Bing Image Creator (DALL-E 3) can now create text

in #hive-163521last year

Microsoft hat vor Kurzem den Bing Image Creator auf DALL-E 3 upgegraded, woraufhin das Service seit einigen Tagen mit Überlastung zu kämpfen hat.

https://bing.com/create

DALL-E 3 ist deutlich besser als die Vorgängerversion und kann das erste Mal auch teilweise Text generieren, was früher überhaupt nicht gegangen ist.

Die KI macht aber noch häufig Fehler, hält sich nicht genau an den Prompt (handgeschrieben) und macht auch Rechtschreibfehler. Fast wie ein Mensch 😂

Was sagt ihr dazu? Habt ihr schon Bing / DALL-E 3 probiert, um damit Text zu generieren?

Prompt

illustration of a blackboard with the handwritten text "VIKI SECRETS"

Bing (DALL-E 3)

bing-viki-secrets-1.jpg

Spelling Mistake (same prompt)

bing-viki-secrets-2.jpg

English

Microsoft has recently upgraded the Bing Image Creator to DALL-E 3, whereupon the service has been struggling with overload for a few days.

https://bing.com/create

DALL-E 3 is much better than the previous version and for the first time can also partially generate text, which was not possible at all before.

However, the AI still makes frequent mistakes, does not adhere exactly to the prompt (handwritten) and also makes spelling mistakes. Almost like a human 😂

What do guys think? Have you tried Bing / DALL-E 3 to generate text with it?

Live your Secrets and Hive Prosper 🍯

xx Viki @vikisecrets

Posted with STEMGeeks

Sort:  

It's wonderful to see that they have updated the DALL-E version, although there are still certain flaws that need improvement, I believe it's a good option for creating images with this AI. I haven't had the opportunity to use this DALL-E 3 version yet, but I hope to be able to run some tests soon.

I did already some images and am quite impressed by the new version.

I love this, never even new it existed so thanks for sharing - this is what I asked for "alien spaceship in a night sky above mountains"

_99de22ff-7d07-477d-9943-1d0f0fd59641.jpeg

that looks really cool :)

This is a great development 😁

Wirklich praktisch und das mit dem Rechtschreibfehler ist sehr lustig! 😂

Thank you for this update. AI are not above making mistakes, right? Moreover it is the same human that makes mistakes that invented it. We are not perfect not even the AI

I have not tried it but I don't really think I will try it out either. It's nice to see the development on AI though.

I haven’t used the image creator till now. However, I am sure the next version will be better. The AI is developing so fast, it will soon be able to do the work without spelling mistakes

I have not tested it yet.
But I hope they will solve your problems.
It looks like a good tool for many things.
Regards

I haven't picked it up yet but the way you say it sounds interesting, I'll definitely give it a try.

Sieht cool aus 👍

I didn't try it but I'm going to ask him something crazy right now.

i tried it. you just have to describes specifically the image you want.

Verstehe gar nicht wieso es diese Fehler gibt, hatte das auch schon vor einer Weile getestet, glaube bei night Cafe.

Einen Text drauf zu schreiben müsste doch eigentlich der einfachste Teil an dem ganzen sein.

Wollte für mein geplantes Spiel damals dessen Namen drauf haben, was nicht funktionierte.

Sollte ich vielleicht nochmal nun testen.

I'm actually happy that you spoke the truth about AI still making mistakes
That happens a lot of time


Artificial intelligence is here to stay and I hope it does not become a surveillance weapon.


hey that looks cool! but i havent tried the apps yet.

DALL-E 3 is really something else compared to DALL-E 2 !
But text is very difficult for every models. It is funny to see that you have some Japanese-based gibberish on the bottom lest corner of the 2nde image !

haha, that's funny :)