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.

Show HN: Playing Telephone with GPT-4 and Dall-E-3

34 pointsby elijahbenizzyover 1 year ago
Hello Hacker News!<p>I’ve been having some fun recently and I wanted to share with you. My idea was to feed ChatGPT an image and ask for a caption, feed that back to DallE, then call this in a loop, and observe how the image that was generated changed over time. Similar to playing the game “telephone”. The results were really intriguing, so I built something that you can play with.<p>Note this is best on a desktop (streamlit is optimized for a large screen), but if you’re on mobile you’ll want to expand the sidebar to start — it’s the carrot on top.<p>My initial intention was for you to be able to submit seed images on the site and watch them progress, but that was too expensive and slow. So instead I built an explorer. While this also illustrates a framework I’m developing that helps structure code, it should be interesting regardless of whether or not you’re looking for new tools.<p>I’m happy to add more images — feel free to suggest starting ones you’d like to see!

10 comments

mcbritover 1 year ago
This is funny and interesting AND GPT4 gives such a bad photo for the prompt.<p>Is this because GPT4 is out of a highly stylized space? I really haven&#x27;t played with it. My objections include: water droplets on the mug, bad framing, B- latte art, mug looks basic (and the handle has geometry that makes you question geometry), table surface looks sus, unidentified brown splotches on the table, what&#x27;s going on with the chairs, what&#x27;s up with the floor, what&#x27;s the small bright vertical plane that descends from halfway in the plane about, and so on. I don&#x27;t know if the caption is random, but here is the caption I&#x27;m responding to:<p><pre><code> An exquisitely detailed photograph captures the essence of a calming coffee break. In the foreground, the star of the scene is a matte black ceramic mug with a lustrous outer finish that gently reflects the ambient light, its thick walls hinting at the mug&#x27;s thermal-retentive quality. Nestled inside the rim, a creamy surface of latte art presents a delicate feathered pattern, comprised of swirling strokes with varying shades ranging from the rich dark brown of the freshly brewed coffee to the light beige of the perfectly frothed milk. Each line in the pattern is smooth and purposeful, evidence of a skilled pour, and comes to a focal point with a singular, tiny coffee bubble, an accidental yet charming centerpiece. A soft shadow wraps around the mug&#x27;s rim, adding depth and dimension, while the subtle sheen on the milk&#x27;s frothy canvas catches the light, drawing the eye to the textural contrasts. In the softly blurred background, everyday domestic tranquility is hinted at—a pristine white countertop underpinning the scene, a red-rimmed napkin holder casually placed to the left hints at a touch of color and homeliness. Behind, the clinical metallic sheen of a laptop lies closed, inviting its user to savor the moment of respite it offers. The space is intimately peaceful, a gentle invitation to pause and enjoy the craftsmanship and simple pleasure of a well-made cup of coffee.</code></pre>
评论 #38712989 未加载
tivertover 1 year ago
This one is kind of interesting:<p>It very quickly goes to nuclear cooling towers; then a weird nuclear cooling towers&#x2F;Thomas Kincaid mashup; then Thomas Kincaid cabin; then Orthodox church; then Orthodox church, ice-palace edition; then finally cathedral interior.<p><a href="https:&#x2F;&#x2F;image-telephone.streamlit.app&#x2F;?seed_image=first_photo_ever_taken_20231216" rel="nofollow noreferrer">https:&#x2F;&#x2F;image-telephone.streamlit.app&#x2F;?seed_image=first_phot...</a>
评论 #38710687 未加载
temp00345over 1 year ago
I like how the images seem to go towards magical, mystical and psychedelic with more iterations. Mushrooms just start appearing out of the blue..
评论 #38712996 未加载
xp84over 1 year ago
Oh, my goodness. DALL-E really loves bombing cities. Both the Golden Gate and the &quot;DAG Diagram&quot; starting image turn into disturbing mushroom-cloud series.<p>It looks like &quot;Sun centered over city -&gt; Eye of Sauron -&gt; Energy Beam of Destruction -&gt; Mushroom Cloud&quot; is a reliable pathway.<p>Oh well, at least it didn&#x27;t blow up Royce [Taj Ma]Hall.
评论 #38729211 未加载
mkgeorge7over 1 year ago
I am VERY curious to see how this recursion plays out, but I&#x27;m seeing the below error.<p>RuntimeError: This app has encountered an error. The original error message is redacted to prevent data leaks. Full error details have been recorded in the logs (if you&#x27;re on Streamlit Cloud, click on &#x27;Manage app&#x27; in the lower right of your app).
评论 #38710641 未加载
joshstrangeover 1 year ago
There was a previous HN discussion [0] about a similar service named Dall-E Party [1]<p>[0] <a href="https:&#x2F;&#x2F;news.ycombinator.com&#x2F;item?id=38432486">https:&#x2F;&#x2F;news.ycombinator.com&#x2F;item?id=38432486</a><p>[1] <a href="https:&#x2F;&#x2F;dalle.party&#x2F;" rel="nofollow noreferrer">https:&#x2F;&#x2F;dalle.party&#x2F;</a>
评论 #38710700 未加载
flirover 1 year ago
Just in case you haven&#x27;t realised, ChatGPT&#x27;s responses are influenced by the filename you hand it. If you&#x27;re sticking with the default filename it gives you, you&#x27;re passing hints to the next generation via a backchannel.
评论 #38715531 未加载
gerashover 1 year ago
This was an interesting experiment. The images tend to converge to a specific style
评论 #38711549 未加载
elijahbenizzyover 1 year ago
Hey folks! Hitting some scaling issues. It should work if you refresh enough (and also probably make it worse :)).<p>Hoping to get some help from Streamlit (see if they&#x27;ll be nice enough to press the magic scale button), but in the meanwhile I&#x27;ve just made the source data public. Not as pretty of a viz, but feel free to download&#x2F;play around with it! Has prompts, embeddings, etc. In the process of uploading -- not a small dataset. 11gb, currently uploading...<p><a href="https:&#x2F;&#x2F;drive.google.com&#x2F;drive&#x2F;u&#x2F;0&#x2F;folders&#x2F;1pZBRHxvygHPAjACSBUpGnPZqcmIkmEdV" rel="nofollow noreferrer">https:&#x2F;&#x2F;drive.google.com&#x2F;drive&#x2F;u&#x2F;0&#x2F;folders&#x2F;1pZBRHxvygHPAjACS...</a>
评论 #38711922 未加载
webmavenover 1 year ago
I can&#x27;t tell what this is. Other than a couple of incidental UI elements that are part of the hosting platform, I just see a grey placeholder that pulses forever...
评论 #38710487 未加载