Hacker News Reader

Comment by airstrike | Hacker News Reader

I'm no image gen expert but these prompts are downright terrible even by my standards.

Are you really complaining that ", from the British Museum." leads to it a painting in the actual British Museum? Just remove the sentence, and you'll be fine. Now good luck trying to make Midjourney place the image at the museum!

I'm a paying MJ user and am impressed by Nano Banana. They're different models. They each serve their purpose.

This analysis is just noise. Yawn.

Ironically, even an LLM with its fake reasoning capabilities can point out the issue with the prompts if you ask it to critique this article.

View on HN

It is interesting what the nbp model takes away from the prompt, though

Eg instead of focusing on the artist, it focuses on the location

This makes sense! I imagine it was trained in some sort of rlvr like way where you give it a prompt and then interrogate "does this image ..." (where each question examines a different aspect of the prompt)

It's obviously an incredible model. I think there's a limit to how useful another article praising it is in contrast with one expressing frustration

I would also welcome someone writing a short takedown where they fix the prompts and get better-than-2022 results from nbp

> I would also welcome someone writing a short takedown where they fix the prompts and get better-than-2022 results from nbp

NBP (and the new ChatGPT generator) are integrated with LLMs to various degrees, so seems like the obvious starting point is a reverse approach: ask them to describe the old images which has the esthetics that Fernando Borretti likes, and start generating from those prompts. If you can recover the old images, then it was just a prompting issue. ("Sampling can show the presence of knowledge but not the absence.") If you can't even with their own 'native' descriptions, then that points to mode-collapse (especially all of the 'esthetic tuning' like DPO everyone does now) as being the biggest problem.

These sorts of prompts used to be quite important when DALL-E was new. I do feel like a lot of the article is just that prompts should be written differently though I think there’s some truth in the idea that nanobanana feels less artistic in some ways.

Reply on news.ycombinator.com