7 ms·
Show HN: A beetle generator made by machine-learning zoological illustrations
- bayesian_horse 7y agoI wonder if this could be used as an identification tool for beetles like a phantom sketch. You could move some sliders to get it closer to the bug you are thinking of or which you have in front of you. Of course, this method would have to compete with a related model just trying to classify a photo.
- belforn 7y agoit could be used like that. Technically, the model already learned the features of the dataset. (although it was unlabelled). There are some implementations to find features in StyleGAN and enhance/modify those: StyleGan Encoder https://github.com/spiorf/stylegan-encoder https://github.com/spiorf/stylegan-encoder
- deanclatworthy 7y agoThey already have similar apps to this for a range of things. I guess the most popular ones are for plants [1] [1] https://www.picturethisai.com/ https://www.picturethisai.com/
- todd8 7y agoWhat a clever idea! By the way, there are over 400,000 naturally occurring species of beetles. Beetles make up 25% of all animal forms according to Wikipedia.
- onychomys 7y agoThere's been some new work that throws that percentage into doubt: https://www.insidescience.org/news/move-over-beetles-new-champions-diversity-are-parasitic-wasps https://www.insidescience.org/news/move-over-beetles-new-cha...
- mrob 7y agoReminds me of the (possibly apocryphal) Haldane quote, on what nature revealed about the mind of God: "An inordinate fondness for beetles." https://en.wikipedia.org/wiki/J._B._S._Haldane https://en.wikipedia.org/wiki/J._B._S._Haldane
- mc3 7y agoHumans: by mass, a very insignificant species.
- benibela 7y agoNot among land mammals: https://xkcd.com/1338/ https://xkcd.com/1338/
- mensetmanusman 7y agoI had the thought of how obfuscating this technology could be to hide information. Imagine this going off and publishing an unlimited number of images with generated descriptions that are indistinguishable from what a human would do. There would be no way to verify what the correct information is if you are someone just casually searching the Internet. (This could also almost apply for any type of information...)
- belforn 7y ago-> https://thisrentaldoesnotexist.com https://thisrentaldoesnotexist.com All of the dynamic content on each listing is generated via a series of different machine-learned AI models.
- chaosmachine 7y ago"The laundry are converted into a mini central heating for every guest."
- csande17 7y agoYeah, the listings on this page don't quite make it up to a human level of coherence: "Thus serve a 24 hour security pournising. Also hype is requested as much as you please empty and have a prestige less um be restriction, day or night." It's close, though! But maybe not close enough for people to worry -- the last 10% of the security pournising always takes 90% of the development time.
- mipmap04 7y agoI've found that creating Markov chains are more human feeling (but maybe a bit overfit) if I use a n[0]-level deep look ahead for my generators. [0] where n is normally 2 or 3
- Aeolun 7y ago“there will be a BREAKFAST Walk into the block in the room”
- 7y ago
- CSactuary 7y agoTo the creator: since most (all?) of the beetles are symmetrical, couldn’t you generate left halves and then reflect it to create the right half? This could help you prevent asymmetric generations
- dan-robertson 7y agoMost of the images have lighting which isn’t symmetrical and I think that makes it pretty obvious when an image is made by mirroring half a beetle. But maybe there’s some way to deal with that.
- erikpukinskis 7y agoNaively, you could train another network to correct the lighting on a mirrored beetle. Although I’m sure there’s a smarter way.
- lopmotr 7y agoThe network should just learn that they all have symmetry and only encode the unique information. Once you start hand-coding priors like this, where do you stop? Maybe also constrain the range of colors? Size? Other geometric features? Eventually you're just doing old fashioned programming, not ML. And since generating beetle images isn't really the important goal anyway, why would you use dirty tricks to achieve it?
- ArtWomb 7y agoThis is a beautiful write-up. And I don't wish to detract from the author's work. Nature's result side-by-side with that of the Machine. Makes one feel as though we've taken a step backwards from Alan Turing's "The Chemical Basis of Morphogenesis". Consider the "morphogenetic puzzle" of a bi-valved seashell that shuts with perfect water-tight seal. There is a constraint to this design: survival! https://twitter.com/AlainGoriely/status/1207210428344029184 https://twitter.com/AlainGoriely/status/1207210428344029184
- erikpukinskis 7y agoReminds me of video game art: every game has two sets of art, the “real” art which is part of the mechanics of the game world, and the “pretty bits” which dangle off the game objects and try to trick you into believing there’s more to the game world than there is. A lot of gameplay involves testing for this boundary... Trying to figure out whether you can actually do things that are implied by the art. Are there any modern games where 100% of the art exists inside the game world?
- skykooler 7y agoI'd argue that that's definitely the case for Dwarf Fortress.
- l_t 7y agoCheck out Noita -- it's a pixel-art-style game where every pixel is subject to the physics engine. Leads to some cool emergent gameplay.
- nbardy 7y agoDeep Rock galactic. If it's not all game objects its damn close.
- hayksaakian 7y agoMinecraft?
- drdeca 7y ago
- auton1 7y agoReminds me of Dawkins' biomorphs: http://www.emergentmind.com/biomorphs http://www.emergentmind.com/biomorphs
- stcredzero 7y agoFor more info on things like this: https://en.wikipedia.org/wiki/L-system https://en.wikipedia.org/wiki/L-system
- jonplackett 7y agoI thought it was now basically the law that this must be named thisbeetledoesnotexist.com
- belforn 7y agoSomeone already set it up: http://thisbeetledoesnotexist.com/ http://thisbeetledoesnotexist.com/
- jonplackett 7y agohaha, about time!
- nsxwolf 7y agoWhat I never understand about these things is ... what actually does the drawing? The AI decides what the beetle looks like, at what level of abstraction? When/how does it go from beetle idea to pixels? Does this network "know" what the beetle's "leg" is, or does it just "know" "this pixel here should be this color"?
- _fullpint 7y agoMuch closer to the latter. I haven’t read this yet, but it sounds like an encoder model.
- belforn 7y agoCorrect. The machine here doesn’t even know that those are beetles (because nobody told it), it is “just” arranging pixels in a similar manner as the pixels from the source images. It does understand that each generated image must have “legs”, “eyes”, “shells”... and other features that it detected are common in the original images.
- tlack 7y agoFrom what I understand there are two networks in a GAN like this one. One (the discriminator) is trained with a bunch of images showing what beetles can look like. It detects a real or fake image of a beetle. The other (the generator) is just generating images with a convolutional neural network. The generator optimizes itself based on how close it is to passing the discriminators test - that is its "loss function". So over time, the generator gets better and better at making things that look like beetles. The process takes a very long time and is aided by many GPUs (as mentioned in the article)
- slynn12 7y agoVery clever. Do you know if a machine does the drawing? Wasn't totally clear on that bit.
- eskimobloood 7y agoWould be interesting to see the result trained on images from this book[1] about mutation on insects in the chernobyl area. [1]https://www.amazon.com/Heteroptera-Beautiful-Other-Images-Mutating/dp/3908247314 https://www.amazon.com/Heteroptera-Beautiful-Other-Images-Mu... [2]https://atomicphotographers.com/cornelia-hesse-honegger/ https://atomicphotographers.com/cornelia-hesse-honegger/
- jamesfisher 7y agoIs the source/model/network available to run?
- belforn 7y agoyes, it is published. Available via @RunwayML https://open-app.runwayml.com/?model=cunicode/confusing_coleopterists https://open-app.runwayml.com/?model=cunicode/confusing_cole...
- fiter 7y agoWas this model tested for overfitting? I do not have any sense of whether the beetles that I'm seeing match some source pictures exactly. I noticed that the transformations seem to be fast through a transition and then seemingly paused. Is this intentional or does this have something to do with the model?
- enchiridion 7y agoIs there a good way to test a gan for overfitting?
- cortesi 7y agoExcellent. I did something similar a few months ago using a dataset of zoological silhouettes, resulting in a menagerie of mammals, bugs, spiders and other mutant wonders. https://twitter.com/cortesi/status/1153075801891278848 https://twitter.com/cortesi/status/1153075801891278848 https://twitter.com/cortesi/status/1153088629972934656 https://twitter.com/cortesi/status/1153088629972934656
- belforn 7y agonice results! could be interesting to see another network naming the mutant creatures :)
- proc0 7y agoInteresting to see how it handled the legs. Looks like it had problems with the fact insects have 3 pairs of symmetrical legs.