I Got DALL-E 3 Access: My First Impressions and Tests

I Got DALL-E 3 Access: My First Impressions and Tests
👊 Become a member and get access to GitHub:
/ allaboutai
Get a FREE 45+ ChatGPT Prompts PDF here:
📧 Join the newsletter:
www.allabtai.com/newsletter/
🌐 My website:
www.allabtai.com
In this video I do test and get my first impressions of OpenAIs new text to image model Dall-E 3.
00:00 Dall-E 3 + ChatGPT Intro
00:14 Dall-E 3 Itteration Test 1
02:25 Dall-E 3 Itteration Test 2
04:37 Dall-E 3 MEME Creation
06:39 Dall-E 3 Text Illustrations
08:02 Dall-E 3 Prompting Test 1
09:15 Dall-E 3 Prompting Test 2
10:09 Dall-E 3 KZbin Thumbnails
11:46 My Conclusion

[@trud811]

The difference with Midjourney is still huge. The girl looks like a generated image, compared to a photo from Midjourney. However, the ability to work with text and adjust already generated images is a killer feature for illustration. Will be waiting from you more videos regarding this

[@carkawalakhatulistiwa]

We need midjourney v6 with gpt4 API 😎.#combination

[@vainezaiven6677]

If you're looking for something that is deliberately not realistic, though (something that is clearly artwork and does not look like a photograph), what would be better, in terms of pure quality of output?

[@trud811]

@@vainezaiven6677 I used both Midjorney and Dalle3(bing version) to create an illustration for my blog and Dalle3 performed much better. For Midjorney you need to know how to build a prompt, but Dalle3 just guess what you need

[@vainezaiven6677]

@@trud811Thanks for the quick reply! Have you tried getting the program to remake the same character over and over? Does it keep the appearance consistent? That's basically my big remaining question at this point. The guy in the video does do several images with the same female model at the beginning, but I'm curious just how consistently we can have the same character appear again and again. Since I'm planning on using the same original character in most of the illustrations that I do, I would really like to know this.

[@trud811]

@@vainezaiven6677 Bing or Midjorney can't do this. Probably ChatGPT can(as we saw on the video), but I still don't have access to try. Hope they provide access in 1-2 weeks

[@getgonged]

I got it too and it is insane! Love it , it's not exactly midjourney but highly versatile and i don't have to be in discord or be a prompt engineer to use it. Your prompts are a great addition to try!

[@gilan4287]

please bro do as much as you can when it comes to consistency. that's the hardest thing to achieve in MJ

[@watcher1326]

Great expansion on my understand of what Dalle3 can do.

[@user-ik8vy1rg8f]

Glad you started off with testing the 16:9 aspect ratio. I guess that's one thing that sets it apart from the free Bing version.

[@ecoruben]

Still waiting here too. Such an amazing advancement. Extrapolating the whole AI, AGI and quantum computing thing out over the next 10 years... woah @#$% this civilization is about to change sooooo rapidly compared to anything we've seen before..... if we don't screw it all up. :)

[@chanclaverde]

You don't need to assign tags to the images just so you know. You can say #1 #2 #3 #4 (or 1st, 2nd, 3rd, 4th) and it already knows.

[@ca6360]

We all did! BING

[@SergeGolikov]

So how do we access this version for the 16x9 AR? The Bing version is still only1:1, image quality is amazing, been a fan of E-2 for months.

[@olagunjujeremiah4003]

When are we getting in the uk 🇬🇧

[@jaffers8593]

that pickle one. holy shit

[@MudroZvon]

Very cule!

[@MissChelle]

How are you accessing this? Could you give us a link please? I can only get 1:1 ratio, maybe I haven’t got the correct place. It does say Dalle 3 tho…… ❤️🇦🇺

[@hldfgjsjbd]

What about comparing with new SD also?

[@digidope]

Compare how? Dalle3 via Bing is the first AI image generator with natural language prompting. It does exactly what you want. SD and MJ neither can do that. In SD it depends so much what model and LoRa and ControlNET you use. So how to compare is quite difficult. In most comparisons what people do in X, Dalle3 have been better than Midjourney on pretty much everything. Note: Its seems that ChatGPT is not very good at Dalle3 prompting. You get better results via Bing.

[@puja1985]

please let me know how to access dall e 3 in chat gpt plus , I just subscribe to chat gpt plus and It doesn't show any dall e 3 plugins beta option.. thanks

[@vainezaiven6677]

It's not widely available yet.

[@DarkandTwisted]

They are rolling it out slowly to us Plus members. In have been a Plus member since day one, but I still don't have these new features yet. I have been waiting for like a week, and still nothing.

[@vainezaiven6677]

@@DarkandTwisted Keep checking. I found it available to me out of the blue last night. You'll have to start a completely new chat and select DALL-E 3, though. It won't let you just start using it in an existing chat.

[@kritikusi-666]

Looks awesome. Thanks for sharing. I am still waiting for my turn to have at it.

[@sachindatt4045]

How you got dalle 3, I have gpt plus but no dalle 3 in it yet....aarghhh.

[@jpmallick367]

Wait when did Kirby become a pokemon

[@happysuperhappy2091]

how do you get dall-e 3 access in chatgpt???

[@vainezaiven6677]

It's not widely available yet.

[@happysuperhappy2091]

@@vainezaiven6677 why not?

[@marki2325]

In October sometime. Still waiting here also…..

[@jonathan-6948]

Strange I got it already several days ago, and I'm in Germany. Maybe business accounts get it first

[@TheMortzCanberra]

this would make me pay for GPT

[@trashcanmucous5153]

why cant DALL-E or even midjourney make a single nuanced object like a Butterfly knife/balisong? it's always a generic weird knife with decorations