Grok 2 — 8 Best Images and Videos Generated So Far

Why Trust Techopedia

AI-generated images are getting better all the time. Grok 2’s release last week has contributed to some of the most convincing synthetic images ever created.

Users have been creating everything from photorealistic images of models and public figures to cartoons and anime-style outputs. Many of these are on par with, if not better than, those produced by leading tools like Midjourney, DALL-E 3, and ImageFX.

Below, we look at some of the top images and videos created with Grok 2 so far. This includes examples that use both Grok 2 and a third-party tool like Runway Gen-3 Alpha or Kling AI.

Key Takeaways

  • Grok 2 excels in creating highly detailed and lifelike AI-generated images and videos, comparable to top-tier tools like DALL-E 3.
  • Examples of Grok 2’s versatility include realistic product ads and animated characters.
  • However, the platform’s lack of content moderation will result in controversial outputs, including deepfakes of public figures in unsettling scenarios.
  • The platform’s ability to generate convincing deepfakes raises serious questions about the need for regulation and protective measures like watermarking.

Grok 2 — 8 Best Images and Videos Generated So Far

8. Modern Viking Woman

Creating AI-generated faces is tough, but in this example, shared by @jeffplinke, Grok 2 does an incredible job. The original image offers a highly detailed depiction of a young blonde woman wearing a hoodie and holding a shield. She also appears to have some kind of weapon — like a warhammer or a battle ax — on her back.

However, this example really comes into its own with the video version of the image. Using Kling AI’s animation capabilities, the user has created a short clip that turns this basic image into a short clip that appears to be of a real woman.

7. Trump, Kamala Harris and Elon Musk Deepfakes

Our next pick, shared by @thedorbrothers, is a little more controversial and displays deepfake images of public figures like President Trump, Kamala Harris, and Elon Musk holding up a local store with a gun and being arrested by police officers.

Advertisements

These examples show how Grok 2 can be used to create highly realistic deepfakes. While this isn’t necessarily a good thing — and goes against the grain of most generative AI tools banning the use of well-known figures — it is still interesting to see how deepfakes can morph public figures into such unusual scenarios in a convincing way.

6. Portrait of a 1987 Woman

Our next example, shared by @PixelWiseAI, shows how Grok 2 can be used to create portraits. In the initial image prompt, the user asked the large language models (LLMs) to create an image of a “person in 1987 subtly incorporated in the background.”

If this result wasn’t impressive enough, the user then combined the output with Runway Gen-3 to add realistic motion to her face and eyes.

At first glance, the video version appears just like a real person. The only real clues that this clip is AI-generated are the unusual appearance of the cheekbones and the movement of the hands.

5. McDonalds and Burger King Ads

Our next example, shared by a user known as @anferneeck, demonstrates how Grok 2 can be used to develop realistic product images. The images above show AI-generated interpretations of meals from McDonalds and Burger King.

These images are so detailed and well put together that they could easily pass for something out of an official marketing campaign from either fast food giant. The only indication the images were fakes was the misspelling of McDonald’s on one of the background cups.

4. Protein Shakes

Following on from our previous example, this next example, taken from a post published by @footballn4life1 shared two images of protein shakes that were created with Grok 2.

Everything in the images would look perfect as part of a digital or print ad, making the shakes look delicious and organizing the background ingredients in an aesthetically pleasing way.

This image could easily be used as part of a menu or an Instagram post to promote a brand’s products. It could also be used as a reference image to help inform a designer’s creation.

3. Cartoon Gecko

This next image, originally shared by @Arteriallion, shows off Grok 2’s ability to create cartoony images. The image depicts a spotted Gecko, and looks incredibly detailed, all the way from the texture of the scales under the chin to the ridges of the eyes, and the form of the toes on the feet.

Above all, this example illustrates how Grok 2 can be used to create cartoon animals that could easily form the basis of a character in a story or marketing campaign.

2. Cave Men

Coming back to reality, our next example, shared by @vincent_serpico provides a photorealistic clip of two cavemen walking through a frozen valley.

Both the cavemen and the environment in the background are incredibly detailed and could easily pass as a real image.

That being said, the unnatural motion of the water do give some indication that this is a synthetic image.

1. Michael Jackson in Cape Town

Finally, in this example a user known as @VinnyLingham shares a highly-convincing black and white deepfake of Michael Jackson and Nelson Mandela at a concert in Cape Town.

This image sticks out because both depictions of Jackson and Mandela appear to be spot on. It also shows that even in black and white, Grok 2 has the potential to be a gamechanger in the text-to-image market.

But, as the creator points out, “History is now so easily rewritten…”

Our First Impressions of Grok 2

Grok 2’s images are incredibly detailed and lifelike. Based on what we’ve seen, they are up to the standard of anything from Runway, Kling AI, ImageFX, or DALL-E 3 — and perhaps even better (particularly if they are combined with Runway or Kling AI).

However, what really caught our attention about Grok 2 was its lack of content moderation guidelines and the sheer variety of shocking images it’s been used to produce so far.

While any model can be jailbroken, we found countless examples of public figures being depicted committing terrorism or placed in derogatory scenarios.

Though there may be an argument for LLM outputs being protected under U.S. law as an extension of free speech, it also doesn’t seem right that political or public figures can have their likeness used in potentially incendiary ways without their consent.

As it stands, Grok 2 poses a big question over whether deepfakes can exist in their current form and whether or not further regulation is needed to encourage watermarking or other protective measures.

The Bottom Line

Grok 2 is a fantastic addition to the image generation market. If there’s any takeaway from this release, it’s that not only are AI-generated images and video getting better — but we are going to need to get much better at identifying and navigating AI-generated content as a society.

FAQs

What is Grok 2?

What can Grok 2 do?

How do I use Grok 2?

Advertisements

Related Reading

Related Terms

Advertisements
Tim Keary
Technology Specialist
Tim Keary
Technology Specialist

Tim Keary is a freelance technology writer and reporter covering AI, cybersecurity, and enterprise technology. Before joining Techopedia full-time in 2023, his work appeared on VentureBeat, Forbes Advisor, and other notable technology platforms, where he covered the latest trends and innovations in technology.

',a='';if(l){t=t.replace('data-lazy-','');t=t.replace('loading="lazy"','');t=t.replace(/