TE
TechEcho
Home24h TopNewestBestAskShowJobs
GitHubTwitter
Home

TechEcho

A tech news platform built with Next.js, providing global tech news and discussions.

GitHubTwitter

Home

HomeNewestBestAskShowJobs

Resources

HackerNews APIOriginal HackerNewsNext.js

© 2025 TechEcho. All rights reserved.

How Good Is DALL-E Mini at Origami?

51 pointsby mkosmulalmost 3 years ago

11 comments

blagiealmost 3 years ago
One related property of GPT-3: It&#x27;s very bad at traditional computational tasks.<p>* &quot;Make a list of 20 items&quot; results in a list. The number of items is as accurate as if you asked a toddler the same question.<p>* If you ask GPT-3 a simple combinatorics question, it will be 100% confident in the wrong answer.<p>Origami is sort of the same. It takes a conceptual understanding of how paper folds, which DALL-E Mini doesn&#x27;t have. It has a feel for the general origaminess of a picture.<p>If I showed a human being a few pieces of origami, including a paper crane, and they had never seen origami before, they&#x27;d likely result in similar pictures.
评论 #31789138 未加载
hanswordalmost 3 years ago
I know this is a little bit banal, but I feel like: (1) the author is thinking about &quot;origami&quot; (2) the model is only able to create &quot;pictures of origami&quot;<p>The model can only ever be trained on pictures of origami. Thus, the model can generate images that are getting close to &quot;pictures of origami&quot;, but (as pictures necessarily are abstracted 2d projections) this might still be way way way off from &quot;origami&quot;. Not knowing about actual origami, only ever having seen pictures, I thought most of the generated images were quite good. The actual experienced origami-folding person doesn&#x27;t see it that way.<p>I hope my thought is phrased clearly enough, I am having trouble finding the right words here.
mysterydipalmost 3 years ago
Semi related question for those more familiar with current AI capabilities: Has there been any attempt to &quot;see&quot; what dinosaurs looked like from their fossils? Using existing known animals and their skeletons as a training set.
thomalmost 3 years ago
I don&#x27;t mean this to sound overly negative, because I absolutely think DALL-E is a killer app amongst recent AI advances. But the thing that made DALL-E astonishing is that it was... good. While DALL-E Mini mimics a lot of the technical advances and you can kind of see what it&#x27;s getting at with its outputs, they&#x27;re still mostly garbage. Very clever garbage! But they lack the emotional impact that - woah! - this is doing something superhuman.<p>Obviously the hope is that somehow this and future advances can be democratised. It was funny that Asimov&#x27;s The Last Question has been posted here a couple of times recently because it makes such a big thing about world-sized computers and how advanced minicomputers would be. It&#x27;s easy to read and scoff at the naivety... before realising we could easily be heading back in that direction for many impactful future technologies.
评论 #31789894 未加载
Thorentisalmost 3 years ago
Honestly, I thought the images generated were actually pretty good. The shadows of the paper folds, the types of folds typically used. It all felt &quot;close enough&quot; to be very impressive for an AI model.
codemonkey-zetaalmost 3 years ago
Checked the model, and the &quot;model card&quot; <a href="https:&#x2F;&#x2F;huggingface.co&#x2F;dalle-mini&#x2F;dalle-mini#bias" rel="nofollow">https:&#x2F;&#x2F;huggingface.co&#x2F;dalle-mini&#x2F;dalle-mini#bias</a> is an interesting exercise in sensitivity absurdity:<p>&quot;Bias<p>CONTENT WARNING: Readers should be aware this section contains content that is disturbing, offensive, and can propagate historical and current stereotypes.&quot;<p>Spoiler alert, nothing contained in that section requires a warning. It&#x27;s just abstract descriptions of &quot;potential&quot; negative stereotypes in images.<p>&quot;initial testing demonstrates that they may generate images that contain negative stereotypes against minoritized groups&quot;<p>Minoritized is a new word for me. As though minority status is something actively attached to someone. But no duh I can ask dalle to generate &quot;images of klan members at a lynching&quot; or &quot;inner city police brutality&quot; and get negative images.<p>&quot;When the model generates images with people in them, it tends to output people who we perceive to be white, while people of color are underrepresented.&quot;<p>I&#x27;d like to see real testing, because from what I can tell this is not true. Ask for &quot;white people&quot; and you get weird abstract models of white figures. Ask for &quot;black people&quot; and you get beautiful photos of smiling black faces.<p>Is this the kind of exercise AI researchers have to concern themselves with these days?
评论 #31791800 未加载
hgargalmost 3 years ago
Just tried all of the prompts from the OP&#x27;s post on OpenAI&#x27;s DALL-E 2 - <a href="https:&#x2F;&#x2F;harishgarg.com&#x2F;writing&#x2F;generating-origami-images-using-dall-e-prompts&#x2F;" rel="nofollow">https:&#x2F;&#x2F;harishgarg.com&#x2F;writing&#x2F;generating-origami-images-usi...</a><p>DALL-E 2 beats Mini in almost all of them.
desroalmost 3 years ago
Some of the issues seemingly stem from the model&#x27;s either poor or mis-understanding of the input language... I wonder what a fusion of DALL-E + GPT3 or LaMBDA, where the text-based models perform prompt interpretations, would look like.<p>This may be a naïve thought as my understanding of all models mentioned is superficial at best.
评论 #31789201 未加载
bambaxalmost 3 years ago
Slightly OT, although there might be some sort of connection with origami: does anyone know if DALL-E can produce vector images?
评论 #31787839 未加载
IYashaalmost 3 years ago
This is damn scary! In a way that people might actually start using this technology (which does not really know what it&#x27;s doing)...
xwdvalmost 3 years ago
Is DALL-E publicly available or what? How do I work on generating images?
评论 #31789925 未加载