charlota Profile picture
Jul 4, 2022 • 11 tweets • 6 min read Twitter logo Read on Twitter
I generated this set of icons in 3D style using DALL-E 2 by @OpenAI. 🤖

(or, did an AI generate this set of icons using text prompts by Charlota?🤔)

#dalle2 Image
@OpenAI anyway, I wanted to try to replicate a visual style you can find around the internet a lot recently:

rendered 3D objects levitating on the page softly illimunated with some kind of gradient. ImageImageImageImage
This was the text prompt I used:

an icon of a paper plane in light blue metallic iridescent material, 3D render isometric perspective on dark background Image
Let's decompose the text prompt a bit!

icon - makes sure the object has reduced detail

light blue metallic iridescent material - specifies the finish, you can go wild here

3D render - especially with 'icon' in the prompt, this makes sure you don't get a 2D extruded icon ImageImage
isometric - this helped so-so to get everything in a similar perspective, otherwise, it would be very random

dark background - made it easier to erase background in post-production, as DALL-E can't export images with transparent background (yet?) ImageImageImageImage
With some light retouching and colour correction, I can imagine using these immediately.

The caveat is DALL-E only exports in 1024x1024px, which isn't a great resolution.

But for small icons on a website, it would work fine. Image
What does it mean for designers?

Hard to say.
I can't do detailed 3D modelling.
DALL-E enabled me to find a workaround.

You ofc don't have precise control of details, tweaks and image quality.
But I think this will inevitably come in the future.
It took me around 80 mins to iterate the text prompt, generate images, remove the background and put them into a grid.

These icons aren't perfect, but I could definitely imagine using them for art direction, mood boarding or quick exploration, i.e. in a website mockup.
Thanks for the overwhelming response! I tried another take with a more rainbow metallic finish:

prompt:
icon of a <cute unicorn head> in metallic rainbow iridescent material, 3D render isometric perspective rendered in Cinema 4D on dark background Image
including 'rendered in Cinema 4D' in the prompt somehow adds more definition and sharpness to the object - and also reflections and ground shadows.

adding words like 'cute', 'simple' or 'tiny' removes details and complexity of an object.

(though these are mostly hit or miss) ImageImageImage
here's a work-in-progress pic of all the images generated by DALL-E before I removed the background and slightly resized them.

I haven't done any colour correction or retouching.

#dalle2 Image

• • •

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

Keep Current with charlota

charlota 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!

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!

:(