Lemon8Lifestyle community
Kelley Dean
Kelley Dean

1312 followers

Follow
Follow
Follow
A woman with long blonde hair and a light sweater is shown, with the text 'ChatGPT VS. GROK' overlaid in a brown, 3D-style font. 'lemon8 @kel.beans' is in the bottom left corner.
A black and white close-up portrait of a woman's face, covered in water droplets, with wet hair and long eyelashes. The text 'ChatGPT' is in a brown cursive font at the bottom right.
A black and white close-up portrait of a woman's face, looking to the side, covered in water droplets, with wet hair and long eyelashes. The word 'GROK' is written vertically in brown cursive on the left.
1/4

You may also like

No content
See more on the app
See more on the app
See more on the app
0 saved

chat v grok

i pasted this into both apps with the first pic. I think Grok looks most like me, but made me low key look like sid the sloth.

Without changing my facial features AT ALL, but just adding lashes, create a high contrast very close up portrait of my face focusing on front head, in black and white closeup, 35mm lens, 4K HD quality, giving proud expression, water droplets on my face black shadow background only face, is visible with my profile looking sharper, adding wet hair, Ratio 4:3

4/2 Edited to

... Read moreOkay, so after my little AI face-off with ChatGPT and Grok, I've had some thoughts about what it really takes to get a good AI-generated portrait. It's not just about typing a few words; it's an art in itself! I've experimented a lot with different AI image generators lately, and honestly, the nuances in prompting can make or break your results. For instance, when I asked for a 'high contrast very close up portrait' with 'water droplets' and 'wet hair,' I was trying to push the AI's understanding of texture and mood. ChatGPT gave me a decent attempt, but Grok, despite the 'Sid the Sloth' comment (don't worry, it was affectionate!), actually captured more of that intense, artistic vibe I was going for, especially with the deep black and white close-up portrait. It really highlighted the dramatic effect of the water and the strands of wet hair. One thing I've noticed across various AI platforms is their struggle with consistency and *subtlety*. You can ask for 'no change to facial features AT ALL,' but sometimes the AI still takes creative liberties. It's like it has its own idea of what a 'proud expression' should look like, and it might subtly tweak features to fit that. My advice? Be super specific, but also be prepared for surprises! Adding details like '35mm lens' or '4K HD quality' definitely helps guide the aesthetic, but the AI's interpretation of those artistic terms can still vary wildly. If you're trying to generate your own AI face, here are a few tips I've picked up. First, really break down what makes a photograph powerful: lighting, angle, expression, and specific textural elements like those 'eyelashes' or 'water droplets.' Second, experiment with negative prompts – telling the AI what not to do can be just as important as telling it what *to do*. For example, 'no cartoonish features' or 'avoid exaggerated expressions.' Third, try iterating. Generate a few options, pick the best one, and then ask the AI to refine it further, adding more detail or adjusting the style. I've also found that the AI's ability to handle specific artistic styles, like a stark 'black and white close-up portrait,' can depend a lot on its training data. Some models excel at photorealism, while others lean towards more stylized interpretations. The key is to keep experimenting and learning what works best for each AI tool. It's a fascinating journey seeing how these algorithms interpret our creative visions, and honestly, even when they get it a little 'off,' it's still pretty amazing to witness!