Is Adobe Firefly better than Midjourney and Stable Diffusion?

Summary

The video discusses Adobe Firefly as a new generative AI tool that can analyze images, text, and user preferences to create customized designs within the Adobe Creative Cloud ecosystem. The host Maria Prohazka runs comparisons with leading AI art tools Midjourney and Stable Diffusion to evaluate if Firefly is a formidable competitor in this space. Prohazka first highlights Firefly's key capabilities - it can generate images, vectors, patterns, brushes, and more based on textual descriptions, making design ideation rapid. She sees strong potential in upcoming vector, sketch, and image extension capabilities. The tool also allows easy tweaking of style, composition, and other attributes. Comparing sample outputs, Prohazka finds Midjourney and Stable Diffusion generate more polished images, especially for complex figures like human portraits. However, Firefly allows finer prompt tuning, works natively in Creative Cloud, and doesn't need coding expertise. So for professional creatives using Adobe tools, it enables tighter integration in existing workflows. Prohazka sees all 3 tools having unique strengths in the AI art space - Midjourney for stunning conceptual art, Stable Diffusion for character design/comics/animation due to custom model training, and Firefly for ideation-to-completion for Adobe users. While casual users may prefer Midjourney, Firefly brings generative design firmly into the professional creative's toolkit. Its upcoming capabilities like vectors and image extensions also promise to redefine artistic workflows. So while it has some catching up to do on image quality, Firefly marks a significant milestone in making AI a force multiplier for human creativity.

Takeaways

• Adobe Firefly is Adobe's new AI generative design tool that can quickly create unique designs
• Firefly analyzes input data like images and text to generate new designs in desired styles
• Firefly shows potential to transform creative workflows and open new design possibilities
• Firefly does not yet match Midjourney or Stable Diffusion in raw image generation quality
• Firefly has powerful upcoming text-to-image and text-to-vector capabilities
• Firefly offers helpful design suggestion panels to refine outputs
• Stable Diffusion remains strongest for tasks like animation and comic design
• Midjourney excels for stunning, artistic imagery with little effort
• Firefly aims to deeply integrate into existing Creative Cloud workflows
• The prompt cheat sheet can significantly improve Midjourney and Stable Diffusion prompts

Q & A

What types of input can Firefly analyze to generate new designs?

Firefly can analyze images, text, user preferences, and other data to inform its AI-generated design outputs.

What unique capabilities does Firefly offer compared to Midjourney and Stable Diffusion?

Firefly will soon offer text-to-image, text-to-vector, extended image generation, and other capabilities those tools lack.

How could the prompt cheat sheet help Midjourney and Stable Diffusion users?

The cheat sheet allows systematically structuring and remembering detailed prompt parameters like styles, focal points, and more for higher quality outputs.

Why might professionals prefer Firefly over consumer-focused tools?

Firefly aims to integrate AI generation into existing Creative Cloud apps like Photoshop for more seamless professional creative workflows.

What tasks is Stable Diffusion still superior for compared to Midjourney and Firefly?

Stable Diffusion remains superior for detailed control and customization needs like animation, comics, and creating assets with specific people or objects.

How does Firefly simplify the AI generation process compared to other tools?

Firefly provides suggestion panels to refine outputs instead of needing manually tuned text prompts like Midjourney and Stable Diffusion.

What unique text generation capabilities does Firefly offer?

Firefly can generate styled text as transparent PNGs to layer into other designs, which other AI tools cannot yet match.

How might Firefly compete with existing stock image libraries?

Firefly's ability to generate high quality photos and graphics on demand may compete with paid stock image sites like Shutterstock.

What are some key upcoming Firefly capabilities to watch for?

Notable upcoming capabilities include text-to-vector, sketch-to-image, 3D-to-image, and personalized outputs based on custom data.

Why is RunDiffusion recommended for using Stable Diffusion?

RunDiffusion simplifies getting started with extensions like DreamBooth without coding, updating, or troubleshooting complications that often arise.

Outlines

00:00

Introducing Adobe Firefly and Comparing it to Midjourney and Stable Diffusion

The introductory paragraph sets the stage for a comparison of Adobe Firefly to Midjourney and Stable Diffusion. It introduces Adobe Firefly as a new generative AI tool from Adobe and hints that it could be a strong competitor to existing solutions like Midjourney and Stable Diffusion.

05:01

Overview of Adobe Firefly's Capabilities

This paragraph provides an overview of Adobe Firefly's capabilities. It highlights that Firefly uses machine learning to generate unique designs from various inputs like images, text and user preferences. The goal is to quickly create personalized designs that match the desired style.

10:02

First Impressions of Adobe Firefly

The narrator shares their initial skeptical impressions of Adobe Firefly, expecting it to be lackluster since it's Adobe's first generative AI tool. However, they were surprised by the high quality output, suggesting Firefly could threaten existing stock image providers.

15:03

Firefly vs. Midjourney and Stable Diffusion

This section makes direct feature and output comparisons between Adobe Firefly, Midjourney and Stable Diffusion. It notes Firefly's simple interface but powerful styling options, Midjourney's stunning image generation, and Stable Diffusion's unmatched photorealistic people generation with custom models.

20:05

Exciting Upcoming Firefly Features

The narrator highlights exciting upcoming features for Adobe Firefly, including text-to-vector, sketch-to-image, 3D-to-image and more. These additions will make Firefly even more valuable for creative workflows.

Using Firefly's Text Effects

A demo of Firefly's text effects feature shows how users can generate stylized text by providing a text phrase and description. The narrator highlights options like changing fonts, styles, and background to customize the text design.

Firefly vs Midjourney vs Stable Diffusion - Conclusion

The concluding paragraph gives final thoughts on how Firefly compares to Midjourney and Stable Diffusion. It notes Midjourney's ease of use for beginners, Stable Diffusion's power for professionals, and Firefly's potential as an all-in-one solution for creatives already using Adobe tools.

Keywords

💡Adobe Firefly

Adobe Firefly is Adobe's new AI generative design tool that can analyze input data like images, text, and user preferences to automatically generate new designs. It is presented in the video as a potential competitor to existing tools like Midjourney and Stable Diffusion.

💡Midjourney

Midjourney is an existing popular AI art generator. The video compares Adobe Firefly to Midjourney in terms of ease of use, quality of output, and overall capabilities.

💡Stable Diffusion

Stable Diffusion is another leading AI art generator. Like Midjourney, Adobe Firefly is compared to it in the video in terms of features and use cases.

💡text to image

One of the upcoming capabilities of Adobe Firefly is text to image generation where text descriptions can be turned into images. This is not yet possible in Midjourney or Stable Diffusion.

💡generative AI

The overarching category of AI tools like Adobe Firefly, Midjourney and Stable Diffusion. They use AI/ML to generate new images, designs, art etc based on input data.

💡design tool

Adobe Firefly is positioned as a design tool for creatives to quickly generate designs. Its integration with Creative Cloud makes it appealing for designers.

💡vector graphics

One of Firefly's upcoming features is generating vector designs from text descriptions. This will be valuable for creatives working with vector formats.

💡text effects

A current capability in Adobe Firefly to generate stylized text effects and fonts using AI. This makes it stand out from Midjourney and Stable Diffusion.

💡custom models

Stable Diffusion allows users to train custom AI models on their own data. This gives more control over output styles and is cited as a key advantage over Midjourney.

💡RunDiffusion

RunDiffusion is promoted in the video as an easy way to run Stable Diffusion without needing to code. It makes Stable Diffusion more accessible to general creatives.

Highlights

Adobe Firefly is a generative AI tool that helps designers quickly generate unique, personalized designs using machine learning.

Firefly analyzes images, text and user preferences to generate new designs in a desired style.

The cheat sheet for writing prompts in Midjourney and Stable Diffusion will change how you create prompts.

Adobe's photo realistic renders are very good - they are easy to work with and control.

Much of the power in Adobe Firefly comes from tweaking parameters on the right side of the interface.

With text effects in Firefly you can generate text designs and download them as transparent PNGs.

Exciting upcoming Firefly features include re-colorization, personalized results, text to vector, sketch to image, and more.

Midjourney creates stunning imagery easily, while Stable Diffusion excels at close-ups of people with trained models.

Firefly runs in the cloud with no setup, generates great images, and will soon have vector capabilities - a game changer.

Stable Diffusion remains best for comics, animation, specific people/objects since you can train custom models.

With solutions like RunDiffusion, Stable Diffusion is extremely powerful but harder to learn than Midjourney or Firefly.

The Multi Control Net in Stable Diffusion lets you tweak the angle, composition, and style of generated images.

Adobe Firefly has potential to transform creative workflows - it complements Midjourney's ease of use and Stable Diffusion's customizability.

Sign up for the Adobe Firefly beta to experience the new possibilities it opens up for innovation and creativity.

Creativity doesn't wait for the perfect moment - start exploring Adobe Firefly's generative capabilities now.

Transcripts

Are you wondering about Adobe Firefly and how it measures up to Midjourney

and Stable Diffusion?

Is it a clear competitor?

Well, I'll get into that in a moment.

Also, I'm sharing a great cheat sheet with you for Midjurney and Stable,

Diffusion that will change the way you write prompts.

It's a new cheat sheet and you don't want to miss out on this.

Is Adobe Firefly the new big competitor to Midjourney A.I.?

And what about Stable Diffusion?

Welcome to our in-depth video comparison of Adobe Firefly with Midjourney

and Stable Diffusion, three cutting edge design tools that are transforming

the digital art landscape, leaving people in awe.

Finally, Adobe is getting into A.I!

About time, some would say.

As Julie Andrews puts it, a queen is never late.

Everyone else is simply early.

So you are probably here because you are already using A.I.

tools like Midjourney and Stable Diffusion,

and you want to see whether Adobe Firefly is worth all the fuss

and can measure up to those very powerful solutions.

And the answer?

Well, it's coming up in a bit.

It's supercalifragilisticexpialidocious.

I went into this thinking, Oh, well, it's Adobe.

It's their first try.

It's probably not very good.

Well, was I in for a surprise?

Because Adobe Firefly is good.

And as one of my colleagues here in the company put it.

Shutterstock should be shaking in their boots right now.

And it's true.

I do foresee that Adobe Firefly is going to be a big competitor

to companies like Shutterstock and Freepik,

because Holy Merde, Adobe's photos realistic renders are good.

They're realistic, they're easy to work with, and you can control them.

Adobe Firefly is a generative AI tool that is designed to help designers

and creatives generate unique and personalized designs quickly and easily.

Using cutting edge machine learning techniques, Firefly can actually analyze

a wide range of input data, including images, text and user preferences,

and use that data to generate new designs that match the desired output style.

Right.

So the cheat sheet I want to show you is something I downloaded from GumRoad.

See the link in the description below for a 30% discount on that.

So I go and open the downloaded PDF right here.

I'm going to open that with my Google Chrome app.

That is going to load this awesome PDF up.

And what I do here is I am going to write in here detailed scene

description “Woman on a motor

bike” and maybe I should write “a woman

wearing a black

the biker biker

jackets on a motorbike”.

And then - this is the cool stuff.

Every time I click somewhere on the PDF, this is going to refresh.

And down here, in the end, I will be able to copy my prompt from that.

So I can go with an aerial photo.

I can go with architecture.

I can go with everything you like.

So I once told you that I needed a list when I create my prompts.

I sort of needed a list,

a checklist for all the stuff that I would like to have in my image.

And this is my checklist, so it's pretty cool.

I can go with aerial photography.

I can choose whatever I would like here.

I like my Sony. Actually, I use Sony Alpha

A7mark iii

for creating what I what you're looking at right now.

So I love that camera.

I can go with specific lenses.

Now I know that if you use a wide lens for this,

this is well, actually 50 is what I'm using, I think.

But I can go with like large lenses and see what that produces.

I can go with f stop.

That is like if I want it to blur a bit so I can create

sort of a really low f stop that's ridiculously low.

So let's try that.

And I can go with balanced symmetry or focal point.

Well, it should be, I guess, a medium shot 3.1..

I can go with all sorts of things.

Question is, does this work when you put it into, for instance,

Midjourney or Stable DIffusion And that is what we're going to test. So

let's go and

say a panoramic overhead view.

And every time I click in here - as you can see

this produces

something that I can just copy.

Let's see how this works in Adobe Firefly.

So I'm just going to go in here

and I'm going to click the Generate and I have my copied text, right?

So I'm just going to throw that in here.

And while that generates, I can go into my discord here,

I can say slash imagine and I can

create my prompt in here.

See what this does, right?

I'm just going to hit enter and see what this comes up with.

This is going to change how you write prompts

because this will make it way easier for you

to actually remember all this stuff and to tweak all this stuff.

Because sometimes when I create my prompts, I keep forgetting stuff.

And here you can actually work with the prompt on forehand

- before putting it into Midjourney or Stable Diffusion.

And it works for Stable Diffusion too.

And you can see this does create something that is, well,

absolutely awesome.

Oh, and we're done. Let's see.

Wow, wow, wow.

Just.

Oh, my God. Wow.

As you can see, this is Adobe Firefly.

So in here,

well, you can see they're not as good,

but there is something I would like to show you.

Don't sign off Adobe Firefly just yet because

there's not so much going on in the prompt field

/ box as you're used to in Midjourney or Stable Diffusion.

Actually a lot of the action that goes on in Adobe

Firefly goes on over here to the ritgh.

Over here you can choose

what aspect rate you would like.

You can choose styles, throw like lots of styles, right?

So movement styles, steampunk, whatever, digital art.

Well, let's just see.

I'm looking for a hyper realistic photo and you can see down here

that is going to throw in some styles on top of that

and then I can just close the drop down.

There are lots of stuff.

You can go in here with techniques too, and you can add that in here.

And that actually really works well for creatives.

I think. So

one thing that I really like about

Adobe Firefly is that you can go in here

and you know, turn on the style, concept

and type of content, color, tone and composition.

Let's generate this again and actually again.

I found that with Adobe Firefly, the simpler

you write the prompt, the better the output.

As you can see, this creates something really nice as well.

So what about Stable Diffusion?

Well, I'm just going to get to that right now because in here

I have my RunDiffusion

and I'm going to get started with that.

So what is RunDiffusion?

RunDiffusion is a no-code solution for running Stable Diffusion.

It runs in your web browser.

It's an online service for creatives like me

who don't want the trouble of dealing with broken code all the time.

I just want my setup to work,

to be up to date, and to be easy to use and fast to use.

With all the new extensions, like multi control and dream booth.

Be sure to check the link in the description below to get 50% off

when you sign up for Creators Club, which provides additional features

like the dreambooth, LoRAS training and a 100 gigabytes of disk

space to save all your extensions, work and models.

And let's just see, let's go back to my Adobe Firefly

and then let's just twirl this up again.

The hands are horrible on these,

but the hands are

horrible.

Let's just jump to in here

and then just give this another go.

So again, imagine

because maybe it was just a lucky shot, right?

I have to say, it was my first try.

I did not do sort of like a thousand takes on this.

This was the first prompt I put into Midjourney

and I just waited for it to create and it just came up

with something that was completely mind blowing.

Also, let me just know that I am

running - right now - I am running the version five

and it does of course create better hands than we have seen before.

And way better eyes.

So obviously it is going to create something

that looks absolutely craz good.

While that generates - this is addictive, right?

It's addictive - I'm just going to show you

while this generates, I'm just going to go back in Adobe Firefly.

So in here I just want to say they have some very exciting things coming up.

So we have text to image.

We have text effects, which is absolutely crazy.

And I will get back to that in a moment.

And upcoming is re-colorization of vectors and that's going to be awesome.

Also, you can get inspired and you can see the video.

But the really cool stuff is here we have in painting coming up,

we have personalized results coming up

that is sort of to generate image based on your own objects or styles.

And that sounds in my ears a lot like training.

So then we have the text to Vector, which I am looking so much for what to because

I use

vector all the time when I work with clients.

Stuff just has to be in vector when I use them for animation videos

as well.

So I am looking very much forward to the text to vector

and we have the extended image that sounds a lot like our painting to me.

Yes it does. And that's coming up soon.

And we have the 3D two image, which is also cool.

You can bring in 3D object and that is going to try

and create an image from that 3D object that's going to be so cool.

Also text pattern.

A lot of people will

probably want to check that out because if you work with Adobe already,

you know that a lot of designers use patterns all the time,

so that's going to be cool. I'm probably not going to use it that much.

Then we have the text to brush coming up

and that is where you just write in text what you want the brush to look like

and it's going to create a brush for you

to use in Adobe Illustrator and Adobe Photoshop.

That's going to be cool.

Then we have sketch to image and I am looking so much forward to that.

That's going to turn my world around

when I can just create sketches and get Adobe to recolor them.

Right now I'm doing that and stable diffusion of course,

then we have the text template.

I'm probably not going to use that, but people who create

or print designs can probably make a lot of use - eh, good use of that.

So I'm just going to show you

on that note, I am getting back to let me just put on a

let me just put on a Stable Diffusion prompt and then get back

into showing you the text effects because that's pretty awesome.

So in here I am going to write actually, I'm going to write,

I'm going to take my own trait model on me,

and then I'm going to take the prompt from before.

And instead I'm going to change it because I've trained this on “sks

person” and not woman.

And we need to take out five because that is from, of course, right?

So in here I'm just going to write a lot of stuff

that I normally write with negative prompts.

You can use negative prompt in Midjourney too, but I don't use them that much,

I have to admit, because you don't need to.

In Stable Diffusion, well you do need to.

And then I'm just going to tweak - oops, some of the settings in here.

Because of course, with Stable Diffusion,

it's not a plug and play solution like that.

In here I have multi controlnet and

I'm actually going to use an image

that I'm going to use as input for this

because I want let's

just say I want something like

this maybe.

So I'm going to enable that.

I'm going to use the Canny model.

And for those of you who have not yet tried controlNet,

check out this great video that I did on that topic.

It's it's not easy to learn, but it's cool to learn.

Okay, so this all looks good.

Let's just give it a smaller weight.

And while this generates, I am going to jump into Adobe

Firefly again because I do want to show you the text effects.

And then here I can write my text

and over here I can write the description of what I want the text to look like.

I want something that is modern with an extrinsic details.

Um, black and yellow.

Yeah.

And I'm going to click generate and now the magic happens because

when you are in here, as you can see, we have a lot of stuff going on.

It's rendering.

This is going to render an image right now.

The background of the image is transparent.

So this is going to be sort of a ping format, png-format.

What I really like about this is that you can choose the different letters.

As you can see, I can choose different detail heights.

I really like that one.

When I click that type or style, it is going to change how this all looks.

Also, another thing I really like about this

and let me just mention, of course

it's not possible to create text with Midjourney or Stable Diffusion yet.

So this is really, really a game changer that Adobe has created this.

Of course you still need to do some layering to get this to work.

So it's going to create sort of an alpha layer

that you can put on top of what you have created with A.I..

That's good, but you still need some sort of

Photoshop-ish thing or illustrator thing where you can put in two layers

so you can layer them on top of each other to use this

type tool on top of what you created.

And that's, well, pretty genius of Adobe.

So in here you can see the ease.

They're the same and that works very well for me.

Over here you can change the overall

sort of design of what you have.

You can like use different fonts, which is pretty cool.

There are sort of different types to use.

I like the source sans so that is again going to

when I click that over here, that's going to generate

something new for me and I can just click out of that fun again.

And as you can see now, the entire layout of the text has changed

and I think maybe I like this one better

over here again, in the right

side of that sort of screen, you can go in and you can check out sort of check out

how loosely or tightly this effect will fit to your design.

And that is another very cool feature.

As you can see right now, it's medium.

There's a lot of stuff going out over the letters.

I don't like that.

I would like it to be sort of a tighter fit so that it doesn't

have all the intrinsic detail going over sort of the letters.

I like this better.

I like it to be tight.

Also, you can style the background, you can change the background,

you can put a background on it or leave it as it is.

And of course you can refresh it. Right?

New text here, change the overall design.

If you go up here and click some of these, you can view

all of them, then you can change.

Well, we have a sushi design.

I would love to see what that even a spot looks like with sushi.

Let's just try that.

That's going to change my design again.

Again, those of you who don't know me, my name is Maria Prohazka

I am founder and creator of the visual agency, Levende Streg.

It’s a Danish word.

I am in Copenhagen, Denmark, Scandinavia.

Levende Streg means lines that are brought to life or lively lines.

I am an illustrator, an animator, and now I am also an AI artist.

And here on my channel

you can follow my never ending quest to adapt new technologies

and use them in my professional workflow as superpowers.

I’m brining it to you,

so that other creatives can learn and do the same stuff that I do.

Don't forget to hit that subscribe button.

We have lots of cool stuff coming up

in the landscape of A.I.,

so right now I have the text as sushi.

I'm not quite sure that the people I work with

are going to like this version, but it is pretty cool.

I can click here to download it,

right?

So I now have Levende Streg as sushi.

Oh, no, no, no, wait.

The popcorn.

I love popcorn.

Yes, I like it.

The popcorn are great.

So I'm just going to leave that rendering.

I'm going to go back into my Stable Diffusion,

RunDiffusion and see what that looks like.

Oh, would you look at that?

Is it absolutely crazy?

Okay, so the details of the face is not good, but.

Well, I got what I asked for.

When you use that, the cool thing about using these,

the multi control network is that you will be able to control the output,

the sort of the angle of view you have off.

I forgot to sign in.

So one of the cool things about this as well

is that you actually with RunDiffusion, you get this cool file

browser and over here it sort of saves the images you have

and you can import models and you can train models.

And as you can see up here, we have the open pose editor.

We have dream Booth.

We have sort of

all the newest extensions installed and I don't have to do any of the coding.

I can just leave it up to the people who are way better at that stuff.

I'm good at art creating.

Oh, I need to see what's going on with my Midjourney.

Right.

So Midourney is still like

mind blowing.

Absolutely beautiful images, right?

So to check out this prompt editor.

I can just reset it, I can go and click again,

see what I want, lots of techniques and you can go in

and find all these cool techniques for painting and styles

and it's absolutely mind blowing what this prompts

Generator can do for your pormpts in Midjourney or Stable Diffusion.

And let me just run this again, by the way, if you're wondering what

the Canny model does, it sort of creates sort of a scribble thing where A.I.

or Stable Diffusion can figure out what kind of type of visual I would like.

Stable Diffusion is

still better at creating close ups of people.

It's like crazy good at doing that with trained models.

So if I like change this model to...

let's say I wanted to change to the deliberate model,

that's a really nice model.

Then I would just instead say here “woman” like we did in the first prompt.

Because this is not a model I trained.

Actually, there is a cool model called Midjourney,

and it does create sort of more of a Midjourney like look, as you can see.

Yeah, but they are really good.

Those images would you look at that!

Absolutely beautiful.

And well, the text to image generator

in Adobe Firefly it's not as impressive

but when they get this entire workspace, figure it out...

...this is going to be like crazy for creatives.

So is Adobe Firefly a match for Midjourney or Stable Diffusion?

Well, it depends on what you are looking to achieve.

I am guessing that a lot of people will stick to Midjourney

if they are already familiar with that solution.

Maybe they dabbling arts and they want to be creative

because it creates stunning imagery.

It is in the cloud and it's easy to use.

But if you are a professional, creative and already working with Adobe

Creative Cloud like I am, this is going to be crazy powerful

because you will be able to start your idea, process and work process

in Adobe Firefly.

It runs in the cloud.

It's easy to use - no setup, no code, It creates great imagery and soon

you will be able to generate vector designs and vector patterns.

It's going to be a game changer

and I can't wait to see how awesome that will be.

And about Stable Diffusion.

Well, it's still the best you can get for something like

comic book creation or animation, character design,

commercial campaigns and stuff like that, where you have to show a specific person

or specific object as it is because you can train Stable Diffusion

and create your own models.

And that is just so powerful.

And with Stable Diffution you can even create gifs and video.

It's still a new technology, but Stable Diffusion is the hardest to learn

and the best to learn.

It is crazy complicated with all the colors you need to install

and sometimes you need to reinstall it because it breaks.

But that is why RunDiffusion is a great solution to me.

Adobe Firefly, in conclusion,

is an exciting new development in the world of AI and design.

Firefly has the potential to transform the way you work and create

opening up new possibilities for innovation and creativity.

I learned so much from all of you.

Please let me know in the comments below how this worked out for you.

And remember, creativity doesn't wait for the perfect moment,

so go and sign up for the beta testing of Adobe Firefly now.