Share this page!

Maybe Scrolly?
hardmaru Profile picture
Twitter logo
Jun 14 • 8 tweets • 5 min read
Good Morning!

“Professional photograph of bears in sports gear in a triathlon in Kyoto” made using both #Imagen and #Dalle
Triathlon Bears in Kyoto
#Imagen #Dalle
The real trophies are the friends we make along the way. #Imagen #Dalle
Practice makes perfect.

“Professional photograph of bears in sports gear in a triathlon in Kyoto in a world full of cute bears” generated using #Imagen
Some samples of Triathlon bear includes a bicycle. #Imagen
They look so relaxed. #Imagen
The Final Stretch. #Imagen
Buddies. #Imagen #Dalle

• • •

Missing some Tweet in this thread? You can try to force a refresh
　

Keep Current with hardmaru

hardmaru Profile picture

Stay in touch and get notified when new unrolls are available from this author!

Read all threads

This Thread may be Removed Anytime!

PDF

Twitter may remove this content at anytime! Save it as PDF for later use!

Try unrolling a thread yourself!

how to unroll video
  1. Follow @ThreadReaderApp to mention us!

  2. From a Twitter thread mention us with a keyword "unroll"
@threadreaderapp unroll

Practice here first or read more on our help page!

More from @hardmaru

hardmaru Profile picture
Jun 13
Tried to use #Imagen to generate collectable Japanese postage stamps about VR cats. I love these results!

“Ukiyo-e painting of a cat hacker wearing VR headsets, on a postage stamp” ❤️
These metaverse cats come in all shapes and sizes.

#Imagen
Don’t worry, these VR cats are only slightly sentient 🙃
Read 7 tweets
hardmaru Profile picture
Jun 1
“Finish the cat drawing” viral meme tweet has replies with all sorts of nice, creative ‘out of the box’ thinking.

I use #Dalle’s inpainting function to do this task, and was impressed at what it can do. Here is the output using the prompt “cats”

🧵An entire thread of results 🐈
Oh boy, this is going to be a fun thread.

Let’s start with “James Bond”

#Dalle
I wanted it to come up with some photo-realistic image in the body, using all sorts of tricks, like text conditioning with the words photo, natural, etc., but it was hard.

If I used the prompt "DSLR photo" it simply gave me a photograph of the cat operating a digital camera :)
Read 31 tweets
hardmaru Profile picture
Jun 1
Good Morning!

Here’s an “Oriental painting of a dragon programming on a laptop in the Song dynasty” produced by #Imagen
More dragon coders from #Imagen🐲
“Oriental painting of a dragon programming on a laptop in the Song dynasty” by #Dalle
Read 6 tweets
hardmaru Profile picture
May 31
From Monday, the UK government has said people who have graduated, in the last 5 years, from one of the eligible universities listed on its website, will be able to apply for the UK's "high potential individual" visa.

Universities: gov.uk/high-potential…
cnbc.com/2022/05/30/uk-…
Countries, not just businesses, also have to do what they can to bid for talent.
It’s kinda weird how most of the universities are in the US though.
Read 4 tweets
hardmaru Profile picture
May 30
In a decade, most of the creative content we see will be at least partially created using tools that incorporate machine learning models, simply due to the efficiency in which content can be created, whether we like it or not.
This is similar to the trend of how most illustrators, designers, and artists, professional or amateur, now use software tools now for most creations, and how most photos are taken using smartphone digital cameras, creating the abundance of content that we have now.
But unlike previous trends, machine learning models are constantly updated with new data, which is produced by our collective intelligence reflecting the current state of our culture. If most of this new creative content is made using ML, it will lead to this weird feedback loop.
Read 4 tweets
hardmaru Profile picture
May 29
“The newest GPT-3 version (May 2022) actually did the worst at this task—they kept presenting me with real donuts that they’d seen during their training, and not even particularly weird donuts… The original early-2020 GPT-3 models were more willing to deliver the weirdness.”
Interesting observation.

There’s definitely a tradeoff (and also some “Efficient Pareto frontier”) between realism/accuracy axis and creative/weirdness axis. A bit similar to what I discussed in this thread:
A good hack might be to use an older generation language model that can come up with ‘weird text’ that’s deliberately not so realistic, and feed the weird text into a text-to-image model:
Read 4 tweets

Did Thread Reader help you today?

Support us! We are indie developers!

This site is made by just two indie developers on a laptop doing marketing, support and development! Read more about the story.

Become a Premium Member ($3/month or $30/year) and get exclusive features!

Become Premium

Don't want to be a Premium member but still want to support us?

Make a small donation by buying us coffee ($5) or help with server cost ($10)

Donate via Paypal

Or Donate anonymously using crypto!

Ethereum

0xfe58350B80634f60Fa6Dc149a72b4DFbc17D341E copy

Bitcoin

3ATGMxNzCUFzxpMCHL5sWSt4DVtS8UqXpi copy

Thank you for your support!

Follow Us on Twitter!

:(