This is an AI image generation comparison for
text-to-image 
prompt:
Will Smith meme reads "I hate spaghetti", as he is shown screaming and throwing a bowl with spaghetti back at the viewer refusing to eat it. Bowl and spaghetti along with tomato sauce and meatballs are captured mid flight in the mid-background and foregraund. Dramatic, epic, comically exagerated...
Log in to see full prompt.
Tested: April 26, 2025
ChatGPT's new image model can make memes straight up. Here's mine.
Tested: April 11, 2025
Celebrity image and text generated by the Dev model from Huggingface space
Tested: April 24, 2025
Dreamina Seedream 3.0 doing a meme with text. In all 4 generated images it kept the taxt in the exact case it was supplied, without inferring that a meme usually means all caps, so keep that in mind if you want all caps - enter that in your quotes.
Tested: April 26, 2025
"I HATE SPAGHETTI" supplied in all caps
Tested: August 7, 2025
Generated in cartoonish style since there was no style reference, but overall pretty great. I'll do photorealistic next
Tested: August 7, 2025
This is a great one. Add 'photorealistic, cinematic' keywords to ensure the realistic style
Tested: October 17, 2025
Tested: October 17, 2025
Makes very lively images for this scene, text isn't correct though
Tested: November 26, 2025
It's uncensored regarding celebrities. For this prompt, it refused to make the subject face the viewer even when specifically asked, he's looking at the bowl
Check out the results from ChatGPT's Image Generator by OpenAI (GPT‑4o) vs HiDream-I1 vs Dreamina AI (Seedream 3.0) vs Reve vs Qwen Image & Video Generator (Qwen-Image) vs Qwen Image & Video Generator (Qwen-Image) vs Reve (Reve Image 1.0) vs Fal AI (HunyuanImage 3.0) vs Hugging Face (Z-Image Turbo) for similar or identical prompts side-by-side.
