DALL-E 3 Makes INSANE AI Images

DALL-E 3 AI image generator secretly launched on Microsoft's Bing. I recruited the help of some friends to show off its capabilities. It makes some HILARIOUS AI images.

Follow me!
Gaming channel - @Greenskull

#dalle3 #ai #photoshop #aiart #artificialintelligence #comedy #cool #funny

[GreenskullAI]

What a time to be alive. DALL-E 3 is wild, isn't it?

[barneymiller]

Terrible. All cartoon-y. Dall e 2 could give you truly interesting, creative images. Dall e 3 is a step back unless you like porn, candy, cartoon crap. Depressing.

[RandomArtist]

Something I found impressive about these ai images is that they have the correct number of fingers on a hand. As well as the text

[noobi_yumyum]

Missed opportunity for them to call it DALL-3

[FiFiTheFish]

Noir greenskull isn’t real he can’t hurt you
Noir greenskull: 3:27

[TheFruitNewt]

its really good at text, xenomorphs, predators(YAUTJAS! THEY'RE CALLED YAUTJAS!), lego, and jason voorhees.

[Crudely-Drawn-Cupcake]

0:00 Bruh why is he eating the pencil crayons? The pencil crayons taste worse than the normal crayons!

[sap-unny]

I don't know why noir style turned you into Adam Scott

[DeadFishFactory]

The filters are so restrictive now that it's hard to generate anything interesting. Hope y'all just love generating landscapes and generic stock photos, because the AI treats your request for anything with the assumption that you're a perverted weirdo looking for pornographic material and will block everything you make.

[BJBUnicorn]

Wow there is a lot of master chief but you have never played halo right…

[TarunBandela-vrfu]

Now it not working for celebrates image create...😢😢😠

[ZAPcon]

Dall-E 3 inside of GPT is WAY more strict. Wanted to try a couple of these prompts (like the alien one) and they got flagged. I guess Microsoft can platform the exact same product with different content policies... even though OpenAI is the one that made the product? Very confusing. Hope it lightens up, because the conversational flow inside of GPT is far superior to any other prompting method I've used. It feels more like you are communicating to an artist making revisions rather than trying to engineer the perfect prompt.

[azelusnova]

I also do think that it tries to translate GLUBO. It is very similar to Spanish GLOBO meaning balloon, globe.

[qoombert]

i asked for a sandwich cafe and it featured japanese text

[MorrisseysthFriend]

Unfortunately it won't produce an image that looks like real characters. I understand why, but it doesn't satisfy my fan fiction imagination 😂

[nicolaecalin]

4:08 Wow. Just wow. You could show that to me and I would belive it's real. Like there are no visible errors that ai usually has Ike weird hands or faces or body proportions. The Sonic is absolutely perfect and you can't tell me otherwise. That background is beautiful, the only errors are Goku's sonic shaped nose that you pointed out, the foot is too small and the ear is a bit weird. And it's all free? Like bruh

[boomerstrikeforce]

It just creates oil painting looking pics Even when I tell it to use realistic photography, photorealism, or I actually put in variables like camera lenses that easily work in mid journey. What is the secret sauce to make Dolly 3 actually create photorealistic looking people or environments from the initial image generation?

[shadeslayerzycron]

I just tried to make a soul stealing potato, then glubo as an anime girl and… prompt blocked

[nicolaecalin]

I cannot belive the brick restaurant one. It got a lot of text completely right! Sure some is still bad, but you can make out words! And most good ai text images are only good with text you tell them to write, but background text is still bad, but here it's something else

[ryanclemons]

It's too restrictive won't let me make pokemon or anything copyright it's just annoying to try and use will wait for something else.