently leading the field with surprisingly realistic videos. Then there's Sora from Chatgpt-Maker Openai. Additionally, popular AI image generator Midjourney recently introduced its own generation AI video tool.
So how do you imagine Grok compared to its competitors? To dull it, I am not impressed.
Yes, Grok is a brand new imagination, and Musk recently said with X “it should get better every day.” However, at the time of writing, it appears to be far behind its rivals.
Please show me my work.
Imagine Grok and compare AI videos with competition
Mashable recently wrote about the virus' AI video trend: security camera footage of similar antics on animals jumping to the trampoline. So I used a simple prompt to test Grok Imagine, Veo 3, Sora and Midjourney. “Security camera footage of a rabbit jumping to a trampoline at night.” It's easy isn't it?
First, it should be noted that there is a big difference between Veo 3 and Grok Imagine. Google's VEO 3 model can generate videos based on text prompts. A quick guide to the video you need is Veo 3 doing the rest. However, I imagine that tools like Midjourney and Grok only provide the generation of images from text. After generating or uploading an image, the user can animate it and convert it into a short video clip. In this sense, Grok Imagine is already at the back compared to Openai and Google.
Masculine light speed
With these warnings, dive into the results. I also shared this on X.
I put the test prompt in Grok and returned these disappointing images.
Credit: Courtesy of Screenshot Grok/Timothy Beck Werth
Credit: Courtesy of Screenshot Grok/Timothy Beck Werth
I chose the worst of these images and created this short video.
This tweet is currently unavailable. It may have been loaded or deleted.
Is that okay? A kind of intermediate, or ahas the kids say.
But it suffers compared to other AI video tools.
As the video shows, the Google Veo 3 and Sora were far better at the same prompt.
This tweet is currently unavailable. It may have been loaded or deleted.
Finally, Midjourney, animating animated images similar to Xai, was able to create better images and videos, although it took two attempts. The images and videos it creates have a grainy appearance of surveillance footage.
AI-generated images.
Credit: Timothy Beck Worth/Mid Journey
This tweet is currently unavailable. It may have been loaded or deleted.
Audio is also a major drawback to Grok Imagine. VEO 3 can sync with the video to generate coherent interactions with sound effects, but the audio found in Grok Imagine videos is limited to rough sound effects and meaningful things.
Musk compares Grok with modern Vine apps and writes in X.
And in my first test, it appears that Grok Imagine is optimized to create two different images and videos: memes and anime. If you want to animate memes or create a sexually suggestive video of anime girls, I think Grok Imagine will do the trick. But beyond that, I cannot say I was impressed.
There is one area where Grok Imagine shines, and that's in terms of speed. So far, I have discovered that it generates both images and videos much faster than its rivals.
This tweet is currently unavailable. It may have been loaded or deleted.
Mashable reached out to Xai. If we receive a response, we will update this story.
Disclosure: Mashable's parent company Ziff Davis filed a lawsuit against Openai in April, claiming it infringed Ziff Davis's copyright in training and operating AI systems.
