How to Use DALL.E 3 - Top Tips for Best Results

Summary

The video discusses the rollout of DALL-E 3 inside ChatGPT for subscribers, allowing text-to-image generation within the chatbot. It compares DALL-E 3's capabilities to Midjourney and Bing's image creator, highlighting how ChatGPT helps craft better prompts for high-quality images. Key features covered include refining images through additional prompts, adding text overlays, downloading images, changing resolutions, and noting stricter copyright rules versus Bing and Midjourney. Overall, it showcases how DALL-E 3 takes text-to-image generation to the next level with ChatGPT's prompting assistance.

Takeaways

• 😀 Dall-E 3 is now accessible to more Chat GPT Plus users, offering the ability to generate images directly within Chat GPT without needing a separate platform.
• 🔔 Microsoft, owning half of OpenAI, allows free access to Dall-E 3 through Bing's new creative mode, facilitating image creation within Bing's AI chatbot.
• 📝 Chat GPT assists in refining prompts for Dall-E 3, eliminating the need for users to craft the perfect prompt themselves, thus enhancing the image generation process.
• 📸 Users can now add text to images directly in Chat GPT, a significant improvement over previous versions where text addition was less accurate.
• 💻 Bing's integration with Dall-E 3 demonstrates similar output quality to Chat GPT, with additional customization options available through Microsoft Designer.
• 📈 Chat GPT offers a variety of image resolutions and orientations (square, horizontal, vertical) to suit different needs, with simple prompts to change these attributes.
• 🛠 The platform's capability to refine images based on feedback allows users to iteratively improve their generated images by specifying preferences like 'more smiley' or adjusting the composition.
• 📱 Dall-E 3 in Chat GPT adheres to strict copyright rules, avoiding the creation of images that could infringe copyrights or resemble copyrighted styles too closely.
• 💾 The limit for image generation in Chat GPT aligns with the platform's messaging cap, indicating a balanced usage policy to prevent system overload.
• 💭 Mid Journey is recommended as another leading text-to-image platform for users to explore and compare outputs with Dall-E 3, offering diverse creative options.

Q & A

What is Dolly 3 and how does it integrate with Chat GPT?

Dolly 3 is an advanced version of image generation AI that allows users to create images by typing text prompts directly within Chat GPT. It eliminates the need to use a separate platform for image creation and enables users to refine prompts with Chat GPT's assistance for improved image outputs.

How can users access Dolly 3 for free?

Users can access Dolly 3 for free through Bing.com. Microsoft, owning half of OpenAI, provides a feature under Bing's chat that lets users create images using Dolly 3 within the 'More Creative' mode.

What are the key differences between Dolly 2 and Dolly 3?

Dolly 3 provides enhanced capabilities over Dolly 2, including the ability to refine prompts directly within Chat GPT for more accurate and diverse image outputs, as well as the addition of text to images.

How does Chat GPT Plus users' access to Dolly 3 work?

Chat GPT Plus users get access to Dolly 3 through the GPT-4 dropdown menu. Unlike other features that require activation in settings, Dolly 3 appears automatically for users, though its rollout is gradual and not immediately available to all users.

What are the limitations of using Dolly 3 in Chat GPT?

Dolly 3 in Chat GPT adheres to strict copyright rules, preventing users from creating images that infringe on copyrights, such as characters or styles associated with protected brands or artists.

How does Bing's implementation of Dolly 3 differ from its integration in Chat GPT?

While both Bing and Chat GPT utilize Dolly 3 for image creation, Chat GPT aids in prompt refinement, offering a more interactive experience. Bing, on the other hand, provides a straightforward image creation tool without the guided prompt refinement available in Chat GPT.

What are the image resolution and format capabilities of Dolly 3?

Dolly 3 can generate images in square, horizontal, and vertical formats with resolutions of 1024x1024 for square, 1024x792 for vertical, and allows for resolution adjustments through simple prompts.

Can users modify the output of Dolly 3 images within Chat GPT?

Yes, users can refine the output of Dolly 3 images within Chat GPT by providing feedback on the generated images, allowing for iterative improvements to achieve the desired result.

What are the usage limits for generating images with Dolly 3 in Chat GPT?

The current limit for generating images with Dolly 3 in Chat GPT is tied to the GPT-4 usage limit, which is 50 messages every 3 hours, including both text interactions and image generation requests.

How does the free version of Dolly 3 on Bing work in terms of usage limitations?

The free version of Dolly 3 on Bing allows for 100 fast generations per day, after which the generation speed may decrease, but it does not strictly limit the number of images a user can create.

Outlines

00:00

😊 DALL-E 3 Now Available in ChatGPT for Paid Users

Paragraph 1 discusses the rollout of DALL-E 3 image generation capabilities directly within ChatGPT for paying ChatGPT Plus users. It explains how DALL-E 3 allows users to generate images by typing text prompts, with ChatGPT assisting in crafting effective prompts. Examples demonstrate creating a moon image, refining prompts, changing image sizes and orientations, adding text, downloading images, and more.

05:01

😎 Free DALL-E 3 Access via Bing for Basic Image Creation

Paragraph 2 highlights that Microsoft's Bing search engine provides free access to DALL-E 3 powered image generation, allowing users to create 100 images per day without paying. It compares image results from the same prompt in Bing vs ChatGPT, showing how Bing can sometimes generate copyrighted content that ChatGPT avoids. The integration with Microsoft Designer for further image editing is also noted.

Keywords

💡Dall-E 3

Dall-E 3 is the latest version of OpenAI's text-to-image generation model. In the video, it is mentioned that Dall-E 3 is now available in ChatGPT for Plus users. This allows users to generate images by typing text prompts right inside ChatGPT, without needing a separate platform. Dall-E 3 is more advanced than previous versions and can help refine prompts to create better quality images.

💡ChatGPT

ChatGPT is OpenAI's conversational AI system. In the video, ChatGPT is mentioned as the platform where Dall-E 3 is now integrated for Plus users. Users can have a back-and-forth conversation with ChatGPT to iteratively edit and refine image prompts before generating them with Dall-E 3 built-in.

💡Prompt engineering

Prompt engineering refers to the process of crafting the text prompts for AI image generation models like Dall-E. The video mentions how Dall-E 3 in ChatGPT makes prompt engineering easier by providing suggestions and automatically refining prompts. Users no longer have to come up with perfect prompts themselves.

💡Microsoft Bing

Microsoft owns part of OpenAI and has integrated Dall-E 3 into its Bing search engine for free image generation. The video shows how Bing can also be used to generate images with Dall-E 3, providing an alternative for those without ChatGPT access.

💡Image resolution

The video demonstrates how users can specify prompt commands to change the image resolution and dimensions for downloads from Dall-E 3. This includes options like square, horizontal, and vertical formats.

💡Downloading images

The Dall-E 3 integration allows users to download generated images directly from within ChatGPT. There is a download button provided for each image, saving them to the local device.

💡Adding text

A new capability showcased is adding custom text to images generated with Dall-E 3. The video shows an example prompt of overlaying text on one of the created images.

💡Copyright rules

When discussing limitations, the video mentions strict copyright restrictions enforced by ChatGPT's Dall-E 3 integration. Users cannot create images that infringe on existing copyrights and intellectual property.

💡Comparison to Midjourney

Midjourney is cited as another leading AI image generation platform. The creator recommends checking out Midjourney to compare capabilities with Dall-E 3 in ChatGPT.

💡ChatGPT message limit

When testing the Dall-E 3 system, the video creator discovers usage counts against ChatGPT's overall 50 message per 3 hours limit. So Dall-E 3 prompts are constrained by the same caps as regular ChatGPT conversations.

Highlights

Deli 3 is rolling out to Chat GPT Plus users, offering the ability to generate images from text within the chat interface.

Dolly 3 allows users to refine prompts with the help of Chat GPT, eliminating the need for perfect initial prompts.

Microsoft, owning half of OpenAI, integrates Dolly 3 into Bing, allowing free image creation.

Bing's Creative Mode uses Dolly 3 for image generation, showcasing the collaboration between Microsoft and OpenAI.

Chat GPT enhances user experience by suggesting multiple prompt variations for Dolly 3 image generation.

Users can refine generated images by providing feedback directly in Chat GPT.

The integration of Dolly 3 in Chat GPT supports direct downloads of generated images.

Chat GPT offers different resolution options for images, including square, horizontal, and vertical formats.

Comparison between Bing's and Chat GPT's outputs demonstrates the versatility of Dolly 3 across platforms.

Microsoft Designer is mentioned as a tool for further refining images generated by Bing's integration of Dolly 3.

Chat GPT's new capabilities include adding text to images, improving over previous limitations.

Dolly 3 in Chat GPT adheres to strict copyright rules, limiting the creation of copyrighted content.

Despite copyright restrictions in Chat GPT, Bing is able to generate images in specific styles like Pixar.

Chat GPT's usage limit for Dolly 3 matches its general limit of 50 messages every 3 hours.

The video compares Dolly 3 with Mid Journey, another leading text-to-image platform, providing a beginner's tutorial.

Transcripts

Deli 3 is finally rolling out to more

chat GPT plus users I just got access

today and there's an option to get it

for free if you don't have chat GPT plus

now with Dolly 3 you could generate

images just by typing in text but now

you could do it right inside of chat GPT

there is not a different platform that

you need to go to but what Dolly 3 does

that makes the biggest difference is you

could actually type in any version of a

prompt and chat GPT will help you really

refine that prompt so you no longer have

to come up with the perfect prompt to

get really good images out of Dolly and

you could go back and forth inside of

chant GPT to really edit and refine that

image and you could even add text now

let me show you the free version of it I

mentioned so Microsoft the company that

owns bing.com the search engine actually

owns half of openai the company that

owns chat GPT so because of that if you

go to bing.com neww press chat now it

will bring you to the the AI chatbot

here under the more creative mode you

could actually create images here inside

of Bing and that uses now Dolly 3 and

here I'll just jump into the blog post

because they have some examples here so

this is basically Dolly 2 this is Dolly

3 but here's a prompt I'm going to just

copy this prompt and I'll paste this

here into Bing and this uses the Bing

image Creator here but you'll see in a

second that it's actually using Dolly 3

behind the scenes it gives you a little

logo here so let me let this finish up

so you can see

I got four different images here and

right there powered by Dolly 3 so it's

made by beinging image Creator but

powered by Dolly 3 the only thing is

I'll show you the difference inside of

chat GPT it actually helps you see the

prompts that it's coming up with so you

don't have to come up with a prompt like

this you could just say something much

simpler than that and chat GPT will do

the prompting but the side by-side

comparison is actually very close on the

output side the prompting though is very

useful so I'll go back to chat GPT to

show you the rest of what's possible and

if you don't have access to what I'm

about to show you inside of chat GPT

again you could use Bing here completely

for free and this is based on a 100 fast

Generations I believe you could do per

day and then it just kind of slows it

down but it's really unlimited on how

many of these you could generate inside

of Bing so inside of CH GPT first you

need chat GPT Plus in order to get Dolly

3 and it's under the chat GPT 4 dropdown

so this GPT 4 drop down you'll see Dolly

3 you actually don't have to turn it on

in settings like these other options I

had to turn these on inside of my

settings menu this one just appears so

this is again rolling out so as I'm

recording this not everyone has it even

if you have the paid version so if you

don't see it here that just means you

don't have it and you have to wait a

little bit more there's no way of

basically forcing them to give you this

Unfortunately they just roll this out a

little bit slower sometimes just like

any other updates they usually do so as

as long as I have this checked on here

next to it as soon as I type in a prompt

it's going to access Dolly 3 in the

background so here's how simple this is

and I'll show you a couple of different

examples and some tips here that you

need to know about to use this more

effectively so I'm just going to say

create an image of a moon with a smiley

face and look what happens with Dolly 3

up here it actually starts crafting the

perfect prompt for you so here's one

here's another I could see illustration

of a cartoon Moon photo of a full moon

in night sky oil painting of a large

detailed Moon Vector graphic of a

minimalistic moon so these are all

created inside of chat

GPT and then it's going to power Dolly 3

to create them and here I got four

Images now you could see it gave me four

completely different versions of this so

then I could refine it further here is

what's really interesting about this so

this is how I count these this is one

this is two this is three and and this

is four so if I want more like number

three I could say I like three but make

it even more Smiley so it took number

three which is this one the oil painting

and it gave me a much more Smiley

version of this and a couple things you

should note the image here could be

downloaded with this little icon here so

if you press this it's going to put it

right inside of your download folder and

it's available for all the Creations so

you don't have to actually get it to

upres anything for you you could just

press download over here now let me show

you the size because that's important

right now this created it in a square

format and the dimensions are 1024x 1024

so that's a square shape this is your

resolution here and the file size is 1.5

megabytes so it's a pretty good

resolution but you could actually change

this resolution with a simple prompt so

I could say for example make a

horizontal version and here's a

horizontal version so let me download

this to show you the file size and the

dimensions you could also do vertical so

if you ask for a vertical version it

just flips these two numbers so it

becomes 1024 by 792 and this one you can

see it's a bigger file 2.7 megabytes so

that's a really quick way to actually

change the dimensions here with a single

prompt and I asked how many resolutions

it could make so those are the three

different resolution your Square your

wide angle the horizontal and the

vertical the tall version of it and

these are all the different dimensions

by pixel size now for this next prompt

I'm actually going to take a prompt that

I've previously made inside a mid

Journey so I said create this and I

typed in this prompt over here let me

generate this and then I want to show

you a couple other tips here that you

could use to refine this a little bit

more and here's four different images

that I got over here I could click on

any of them to see it bigger here I

could copy any of the prompts that it

created from this page and I could

always press download right over here

this is going to download it to my

computer from from there and I could do

this with all four now let me copy this

prompt and let's take it to Bing here

let's see what Bing comes up with so I'm

going to say create this and I'm going

to type in that prompt let's see how

close this is so let's look at the

difference this is what Bing gave me

again it says powered by Dolly 3 and

this is what chat GPT gave me so this is

a little bit more contrasty uh the

characters a little bit different people

are seated at different positions so you

could refine this further for example I

could then take this prompt and say hey

don't have a table for example or have

people standing around in the library I

could say reduce the number of robots

maybe I just want the one robot it looks

like you create a little R2-D2 type Star

Wars character over here and again you

could download these from Bing as well

just by pressing this right here or you

could press customize here and this

actually takes you to Microsoft designer

which is really interesting Microsoft

designer is a whole different tool that

lets you actually refine images and

create different graphics with it so

using this in beinging is kind of really

interesting as you can see here is

giving me some more options for prompts

or turning it into a poster for example

here you can see all these different

types of Graphics it gives me but inside

of chat GPT again I could refine this by

adding text to it so I could say add

text to number one and this will add a

text to this first image here this one

was created in the horizontal format by

default but if I want to make it Square

I could say make give me a square

version here now adding text before was

just not possible at all it did did a

terrible job dolly2 did a bad job mid

Journey did a bad job but now it's

getting a whole lot better there we go

the future is here very nice right no

typos this is the best part of this

platform is you could refine every image

here with a simple prompt and it will

design that prompt for you in a way that

chat GPT and Dolly understand it best

now one thing to note about Dolly 3

inside of Chan GPT it actually has very

strict copyright rules so I said create

a Pixar looking character if you ask it

anything that has any copyright issues

it's going to basically try and then

stop and it's going to say you can't do

that sorry I can't do that because of uh

content policy restrictions which is

copyright issues so this is one thing

that you just have to deal with inside

of chat GPT as of right now they have

very strict copyrights and you can't

also create things in a certain style so

that's something you could do in mid

Journey For example you could say create

this in a Salvador Dolly style and this

won't be able to do that but if I take

the exact same prompt that chat GPT

basically gave me an error message on

and I take it to Bing somehow Bing could

do exactly that illustration of a Pixar

style character exploring a dense jungle

and he created one for me back inside of

chat GPT though I just basically removed

Pixar created a 3D character in a jungle

and it created some for me and as far as

how many images you could create inside

a chat GPT the limit right now from my

testing is same as the gp4 limit which

is 50 messages every 3 hours so if I

pick Dolly 3 again I'm going to cap out

here espcially if I use these in

combination I still have this 50 message

that's not just exclusive to Dolly 3

it's just exclusive to all of chat GPT

usage at the time that I'm recording

this video and I also recommend checking

out mid Journey if you haven't that also

is one of the leading text to image

generation platforms so I have a

beginner's tutorial so that way you

could compare what you're getting out of

Dolly 3 versus mid Journey the link is

included in the description as well I'll

see you on the next video