Show HN: High-Res Neural Cellular Automata • Hacker News Board

Re: Show HN: High-Res Neural Cellular Automata

#31

Post by esychology » Wed, Jun 17, 2026, 2:38 PM UTC

WhitneyLand wrote: post #20

At a glance it looks like it could be just iterative texture sampling. The difference is when creating each pixel, there’s no coordinate to look up, instead it’s using only a set of rules like Conway’s game of life. But the rules come from a neural network trained on the image, so… it’s kind of memorizing enough information to effectively do the same thing as texture sampling, but using only local information. I’m su…

Texture sampling retrieves pixels by coordinate, while NCA grows them from local rules with no global lookup. The weights are actually ~3× smaller than JPEG-compressed texture maps, so it's not just memorizing the image either. The more interesting part is that it's a dynamic process, you can damage the pattern while it forms and it repairs itself.

Re: Show HN: High-Res Neural Cellular Automata

#32

Post by esychology » Wed, Jun 17, 2026, 2:42 PM UTC

embedding-shape wrote: post #7

Really interesting demo, nicely done :) Would be fun if switching the "Target Image" when using the second brush mode in the Growing Demo didn't erase/reset the existing canvas, so we could "stamp" new things on top of other images. Small thing perhaps but I got sad when it disappeared when I wanted to merge a kitten on top of the chameleon but couldn't :(

That's not possible in the current demo but this sounds like an interesting feature to work on and add!

Re: Show HN: High-Res Neural Cellular Automata

#33

Post by bjourne » Wed, Jun 17, 2026, 2:50 PM UTC

I found your previous work here: https://distill.pub/2020/growing-ca/ For someone (like me) who wants to understand the basics its probably better. It's very well written.

Re: Show HN: High-Res Neural Cellular Automata

#34

Post by moralestapia » Wed, Jun 17, 2026, 3:08 PM UTC

@esychology this is phenomenal work, thank you so much for sharing it. I am working in a similar thing and might reach out about it soon.

Also, what's going on? Why would the community flag and kill this comment[1], from the creator itself. If you're jealous of what the guy built, take it elsewhere. HN will implode with that attitude.

1: https://news.ycombinator.com/item?id=48571171

Re: Show HN: High-Res Neural Cellular Automata

#35

Post by esychology » Wed, Jun 17, 2026, 3:13 PM UTC

moralestapia wrote: post #34

@esychology this is phenomenal work, thank you so much for sharing it. I am working in a similar thing and might reach out about it soon. Also, what's going on? Why would the community flag and kill this comment[1], from the creator itself. If you're jealous of what the guy built, take it elsewhere. HN will implode with that attitude. 1: https://news.ycombinator.com/item?id=48571171

Thank you for the kind comment! Please reach out, I'm happy to have a chat.

Re: Show HN: High-Res Neural Cellular Automata

#36

Post by bjourne » Wed, Jun 17, 2026, 3:45 PM UTC

moralestapia wrote: post #34

@esychology this is phenomenal work, thank you so much for sharing it. I am working in a similar thing and might reach out about it soon. Also, what's going on? Why would the community flag and kill this comment[1], from the creator itself. If you're jealous of what the guy built, take it elsewhere. HN will implode with that attitude. 1: https://news.ycombinator.com/item?id=48571171

It's the stupid spam filter going haywire: https://news.ycombinator.com/item?id=48321198

Re: Show HN: High-Res Neural Cellular Automata

#37

Post by menno-sh » Wed, Jun 17, 2026, 3:46 PM UTC

oersted wrote: post #22

WhitneyLand wrote: post #20

At a glance it looks like it could be just iterative texture sampling. The difference is when creating each pixel, there’s no coordinate to look up, instead it’s using only a set of rules like Conway’s game of life. But the rules come from a neural network trained on the image, so… it’s kind of memorizing enough information to effectively do the same thing as texture sampling, but using only local information. I’m su…

To me, it is intriguing as a toy model for how cells are able to grow into complex tissue and organisms based only on local information, and how they are able to repair and recover harmed tissue. Of course, this is as close to cells, as neurons from neural networks are to real neurons. And I have no idea what it could be applied to (inpainting/outpainting?), but it’s interesting as exploratory research.

Oooh, this made it click for me. Thank you

Re: Show HN: High-Res Neural Cellular Automata

#38

Post by esychology » Wed, Jun 17, 2026, 4:16 PM UTC

WhiteNoiz3 wrote: post #29

amelius wrote: post #2

Why are the images always generated in the same orientation (upright)? Do the cells have awareness of what is "up"?

IIRC training starts with the initial state and the end state, and the end state is always oriented the same way. It would be interesting to see what would happen if the end state was rotated randomly though I suspect it wouldn't work so well.

Checkout the Isotropic NCA blog from the Google Zurich team: https://google-research.github.io/self-organising-systems/is...

Re: Show HN: High-Res Neural Cellular Automata

#39

Post by yorwba » Wed, Jun 17, 2026, 4:21 PM UTC

bjourne wrote: post #36

moralestapia wrote: post #34

@esychology this is phenomenal work, thank you so much for sharing it. I am working in a similar thing and might reach out about it soon. Also, what's going on? Why would the community flag and kill this comment[1], from the creator itself. If you're jealous of what the guy built, take it elsewhere. HN will implode with that attitude. 1: https://news.ycombinator.com/item?id=48571171

It's the stupid spam filter going haywire: https://news.ycombinator.com/item?id=48321198

If it says [flagged], a human pressed the "flag" button. Most likely because OP is commenting too much.

Re: Show HN: High-Res Neural Cellular Automata

#40

Post by CamperBob2 » Wed, Jun 17, 2026, 4:44 PM UTC

yorwba wrote: post #39

bjourne wrote: post #36

Earlier quoted context omitted.

It's the stupid spam filter going haywire: https://news.ycombinator.com/item?id=48321198

If it says [flagged], a human pressed the "flag" button. Most likely because OP is commenting too much.

I don't believe that's the case. HN is running some kind of automated slop detector that classifies any suspected LLM output as "slop" even if it's entirely in-context for the conversation (e.g., a conversation about LLM output.) The comment gets autoflagged the moment it's posted. This has been the case for a while now, I've run afoul of it myself.

It sucks, and I can only imagine they implemented it because they felt they had no other option to deal with the onslaught of bots that's appeared lately.

One thing they could do to improve the situation is to weight vouching heavily by karma. I have plenty of karma but I don't think I've ever successfully pulled a comment out of the [dead] state by vouching for it. Legitimate comments that get flagged by mistake tend to stay that way for the duration of the discussion.