AI Image Generation

[ntxdave]

Saw this topic on the SketchUp forum and thought some of you might find it interesting. Just a warning though, it is both interesting and scary.

Here is the main topic and the software mentioned Dall.E 2 and here is a place to see how it works.

Hope you enjoy

[Rich O Brien]

I have both Dall-E and MidJourney accounts and can honestly say it's fascinating to use both.

Here's what MidJourney made based off the following phrases...

Code: Select all

Twitter troll farm, realistic, 3D, render, volumetric, dystopian

Code: Select all

bulldog sleeping in a bubble on a cloud, stylized, retro, dream, render, octane, realism

Code: Select all

ancient Celtic shield, ornate, damage, metal, natural volumetric lighting, realistic 4k octane beautifully detailed render, 4k post-processing

Here's what Dall-E made based off the same phrases...

[ntxdave]

Wow, really interesting results Rich.

However, here is an interesting post from @gsharp on the SketchUp forum:

Raises another interesting question… who owns the copyright of the image ?

The images used by the software to create the concept?
the DALL-E 2 software that created the image
the user who input the generating statement?

[Rich O Brien]

[pre]Raises another interesting question… who owns the copyright of the image?

Reading the ToS of using the service to establish usage...

https://labs.openai.com/policies/content-policy

https://labs.openai.com/policies/terms

https://midjourney.gitbook.io/docs/bill ... cial-terms

...but at the end of the day we are contributing to an experiment in machine learning. So its quite easy to not partake as it is to partake.

[ntxdave]

Interesting links Rich. You should add them to the SketchUp forum thread. It will be interesting to see where this goes in the future.

As was said before: interesting and scary at the same time.

[pilou]

A free one! Without register...
https://www.craiyon.com

[ntxdave]

Another interesting one Pilou. Thanks for sharing.

[Rich O Brien]

[pilou]

My favorite stays for the moment the free Gaugan 2 for it's speedy results, simplicity, versality!

100 variations/day i will stop after a full year! (248 missing)

[pilou]

Another free one...but maybe not so easy! ;

Video

[L i am]

I find this creepy and scary.
I wonder what the limitations there are? Like what about the words: Death, murder, rape molestation, amputation, death of a child, brutal car accident, smashed body, Chilling

[Rich O Brien]

Your typical controversial words are disallowed. Worst thing you can create is some type of fantasy horror type vibe. Deep fakes are also disallowed.

[pilou]

Why Wombo is the more easy free fun to use!

Video

[pilou]

Half million images overpassed!
by Gulfac

[pilou]

https://beta.dreamstudio.ai/dream

Video

[pilou]

Another free one!

https://enstil.ai
Seems cool! (each image 512 *512 )
Rainy day flowers Moebius style

[pilou]

Ai for textures!

https://withpoly.com/

Video

[pilou]

Free Wombo 1 seconde ! ( Now 200 letters admissible!

Free Gaugan 2 with my pilouteries!

[pilou]

Quasi all results are valid!

[pilou]

Native language permitted!

[pilou]

We can input any numbers or letters without space! (max 200)

[Rich O Brien]

Whatever dataset they trained their AI on its clearly not up for much...

Clay roof tiles?

[pilou]

30 Wombo styles with "futurist architecture in the mountains"

[pilou]

2 new styles for Wombo art! Paint & HDR

[pilou]

A another one...https://stability.ai/blog/stable-diffus ... ic-release

[pilou]

for the pleasure of the eyes

[ntxdave]

Pilou, you sure seem to be enjoying this software.

[pilou]

Yep i am fan!

[pilou]

How crazy is this!
an image each 5 seconds to the infinite!

[pilou]

Seen this
"Realistic portrait of a beautiful woman from [country of origin], highly detailed, symmetrical, intricate, by Serge Marshennikov, greg rutkowski"
k_lms, CFG 7.5, Seed 6, Steps... I think 80.
II don't know the Ai..
But with Wombo A little problem with the eyes in big zoom but...