Discover and read the best of Twitter Threads about #generative
Most recents (23)
The G7 Communiqué is out. It covers a range of topics: trade, climate change, science and technology, artificial intelligence, anti-corruption, and so much more. This short thread 𧡠focuses mostly on the trade/tech/econ security themes.
whitehouse.gov/briefing-room/β¦
whitehouse.gov/briefing-room/β¦
On artificial intelligence, the G7 recognizes the possible benefits but clear drawbacks of #generative AI. They will thus establish the *Hiroshima AI process,* a G7 working group, in cooperation with the OECD and GPAI, for discussions on generative AI later this year.
The trade section offers unequivocal support for the WTO right off the bat, celebrating positive outcomes from MC12 (ie, fisheries) and anticipates progress in other areas like the #JSI ahead of MC13. Unclear where the Biden administration views its relationship evolving re: WTO.
How should product managers respond to generative AI?
I wrote an article about how the emergence of generative AI will change future products, and how product managers should respond.
#generative #productmanager
I wrote an article about how the emergence of generative AI will change future products, and how product managers should respond.
#generative #productmanager
Key takeaways:
1. Generative AI is revolutionizing fields like image generation, copywriting, and code writing, and has significant business implications in terms of quality, speed, creativity, and cost.
1. Generative AI is revolutionizing fields like image generation, copywriting, and code writing, and has significant business implications in terms of quality, speed, creativity, and cost.
Just gave GPT-4 access to use Chrome however it wants (click, scroll, fill forms) thru @LangChainAI Auto-GPT agent. The results were mind-blowing!
Here's what it did: π
π§Bought me an AirPod
πHelped me find an event venue and filled out the contact form
π½οΈBooked restaurants
Here's what it did: π
π§Bought me an AirPod
πHelped me find an event venue and filled out the contact form
π½οΈBooked restaurants
Iβve also open-sourced the project so you can try it out too! ChromeGPT is built using @LangChainAI and @SeleniumHQ , supports Zero-shot, #BabyAGI and #AutoGPT agents. github.com/richardyc/Chroβ¦
More ChromeGPT demos: π§΅
2/4
#ChromeGPT #ChatGPT #langchain #generative #GPT4
More ChromeGPT demos: π§΅
2/4
#ChromeGPT #ChatGPT #langchain #generative #GPT4
Demo using #langchain Auto-GPT agent
Prompt: Find me a bar that can host a 20 person event near Chelsea Manhattan. Fill out contact us form if they have one. π
3/4
Prompt: Find me a bar that can host a 20 person event near Chelsea Manhattan. Fill out contact us form if they have one. π
3/4
What's interesting about this proposal in the context of #generative and Copyright: the place with the most creator-friendly legislation can have jurisdiction. Under the Berne Convention, it's where infringement occursβso that could be the most creator-friendly place you chose.
That is to say, "allowing #generative AI research under ethical standards [...]" that require Copyright are already regulated by the most human-friendly, pro-creator, and jurisdiction is automatically (e.g. that's where data is hosted, service is used) as per Berne.
Thus, we have the tools to make this happen for #generative systems. Just need the guts to enforce them now before the U.S. court system normalizes exploitation that goes against Berne...
Have you seen #videoart on @fx_hash_ before?
Teleport Loopr is my genesisπ«
Mints open in 10 hours!β³
[ Sept 14th / 15:00 UTC ]
Follow this thread ! π§΅#fxhash
#creativecoding #p5js #generativeart #CleanNFT
Teleport Loopr is my genesisπ«
Mints open in 10 hours!β³
[ Sept 14th / 15:00 UTC ]
Follow this thread ! π§΅#fxhash
#creativecoding #p5js #generativeart #CleanNFT
Videos per iteration:
[3], [4] or [5]
Color inversion effect may be applied to:
[ one video ], [ none (rare) ] or [ variable (super rare) ]
[3], [4] or [5]
Color inversion effect may be applied to:
[ one video ], [ none (rare) ] or [ variable (super rare) ]
Have you seen #videoart on @fx_hash_ before?
Teleport Looper is my genesisπ«
Mint opens:
Sept 13th / 15:50 UTC
Follow this thread ! π§΅#fxhash #tezos
#creativecoding #p5js #generativeart #CleanNFT
Teleport Looper is my genesisπ«
Mint opens:
Sept 13th / 15:50 UTC
Follow this thread ! π§΅#fxhash #tezos
#creativecoding #p5js #generativeart #CleanNFT
Link:
fxhash.xyz/generative/192β¦
There are 30 .mp4 files available to be randomly chosen by the code ~
~ videos per iteration:
[3], [4] or [5]
~ color inversion effect may be applied to:
[ one video ], [ none (rare) ] or [ variable (super rare) ]
fxhash.xyz/generative/192β¦
There are 30 .mp4 files available to be randomly chosen by the code ~
~ videos per iteration:
[3], [4] or [5]
~ color inversion effect may be applied to:
[ one video ], [ none (rare) ] or [ variable (super rare) ]
All videos are part of my audiovisual research and were recorded from 2016 to 2021 in Brasil, Chile and Canary Islands ~
They're all short and looped in Teleport Looper project, just like moving photos:
They're all short and looped in Teleport Looper project, just like moving photos:
Coalesce is coming to SuperRare Friday, July 1st. Itβs a collection based on circle packing and simplex noise. What began as a response to a Genuary prompt slowly developed into a larger concept. This project is the culmination of several monthsβ efforts.
#generative #codeart
#generative #codeart
The collection is divided into styles or sets, consisting of five pieces each. They're set apart by the combination of constraints used to place the circles, giving rise to different forms. Coalesce is the name of the collection as well as the first series in the collection.
The four styles are Coalesce, Multifield, Colorshift, and Lens. A new style will be released each Friday. Additional styles may be added in the future, but no further works will be added to the current sets.
vec2 p = FC.xy/r-.5; float N=2e2,n=-N;o+=1.+sign(6.-length(FC.xy-m*r));
for(;n++<=N;)o-=6./N*sign(n)*texture(b,p/.997+.5+(4.+2.*sign(n)+cos(n))*cos(PI2*(n/N+vec2(0.,.25)))/(length(p)+.2)/r);
//run on twigl.app using mode geekest (300 es)
//#γ€γΆγγGLSL #generative
for(;n++<=N;)o-=6./N*sign(n)*texture(b,p/.997+.5+(4.+2.*sign(n)+cos(n))*cos(PI2*(n/N+vec2(0.,.25)))/(length(p)+.2)/r);
//run on twigl.app using mode geekest (300 es)
//#γ€γΆγγGLSL #generative
vec2 p = FC.xy/r-.5; float d=length(FC.xy-m*r), N=1e2,n=-N;o+=.1*exp(-d/3e2);
for(;n++<=N;)o-=1.3/N*sign(n)*texture(b,p*.998+.5+9.*(4.+2.*sign(n)+cos(n))*cos(PI2*(n/N+vec2(0.,.25)))*pow(length(p)+0.3,4.)/r);
//similar code, but different parameters
//#γ€γΆγγGLSL #generative
for(;n++<=N;)o-=1.3/N*sign(n)*texture(b,p*.998+.5+9.*(4.+2.*sign(n)+cos(n))*cos(PI2*(n/N+vec2(0.,.25)))*pow(length(p)+0.3,4.)/r);
//similar code, but different parameters
//#γ€γΆγγGLSL #generative
Oh, wow. Thanks to all of you! Here's some more from yesterday. It's not tweet size, but maybe you know more tricks?
interactive version bit.ly/3KV92Kz
interactive version bit.ly/3KV92Kz
I finally took time to review the 512 Window in detail. The rarity percentage might not be very indicative on @fx_hash_ so here is a highlight thread of emergent behaviours from the system and the most peculiar outputs.
Starting with the unique "Center Cross" (1/512): #69. Nice.
Starting with the unique "Center Cross" (1/512): #69. Nice.
The "Desaturated" (5/512): #17, #119, #188, #509, #510
The less frequent "Tri-shades" (8/512): #117, #154, #161, #222, #364, #385, #390, #470
.@fx_hash_ new art appreciation thread.
Featuring @yiwen_lin, @JuhaniHalkomaki, @lomzart, @sam___tsao, @camillerouxart
#cleanNFT #architecture #generativeart #fxhash #NFT #generative #tezos
πͺ‘
Featuring @yiwen_lin, @JuhaniHalkomaki, @lomzart, @sam___tsao, @camillerouxart
#cleanNFT #architecture #generativeart #fxhash #NFT #generative #tezos
πͺ‘
Hex by @yiwen_lin
Tech π§βπ»: using math (lots of math), to create hexagons and to fill them with lines of various properties
Looks π¨: love that even if some hexagons overlap they still look good, even add to the overall depth of the generated piece
Tech π§βπ»: using math (lots of math), to create hexagons and to fill them with lines of various properties
Looks π¨: love that even if some hexagons overlap they still look good, even add to the overall depth of the generated piece
LΓΆv by @JuhaniHalkomaki
Tech π§βπ»: branch and leaves generation, with p5js
Looks π¨: the various shapes of the leaves, the colours, how the branches interact with the white box 'wall', it's all VERY beautiful
Tech π§βπ»: branch and leaves generation, with p5js
Looks π¨: the various shapes of the leaves, the colours, how the branches interact with the white box 'wall', it's all VERY beautiful
.@fx_hash_ new art appreciation thread.
Featuring @jMarkusiewicz, @notlion, @ambergrids, @ClaudioDallaB, @semicolonartist
#cleanNFT #architecture #generativeart #fxhash #NFT #generative #tezos
πͺ‘
Featuring @jMarkusiewicz, @notlion, @ambergrids, @ClaudioDallaB, @semicolonartist
#cleanNFT #architecture #generativeart #fxhash #NFT #generative #tezos
πͺ‘
Unbuilt by @jMarkusiewicz
Tech π§βπ»: applying sacral architectural rules in code
Looks π¨: excellent colour choice and doesn't look like a generative output! the name and date underneath the generated layout is a nice touch
Tech π§βπ»: applying sacral architectural rules in code
Looks π¨: excellent colour choice and doesn't look like a generative output! the name and date underneath the generated layout is a nice touch
Indirection β€³β€βΆ by @notlion
Tech π§βπ»: using fonts (Unicode Arrows) with algorithms
Looks π¨: the way the art unfolds and generates the paths makes it really interesting, a nice touch is that few of the arrows are in a different/highlight colour
Tech π§βπ»: using fonts (Unicode Arrows) with algorithms
Looks π¨: the way the art unfolds and generates the paths makes it really interesting, a nice touch is that few of the arrows are in a different/highlight colour
ok, so here's why you would buy one of my #NFT that's branded with the current date:
you broke up with your girlfriend today. bad breakup, much shouting, accusations, the usual stuff. you feel like shit, obviously. how about going to a casino and gambling away a stack...
you broke up with your girlfriend today. bad breakup, much shouting, accusations, the usual stuff. you feel like shit, obviously. how about going to a casino and gambling away a stack...
...of money to ease your mind? STOP!
instead, you should relief your grief by buying a #NFT from @days__nft with the CURRENT DATE. because it's a nice piece of #generative art, and it's delivered as infinitely zoomable svg file, you can print it on canvas, and hang it in your...
instead, you should relief your grief by buying a #NFT from @days__nft with the CURRENT DATE. because it's a nice piece of #generative art, and it's delivered as infinitely zoomable svg file, you can print it on canvas, and hang it in your...
...living room! there, the printout serves as visual reminder of the shitty day you're just having. BONUS: we're currently having a limited edition black and white series, so if your breakup happens soon enough (this week! β±), you can stare at a depressingly color-less...
After a succesful and exciting launch of my first blockstyle `Infinite Beauty`: ethblock.art/create/15
on Ethblock.art, I've been having a lot of fun discovering what's out there!
#nft #nfts #cryptoart #generativeart #Ethereum
on Ethblock.art, I've been having a lot of fun discovering what's out there!
#nft #nfts #cryptoart #generativeart #Ethereum
I decided to take a favorite pick every day during March to share it with all of you. The months start with unique entities and no symmetry, evolving to a max symmetry of 101 on months with 31 days.
#nft #nfts #cryptoart #generativeart #Ethereum
#nft #nfts #cryptoart #generativeart #Ethereum
Feel free to join in and post your favorite!
My pick for today on March the 4th:
"Go For Gold"
from ethblock.art/create/15/1196β¦
#nft #nfts #cryptoart #generativeart #Ethereum
My pick for today on March the 4th:
"Go For Gold"
from ethblock.art/create/15/1196β¦
#nft #nfts #cryptoart #generativeart #Ethereum
#Genuary Day 3: Something human. (1/4)
#genuary2021 #generative #generativeart #creativecoding #genartclub
#genuary2021 #generative #generativeart #creativecoding #genartclub
(2/4) Human experience is in fact precisely what a plotter lacks--it can't see what it draws, it doesn't feel the pressure of the pen on the paper, in fact it doesn't even know it's current position. It just knows that the computer tells it how to move its motors.
(3/4) That is why it was up to me to program some of these features in the algorithm that ultimately controls the motor movements. I use signed distance fields and ray marching to give it a feeling of shape, light and shadow.
Let's start our tour of research papers where #generative meets deep learning with this classic by Gatys, Ecker and Bethge from 2015.β¨
A multimedia tutorial & review in a thread! π
π Texture Synthesis Using Convolutional Neural Networks
π arxiv.org/abs/1505.07376 #ai
A multimedia tutorial & review in a thread! π
π Texture Synthesis Using Convolutional Neural Networks
π arxiv.org/abs/1505.07376 #ai
Here's the nomenclature I'll be using.
βοΈ Beginner-friendly insight or exercise.
π³οΈ Related work that's relevant here!
π Open research topic of general interest.
π‘ Insight or idea to experiment further...
See this thread for context and other reviews:
βοΈ Beginner-friendly insight or exercise.
π³οΈ Related work that's relevant here!
π Open research topic of general interest.
π‘ Insight or idea to experiment further...
See this thread for context and other reviews:
The work by Gatys et al. is an implementation of a parametric texture model: you extract "parameters" (somehow) from an image, and those parameters describe the image β ideally such that you can reproduce its texture.
I'll be using these textures (photos) as examples throughout:
I'll be using these textures (photos) as examples throughout:
It appears that the experiment known as creative.ai will be ending soon, roughly four years after it began. Since I got permission to share, I thought I'd write more about my part in particular and what I could have done differently.
Thread < 1 /π§΅>
Thread < 1 /π§΅>
2/ I say "experiment" because on some level everything is just an experiment, with lessons to be learned from the outcomes. Experiments never fail!
With hindsight I feel more able to discuss this rationally...
With hindsight I feel more able to discuss this rationally...
3/ However, when you're in the trenches, everything can feel like a critical moment β with the full weight of potential failure behind them.
ππππππ¦ π πππ ππ‘ π€ππ πππ π¦, ππ’π‘ ππ πππ ππ£ππ π πππ ππ‘ π€ππ’ππ ππ π‘βππ βπππ.
ππππππ¦ π πππ ππ‘ π€ππ πππ π¦, ππ’π‘ ππ πππ ππ£ππ π πππ ππ‘ π€ππ’ππ ππ π‘βππ βπππ.
New paper out!
scirate.com/arxiv/1910.015β¦
It's time to talk about #machinelearning, generative models, statistical physics and spin glasses
scirate.com/arxiv/1910.015β¦
It's time to talk about #machinelearning, generative models, statistical physics and spin glasses
First, I'm very happy to say this is the first purely-@ICFOnians collaboration I'm doing (although Przemek is now in @UAM_Poznan), and I've been lucky to work side-by-side with the great @gorka_mgm, who is also a @BecarioslaCaixa
But let's go to the matter. The problem we deal with has to do with #unsupervised #learning. Given some data, we want to learn its fundamental features, in order to generate more data. This is used in #generative modelling for image denoising, colorization... etc
Always wanted to compile a visual list of #generative artists I admire but don't know what form is better so here you go
#1
Twitter: @dmitricherniak
IG: instagram.com/dmitricherniak/
Github: github.com/dmitric
#1
Twitter: @dmitricherniak
IG: instagram.com/dmitricherniak/
Github: github.com/dmitric
Neural approaches to style transfer struggle with certain types of art, e.g. crisp yet smooth brush-strokes ποΈ. It's likely a combination of factors, including using models pre-trained on natural images. π·
In this thread I'll experiment to learn more! π #neuralimagen #procjam
In this thread I'll experiment to learn more! π #neuralimagen #procjam
Ah. Here's what happens visually when the optimization algorithm (L-BFGS) "explodes" π₯. I have not tracked down the cause yet, likely extending #PyTorch to understand parameter ranges would help.
(Left: previous iteration looks OK, Right: pixels go far out of range β clamped)
(Left: previous iteration looks OK, Right: pixels go far out of range β clamped)
Example outputs of the Picasso drawing with more bias towards upper layers of the convnet (4x), then equalized, then bias towards lower layers (4x).
Interesting, but nothing like the original. Some insights forming!
Interesting, but nothing like the original. Some insights forming!
So many of my sketches begin with the same function: producing a uniform distribution of values along an array, between [0..1] range. Here's how it looks in JavaScript.
I'm calling it "arcs" (for arc lengths) but perhaps there is a proper name for this sort of thing?
I'm calling it "arcs" (for arc lengths) but perhaps there is a proper name for this sort of thing?
I use this for all sorts of things, like generating an N-sided polygon:
Pen plotter symphony #plottertwitter
Box fitting, filling, time lapse #plottertwitter
Box fitting, white gel pen on black paper. I need to find a better way to pause/resume plots #plottertwitter 
