Grok can make images using Flux — 5 examples of it in action
The newest version of the AI chatbot Grok, Grok-2, can now generate images and we of course had to try it out.
Thanks to the FLUX.1 model by Black Forest Labs, Grok-2 (currently in beta) can now create images that rival those generated by the likes of Midjourney — and when it comes to generating AI images of people it may perform even better.
However, one big difference users are pointing out is that they are encountering less of the typical guardrails they’ve gotten used to when toying around with other AI image generators.
FLUX.1 is integrated into Grok-2, in the same way, you can generate images in OpenAI’s ChatGPT and Microsoft’s Copilot using DALL-E 3 or Google Gemini using Imagen. I decided to take it out for a spin and show you the results.
1. Olympic breakdancer
Days after the Paris Olympics, the internet is still mesmerized by the performance of Australian breakdancer Rachael Gunn, known competitively as Raygun. I asked Grok to generate an image of a woman in a green tracksuit performing an airflare breakdance move. Of course I had to include some Olympic rings in the background. My request for adoring fans cheering her on turned into the kind of fans that blow wind. I didn’t specify that the fans had to be human, so fair play to Grok.
2. Skateboarding goldie
While I was already down under I thought I’d stick to the same general theme. Next I wanted a golden retriever wearing a knitted nautical-themed sweater. I asked Grok-2 to have the dog skate along Bondi Beach. I got a pretty standard result – nothing to write back home about but it was all decent enough.
3. Opera house photoshoot
Jumping back to photorealistic portraits of people, I opted for a professional looking photo of a newlywed couple posing outside the Sydney Opera House. I’m happy with the result with a well-dressed couple in an embrace with the iconic opera house clearly recognizable in the distance.
4. Zoo’s roo
No Australian roundup would be complete without an appearance by a kangaroo. I asked Grok for an AI image of a zookeeper feeding a baby kangaroo and the result is as cute as I’d hoped for.
5. Glowing bridge
I wanted to end on a more creative note so I asked Grok to re-imagine the Sydney Harbor Bridge. I wanted the bridge to remain recognizable but requested a cyberpunk style with glowing neon lights. I got the bridge and the lights but was still missing an overall cyberpunk feel.
Final thoughts
Overall the results were what you’d expect to get with a leading AI image generator in the second half of 2024.
On X, Elon Musk said Grok has its own image generation system under development but until it’s completed, users can continue to experiment with the FLUX.1 model.
More from Tom’s Guide
Source link
/cdn.vox-cdn.com/uploads/chorus_asset/file/25628420/Screen_Shot_2024_09_18_at_2.07.32_PM.png)
