Hacker News new | ask | show | jobs
by svachalek 23 days ago
I'll give them images. Gemini/Nano Banana is notably better at understanding and generating images than OpenAI, imo, and Claude can't even generate.

While I'm at it I've got to give them credit for Gemma as well. Stellar, first class model for the size.
1 comments
vunderba 23 days ago
It's a toss-up - NB Pro (and NB2) were in the lead for a long time, but gpt-image-1.5 and then particularly gpt-image-2 pretty much closed the gap on my GenAI Image Showdown benchmark site.

NB still generates better looking images though for the most part - gpt-image series is still affected by the yellow saturation issues though its been heavily mitigated.

link
basch 23 days ago
gpt-image-2 still has a very visible banding/artifacting, but nb2 and gpti2 each have their strengths to the point that its momentarily worth running the same prompt through both and grabbing the better result, then feeding it to the opposite model to tweak it. both of them do a better job of not "getting stuck" regurgitating the same thing over and over when the base image comes from the opposite model. i tend to try and limit how many edits i make with gpti2 because successive iterations degrade the image much much faster.

huge waste of firefly credits tho.

link