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

Show HN: isometric.nyc – giant isometric pixel art map of NYC

cannoneyed.com

Post reply on HN
Print view View original

41–50 of 247 posts

Re: Show HN: isometric.nyc – giant isometric pixel art map of NYC

#41

Post by NelsonMinar »

jesse__ wrote: post #13

> Slop vs. Art > If you can push a button and get content, then that content is a commodity. Its value is next to zero. > Counterintuitively, that’s my biggest reason to be optimistic about AI and creativity. When hard parts become easy, the differentiator becomes love. Love that. I've been struggling to succinctly put that feeling into words, bravo.

I agree this is the interesting part of the project. I was disappointed when I realized this art was AI generated - I love isometric handdrawn art and respect the craft. But after reading the creator's description of their thoughtful use of generative AI, I appreciated their result more.

Re: Show HN: isometric.nyc – giant isometric pixel art map of NYC

#42

Post by NAR8789 »

tptacek wrote: post #19

So, wait: this is just based on taking the 40 best/most consistent Nano Banana outputs for a prompt to do pixel-art versions of isometric map tiles? That's all it takes to finetune Qwen to reliably generate tiles in exactly the same style? Also, does someone have an intuition for how the "masking" process worked here to generate seamless tiles? I sort of grok it but not totally.

I think the core idea in "masking" is to provide adjacent pixel art tiles as part of the input when rendering a new tile from photo reference. So part of the input is literal boundary conditions on the output for the new tile.

Reference image from the article: https://cannoneyed.com/img/projects/isometric-nyc/training_d...

You have to zoom in, but here the inputs on the left are mixed pixel art / photo textures. The outputs on the right are seamless pixel art.

Later on he talks about 2x2 squares of four tiles each as input and having trouble automating input selection to avoid seams. So with his 512x512 tiles, he's actually sending in 1024x1024 inputs. You can avoid seams if every new tile can "see" all its already-generated neighbors.

You get a seam if you generate a new tile next to an old tile but that old tile is not input to the infill agorithm. The new tile can't see that boundary, and the style will probably not match.

Re: Show HN: isometric.nyc – giant isometric pixel art map of NYC

#45

Post by cannoneyed »

bigwheels wrote: post #7

Very impressive result! are you taking requests for the next ones? SF :D Tokyo :D Paris :D Milan :D Rome :D Sydney :D Oh man...

Really want to do SF next. Maybe the next gen of models will be reliable enough to automate it but this took WAY too much manual labor for a working man. I’ll get the code up soon if people wanna fork it!

Re: Show HN: isometric.nyc – giant isometric pixel art map of NYC

#46

Post by ivangelion »

Want to thank you for taking the time to write up the process.

I know you'll get flak for the agentic coding, but I think it's really awesome you were able to realize an idea that otherwise would've remained relegated to "you know what'd be cool.." territory. Also, just because the activation energy to execute a project like this is lower doesn't mean the creative ceiling isn't just as high as before.

Re: Show HN: isometric.nyc – giant isometric pixel art map of NYC

#47

Post by cannoneyed »

NAR8789 wrote: post #42
tptacek wrote: post #19

So, wait: this is just based on taking the 40 best/most consistent Nano Banana outputs for a prompt to do pixel-art versions of isometric map tiles? That's all it takes to finetune Qwen to reliably generate tiles in exactly the same style? Also, does someone have an intuition for how the "masking" process worked here to generate seamless tiles? I sort of grok it but not totally.

I think the core idea in "masking" is to provide adjacent pixel art tiles as part of the input when rendering a new tile from photo reference. So part of the input is literal boundary conditions on the output for the new tile. Reference image from the article: https://cannoneyed.com/img/projects/isometric-nyc/training_d... You have to zoom in, but here the inputs on the left are mixed pixel art / photo textures. The…

That’s exactly right - the fine tuned Qwen model was able to generate seamless pixels most of the time, but you can find lots of places around the map where it failed.

More interestingly, not even the biggest smartest image models can tell if a seam exists or not (likely due to the way they represent image tokens internally)

Re: Show HN: isometric.nyc – giant isometric pixel art map of NYC

#48

Post by blintz »

cannoneyed wrote: post #36
blintz wrote: post #26

I was most surprised by the fact that it only took 40 examples for a Qwen finetune to match the style and quality of (interactively tuned) Nano Banana. Certainly the end result does not look like the stock output of open-source image generation models. I wonder if for almost any bulk inference / generation task, it will generally be dramatically cheaper to (use fancy expensive model to generate examples, perhaps inte…

In my experience image models are very "thirsty" and can often learn the overall style of an image from far fewer models. Even Qwen is a HUGE model relatively speaking. Interestingly enough, the model could NOT learn how to reliably generate trees or water no matter how much data and/or strategies I threw at it... This to me is the big failure mode of fine-tuning - it's practically impossible to understand what will…

I see, yeah, I can see how if it's like 100% matching some parts of the style, but then failing completely on other parts, it's a huge pain to deal with. I wonder if a bigger model could loop here - like, have GPT 5.2 compare the fine-tune output and the Nano Banana output, notice that trees + water are bad, select more examples to fine-tune on, and the retry. Perhaps noticing that the trees and water are missing or bad is a more human judgement, though.
Post reply on HN