Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Nano Banana Pro

blog.google

Post reply on HN
Print view View original

231–240 of 718 posts

Re: Nano Banana Pro

#231

Post by fwip »

famouswaffles wrote: post #136
aqme28 wrote: post #79

Earlier quoted context omitted.

I don't think it will be easy to just remove it. It's built into the image and thus won't be the same every time. Plus, any service good at reverse-image search (like Google) can basically apply that to determine whether they generated it. There will always be a way to defeat anything, but I don't see why this won't work for like 90% of cases.

It's an image. There's simply no way to add a watermark to an image that's both imperceptible to the user and non-trivial to remove. You'd have to pick one of those options.

I'm not sure that's correct. I'm not an expert, but there's a lot of literature on digital watermarks that are robust to manipulation.

It may be easier if you have an oracle on your end to say "yes, this image has/does not have the watermark," which could be the case for some proposed implementations of an AI watermark. (Often the use-case for digital watermarks assumes that the watermarker keeps the evaluation tool secret - this lets them find, e.g, people who leak early screenings of movies.)

Re: Nano Banana Pro

#232

Post by esafak »

jasonjmcghee wrote: post #90

Maybe I'm an obscure case, but I'm just not sure what I'd use an image generation model for. For people that use them (regularly or not), what do you use them for?

I'm creating a team T-shirt from a bunch of kids drawings. The model has synthesize a bunch of disparate drawings into a cohesive concept, incorporate the team's name in the appropriate color and font, and make it simple enough for a T-shirt.

Re: Nano Banana Pro

#233

Post by vunderba »

minimaxir wrote: post #149

I...worked on the detailed Nano Banana prompt engineering analysis for months ( https://news.ycombinator.com/item?id=45917875 )...and...Google just...Google released a new version. Nano Banana Pro should work with my gemimg package ( https://github.com/minimaxir/gemimg ) without pushing a new version by passing: g = GemImg(model="gemini-3-pro-image-preview") I'll add the new output resolutions and other features ASAP…

> The model generates up to two interim images to test composition and logic. The last image within Thinking is also the final rendered image.

I've been using a bespoke Generative Model -> VLM Validator -> LLM Prompt Modifier REPL as part of my benchmarks for a while now so I'd be curious to see how this stacks up. From some preliminary testing (9 pointed star, 5 leaf clover, etc) - NB Pro seems slightly better than NB though it still seems to get them wrong. It's hard to tell what's happening under the covers.

Re: Nano Banana Pro

#234

Post by minimaxir »

Fiveplus wrote: post #224

What can nano-banana do that chatGPT made images can't? Or is it only better for image editing from what I can gather from these comments so far. I haven't used it so genuinely curious.

I made some direct comparisons my Nano Banana post (https://news.ycombinator.com/item?id=45917875) but Nano Banana can handle photorealistic photos with nuanced prompts much better. And there is no yellow filter.

Re: Nano Banana Pro

#235

Post by ceejayoz »

cj wrote: post #82
evrenesat wrote: post #73

I've tried to repaint the exterior of my house. More than 20 times with very detailed prompts. I even tried to optimize it with Claude. No matter what, every time it added one, two or three extra windows to the same wall.

I tried this in AI studio just now with nano banana. Results: https://imgur.com/a/9II0Aip The white house was the original (random photo from Google). The prompt was "What paint color would look nice? Paint the house."

> The prompt was "What paint color would look nice? Paint the house."

At some point, this is probably gonna result in you coming home to a painted house and a big bill, lol.

Re: Nano Banana Pro

#236

Post by pigpop »

rcarr wrote: post #141
aqme28 wrote: post #79

Earlier quoted context omitted.

I don't think it will be easy to just remove it. It's built into the image and thus won't be the same every time. Plus, any service good at reverse-image search (like Google) can basically apply that to determine whether they generated it. There will always be a way to defeat anything, but I don't see why this won't work for like 90% of cases.

You could probably just stick your image in another model or tool that didn't watermark and have it regenerate the image as accurately as possible.

Exactly, a diffusion model can denoise the watermark out of the image. If you wanted to be doubly sure you could add noise first and then denoise which should completely overwrite any encoded data. Those are trivial operations so it would be easy to create a tool or service explicitly for that purpose.

Re: Nano Banana Pro

#237

Post by gpmcadam »

arresin wrote: post #197
simonw wrote: post #193

This thing's ability to produce entire infographics from a short prompt is really impressive, especially since it can run extra Google searches first. I tried this prompt: Infographic explaining how the Datasette open source project works Here's the result: https://simonwillison.net/2025/Nov/20/nano-banana-pro/#creat...

Is the infographic accurate in terms of the way datasette wprks?

None of it was accurate.

But boy was it beautiful.

Re: Nano Banana Pro

#238

Post by embedding-shape »

I tried the same prompt as one of the examples (https://i.imgur.com/iQTPJzz.png), in the two ways they say you can run it, via Google Gemini and Google AI Studio (I suppose they're different somehow?). The prompt was "Create an infographic that shows hot to make elaichi chai" and Google Gemini created a infographic (https://i.imgur.com/aXlRzTR.png), but it was all different from what the example showed. Google AI Studio instead created a interactive website, again with different directions: https://i.imgur.com/OjBKTkJ.png

There is not a single mention about accuracy, risks or anything else in the blogpost, just how awesome the thing is. It's clearly not meant to be reliable just yet, but not making this clear up front. Isn't this almost intentionally misleading people, something that should be illegal?

Re: Nano Banana Pro

#240

Post by psygn89 »

sd9 wrote: post #196

It's crazy how good these models are at text now. Remember when text was literally impossible? Now the models can diagetically render any text. It's so good now that it seems like a weird blip that it _wasn't_ possible before. Not to mention all the other stuff.

I agree, it's improving by leaps. I'm still patiently awaiting for my niche use of creating new icons though, one that can match the existing curvature, weight, spacing, and balance. It seems AI is struggling in the overlap of visuals code, or perhaps there's less business incentive to train on that front. I know the pelican on bicycle svg is getting better, but still really rough looking and hard to modify with prompt versus just spending some time upfront to do it yourself in an editor.
Post reply on HN