16 Comments
User's avatar
Frontier Modal's avatar

Saying something that's just different and a little bit better in some ways a KILLER of a competitor is just dumb trolling at this point. It doesn't add anything of value.

Daniel Nest's avatar

If you know anything about my attitude to these hype-trains, you'll know that I'm using the term very intentionally to mock this very topic. So yes, I agree.

Bette A. Ludwig, PhD 🌱's avatar

OMG, Daniel. I love the old school photo!

Daniel Nest's avatar

Which one's that, exactly?

Bette A. Ludwig, PhD 🌱's avatar

Your main pick with the banana and the shadow. I loved it 😊

Daniel Nest's avatar

Ah yes, ironically enough, it was made with GPT-4o image generation and edited with Nano Banana!

Rationaltail's avatar

I used flash image to edit one image and it worked marvelously. It rejected every other attempt siting harmful or explicit content. The content was definitely not harmful or explicit. Any one else had this issue?

Daniel Nest's avatar

Yeah, Google's often weird when it comes to censorship. For what it's worth, I seemingly ran into fewer rejections when using Nano Banana on LM Arena - see if that does the trick for you?

Rationaltail's avatar

Good to know! I’ll try that.

AI Creator House's avatar

Honestly, I’m not super impressed by Nano Banana. I’m just as well off with Hailuo for reworking images.

I have done two AI-powered feature films, many shorts, and I’m currently putting out four web series.

I use it occasionally (like Vidu), but honestly… it’s not nearly as good IMO as other tools out there.

Google and Meta both just use their trillions of dollars to keep telling people they are leading technology, but it doesn’t mean they actually are in reality.

I’m excited about seeing what China produces next.

Daniel Nest's avatar

I always thought of Hailuo as a primarily video model, so I never paid too much attention to their image models, interesting to hear it handles image editing well!

But it's hard to argue with the fact that Veo 3 is a very impressive video model, and the fact that it generates native audio is cherry on top.

Andrew Sniderman 🕷️'s avatar

Not seeddream, but I just had my first go with nano banana. I had annoying people in a pic for Fridays piece and it disappeared them lickety split. Hadn't noticed the Gemini watermark in the corner before. I like that

Daniel Nest's avatar

Yeah Google is leaning heavily into watermarking with SynthID etc. Probably the way forward with AI for sure.

Andrew Smith's avatar

Man, right as I step away from daily image generation (tho I still average at least one per day), stuff gets really good! I'm going to steal the idea for a four-panel explanation about how a process works and see if I can rubber-stamp that onto some of the business processes I'm working on!

Daniel Nest's avatar

I want a commission for my genius and unique idea!

Andrew Smith's avatar

If we could just get a proper system for attribution set up, that would already be taken care of. I feel like "AI fixes this" is the new "the blockchain fixes this."

But really: AI fixes this!