#community-chat
1 messages ยท Page 2 of 1
Oh and what if we collected marketing data while forcing you to sign up for it? Yeah, why not.
Where does Firefly do its prompting? Very curious to see where you take this. Been on Midjourney since August, 30k+ images... but I also have a CC sub, so very eager to "cheat" on Midjourney and test your capabilities.
Firefly better be good..; here's midjourney v5 alpha
Hey all! You can find the FAQ here: https://firefly.adobe.com/faq
MJ5 still realistic oysters. Was fun to ask for Oysters with v4
luigi disguising in pirate*
MJ5 is now very good at not forgetting any part of the prompt
What does this mean? "We do not currently support tablets or mobile devices." Wouldn't I be able to prompt on iOs Chrome, possibly in Desktop view? I mainly work on an iPad, like most illustrators these days.
a student who is looking at the board
Another midjourney 5, nice faces and clear hands on first prompt
I imagine it would work, yes.
But it isn't optimized for such use and isn't officially supported.
You can't prompt to discord, please check the #announcements and register for Firefly beta here https://firefly.adobe.com/
And answer marketing questions
Hey @zealous cipher, thanks for joining us. Is there something I can help you with?
Firefly is getting pretty solid with faces
tf, girl with black sunglasses is my cousin โน๏ธ
I am here to try Firefly. I have been a paid user for Midjourney, excited to see Adobe version of this. @quaint lily @slim lion
@whole patio you cannot generate images directly in the discord, please check the #faq channel
Hey, welcome @blissful pasture !
Welcome, Erick!
welcome @blissful pasture ๐
Hey @late sail and @honest quiver !
This is so cool. I can't wait to get approved and start checking it out.
Just signed up, will this be integrated seamlessly into Adobe products?
I'm eager to check it out too. I have high hopes.
The Hype train is leaving the station ๐
The plan is to bring some of this tech into apps like Photoshop, Illustrator, etc
checkout the #faq @meager frigate !
Adobe finally came with an image generator AI
Oh thanks!
Logรณtipo
Not sure bud , let's see
they need to eat.
#announcements soon the livestream will start :hype:
๐
waves
https://www.youtube.com/watch?v=c3z9jYtPx-4 for the livestream! It starts in 10 minutes ๐
Today weโre excited to introduce #AdobeFirefly, a new family of creative generative #AI models. Tune in to the live stream to get your questions answered, connect with other creatives and get started with Adobe Firefly!
Learn more: adobe.com/firefly
Eric Snowden is the VP of Design at Adobe and is responsible for the product design teams for t...
Simultaneously excited and nervous to see what Adobe does
Hey Gila! Excited to try it out! ๐
I am sooooo ready to cheat on Midjourney... ๐
So far, to a degree. I often get to where I want to go faster creating images I want in Photoshop, then prompting Midjourney. ๐ I am not very good at prompting.
Another tool in the arsenal 
But you could come to prompt-chat. And ask for help! If you know PSD, it is perfect to fix small things, too, where MJ might take too long to get things right.
i'm excited that they would pay those that get trained on, compared to midjourney, who still don't do that despite costing a lot of cash.
I already pay for CC suite, so... yes please. Plus I am guessing their team is bigger, and they have plans for sketch2vector... it would be perfect for me.
Yeah, I was tired of creating fake discord accounts to use Midjourney
This is one example. I had a very hard time getting MidJourney to create a COD mobile ghost lego. So I made it myself in PS. MidJourney is good with popular characters on the internet, but not very good with not-so-famous characters.
so the firefly gonna be in subscription?
@elfin haven check out the #faq channel but: Anyone who has an Adobe ID and is 18 years or older can apply to join the beta and will be approved for access as space is available. Eventually, Firefly will be available in Adobe applications.
Their 3D apps seems to be separate, so wouldnโt surprise me if they done the same for Ai addons
Add Figma to CC too please. haha
figma should be in CC
alright thanks!
Excited to be here
Adobe does not own Figma yet
Wait, what????
"yet" ๐ค
the news is all over .. how do you say that?
Join the party! Livestream is happening now! https://www.youtube.com/watch?v=c3z9jYtPx-4
Today weโre excited to introduce #AdobeFirefly, a new family of creative generative #AI models. Tune in to the live stream to get your questions answered, connect with other creatives and get started with Adobe Firefly!
Learn more: adobe.com/firefly
Eric Snowden is the VP of Design at Adobe and is responsible for the product design teams for t...
i legit wish Substance 3d apps were part of CC lol
Would be cool if you took a lot of people in after the live event so we could amplify the hype on Social Media hehe! ๐
@tough rock we're working to approve ASAP!
Hi
3d render of a black lego figure wearing a skull on blue background:: face::-0.3 --v 5
Let's keep this about Firefly, but the Figma acquisition is still going through the approval processes. No more Figma talk. ๐
Great one, but that's not Ghost though.
U2
I have no clue who ghost is. ๐ I often use MJ as a starting point...
Ok boss.
Exactly, Midjourney doesn't know who Ghost is neither.
Hey everyone!
Welcome!
We totally want to keep this about Firefly! Just cannot play yet. ๐ Plus, I am still unsure if "We do not currently support tablets or mobile devices." means I cannot prompt on iPad.
I'm stoked for this. I was literally about to subscribe to Midjourney lol. I blasted through the trial period. Hope it doesn't take too long to get access to the beta.
Hi guys just wondering is this free btw?
https://www.youtube.com/watch?v=c3z9jYtPx-4 stream just started
Today weโre excited to introduce #AdobeFirefly, a new family of creative generative #AI models. Tune in to the live stream to get your questions answered, connect with other creatives and get started with Adobe Firefly!
Learn more: adobe.com/firefly
Eric Snowden is the VP of Design at Adobe and is responsible for the product design teams for t...
We're trying to find an answer on this. My guess is that since it's currently developed as a desktop experience, it may not work as intended on other devices.
It would be lovely if it were covered as part of CC subscription.
One problem is always when you encounter ambiguity in terms. Is black a color or a person? Is ghost a dead person or a character?
Yay! Btw FAQ says:
If you successfully submit your access request, it may take time for the confirmation email to arrive. You'll receive an additional email when access has been granted.
FAQ says this, but I don't think I got it and I applied 2-3h ago. What should I do?
wait? ๐
Ommm... we shall see. I am fine running in desktop view, but nothing will get me behind my mac to prompt. Ipad is my ride or die.
@tough rock hang tight! if you submitted, we have your application ๐ I know the team is working furiously behind the scenes to get everything firing on all cylinders.
๐ค
Thank you Gus
Right.
Sounds great! I just wanted to make sure my application doesn't get lost. Thanks a lot for you and team!
Totally understand. My guess is that once Firefly is out of beta, more devices will be supported.
Watching the live stream. I'm in awe.
It's pretty incredible!
I will share the replay to the livestream afterwards, for folks that miss it!
Haven't been so hyped since Max ๐คฉ
Especially the vectorized text
YUM
The vectorized text bit is super cool.
Demoing AI tools live is wild, since every generation is completely different. You never know what will happen ๐
Lot of confidence
I am the queen of text on MJ. But this looks sick. ๐
This is correct ๐
wow this UX/UI is super clean
Stock + Sensei + ????
All of the current day image models are amazing ๐ฅณ
i'd just wait i suppose. I also applied few hours ago but i presume they recieved my request even though i didnt get an email either yet.
Just imagine what they'll be like in 5 years ๐คฏ
In case you missed, we're demoing Firefly LIVE: https://www.youtube.com/watch?v=c3z9jYtPx-4
Today weโre excited to introduce #AdobeFirefly, a new family of creative generative #AI models. Tune in to the live stream to get your questions answered, connect with other creatives and get started with Adobe Firefly!
Learn more: adobe.com/firefly
Eric Snowden is the VP of Design at Adobe and is responsible for the product design teams for t...
Very exciting
im listening, it's great!
Hi Gus, could you give me access for the online testing, so i can make a video about it on my youtube channel?
How do you track copyright on this?
My creatives are going to be pounding at the gates for this.
Really excited about the text effects
Looking forward to hearing more about how they'll prove out "Commercially safe"
@rustic wadi We're working as quickly as we can to approve applications, but we're at the mercy of the system right now. You'll know as soon as you're approved!
what i know so far is that nothing can be used commercially until the app is out of beta, and based on current precedent so far (in US law anyways) as long as the artist/designer has modified output creatively enough to the standard of "authorship" then the copyright is yours.
thank you
The request access button isn't working for me...
@toxic flicker: We train our model by collecting diverse image datasets, which have been curated and preprocessed to mitigate against harmful or biased content. We also recognize and respect artistsโ ownership and intellectual property rights. This helps us build datasets that are diverse, ethical, and respectful toward our customers and our community.
Will we need a law degree to know what that creativity looks like?
In just seven days, three remarkable revolutionary models were brought to life; The pace of innovation is so rapid! ๐คฉ
Correct, while we're in beta none of the outputs are ready for commercial use!
Wen can we download a Beta?
check the #announcements channel for a potential solution
Right?! I literally can't keep up these days.
we're working hard to make sure that isn't the case for Firefly ๐
@noble hound please se the announcement. You'll request access and then join the queue.
Makes sense when you're pulling from Stock, but it sounds like you're pulling from additional datasets that aren't already within your ZOA - are you going to be able to validate that content in the same way you validate your own?
haha yes and no. but generally no, lol. overall just how you used the Ai generated output with post-processing edits, but also precedent still being established to recognize that copyright authorship can also be established with complex prompting, or /blend with references of your own original works, etc. its a lot going on legally atm to establish the nuances of "authorship" in ai.
@toxic flicker Firefly was trained on Adobe Stock images, openly licensed content and public domain content, where copyright has expired.
openly licensedas in there is a list somewhere?
"we have licenses with sites X, Y and Z"
This sounds awesome, but my Art Buying and Legal departments are definitely going to look for more details on that. ๐
or does this just mean CC
how long does it take to get the access for firefly, any idea ?
hell
@sudden garnet no estimates but we're working to get applications approved as quickly as possible
i applied this morning and i'm not in yet, therefore at least six hours

Same her, sadly.
@quaint lily thanks bud!
@shut dew I am not well versed on the specifics but we do have some Adobe folks in here who might be able to give additional details. I'll see if someone can chime in (please be patient, very busy day)
When you signup and are waiting is there a confirmation email you get? Didn't receive one when I signed up so just confirming.
Good questions @brazen coral
Same, I've also been waiting since this morning
Maybe I typed my email address wrong lol, and dont want that to be the issue for me waiting lol
i applied to google bard this morning and got in, and can i just say, woof
@brazen coral not everyone has gotten a confirmation email, but if you submitted, we have it!
i've spent the last year wondering why google was sitting on its ass and not releasing any AI stuff
now i have a good idea why
bard is real, real basic
Today Google released a beta for some new AI, right?
I feel like if we are active in the Discord we shoud get early access IMO
Yes
hi everyone
Are poems already working?
yeah its their GPT competitor ... except right now its like Alabama vs North Alabama College of Dentistry
I joined the Bard waitlist earlier, interested to try it out.
bing isn't the best rapper, chat bots have no sense of syllables and scansion, but it can at least rhyme the last word
Live finished, it was amazing
lol, I got the beta the truth is it's not the best nor is it available in my country but VPN saved me
this is going to be amazing!!
hope we get the access in 4 months
So where do we type the prompt?
The live stream has me even more hyped to check it out now
check the #faq @wooden granite
@wooden granite check out the #faq channel! You need to apply for beta access, before getting access to the tool ๐
the music of the live was made with AI?
I love the concept of AI learning your art/design style and creating something based on your direction. Especially if they can get it in to Adobe Illustrator (probably well down the road). That would be a HUGE timesaver, and a really powerful use of AI.
amaaazing video at the end of the livestream!
That preview was amazing
Will the livestream be shown on YT now that itโs finishedโฆ?!? I missed the live ๐คฆ๐ผโโ๏ธ๐คท๐ผโโ๏ธ
โค๏ธ
Any room on the A.I. team for experienced devs/technologists with pixel-processing, imaging and design experience? :D
It should be available on-demand shortly. YouTube likely needs to encode, etc
Today weโre excited to introduce #AdobeFirefly, a new family of creative generative #AI models. Tune in to the live stream to get your questions answered, co...
That Livestream got every one wanting to try Firefly. Such a good presentation!
It's viewable here still
Thanks mane ๐ฌ
The video was incredible; this is a revolutionary moment in the creative field! ๐คฉ
It starts at about 2:10 in
@wary widget we'll add chapters too
All right, awesome!
Wow! Firefly looks fabulous! I watched along with over 900 of my closest friends just now and am amazed and excited!
Will this be integrated in to Adobe Illustrator from day one? Editable vectors would be a game changer
We don't have a timeline on integrations, but it's definitely a priority to weave this tech into our apps, like Photoshop and Illustrator.
I hope editable vectors is a priority!
When did the people with access sign up?
Today!
Here are just a few things we're currently exploring.
Yayy!!!
"A sad cat who has been waiting to get into the Firefly beta"
cat
Firefly is amazing!
did you take that picture of me ?๐
You can't prompt to discord, please check the #announcements and register for Firefly beta at https://firefly.adobe.com/
Yea, sketch to image would be a pretty big deal, along with text to vector.
Great to see you here, Marsha!
Haha that's awesome, I love the relatively consistent artstyle from the results. Is this intentional?
Is the AI based on openai dalle ?
Do you think you can outmatch Midjourney V5 in short-term? and I didn't see any "portraits" around, can you provide some for examples?
so excited for this Howard - Toronto is waiting
how do you get notified after requesting access?
I bet it's stable diffusion
Curious to hear too, but I am guessing Stable Diffusion under the hood somewhere.
Firefly was trained on Adobe Stock images, openly licensed content and public domain content, where copyright has expired. You can learn more here https://firefly.adobe.com/faq
you will receive an email
that's the training, not the AI itself, tho
Thanks!
You will receive an email confirmation with next steps ๐
This one had no additional styles applied, so no idea ๐
Thanks!
it most likely uses diffusion for the processes.
True, I assume stable too but hmm they got inpaint feature which only dalle has I think? Or wait stable does too?
We cannot predict timing at this moment but the team is doing their best to approve requests. Hopefully in the next 24-48 hours
SD has inpaint for eons
I actually just updated my original question to #1087561412269854840 because it seems /mashup will be the alternative to MJ /blend and thats amazing, do you, or @quaint lily know if there will be a cap on how many ref. images can be used?
So crazy how fast this field is advancing ๐
cool, thats not too long ... tnx for the answer โค๏ธ
It may take time for the confirmation email to arrive. You'll receive an additional email when access has been granted.
Its like Leonard?
Hard to say with how fast things move. MidJourney definitely has a strong 'style' behind it, while Firefly is more 'raw'. We're getting pretty good with faces, though!
The FAQ says "At first, Firefly will be integrated into Adobe Express, Photoshop, Illustrator, and Adobe Experience Manager."
Does this mean that it isn't going to be a web-based app?
more raw, uh? that def reads as SD for me, then ๐ค
Too early to know. Those features are still very much in development and change all the time.
I'll keep an eye out, any idea how long the confirmation email might take? (minutes, hours, etc)
Yea pretty much haha
Raw without styles, but once you start experimenting with prompts/techniques, you can get some really nice outputs
i would expect so in a beta certainly, but def exciting to at least know it's there.
Hi laze ๐
So, in the #1087561263137177640 channel, it references the URL https://community.adobe.com/t5/adobe-firefly-beta/ct-p/ct-adobe-firefly-beta but I can't hit it, even with my Adobe login (I'm an employee). "Access Denied" and This page isn't available. Try checking the URL or visiting a different page. I DO have access to Firefly.
Adobe ID
Hey rendo ๐
I did not say it as a bad thing. I just like knowing what's under the hood. ๐
hi everyone ๐
Cannot estimate timing, it depends on the system as a lot of requests are being processed atm. The team is doing their best to get you access asap. ๐
Oh absolutely! ๐
Midjurney have a lot issues with product like and with 10 fingers , when it come to human anatomy or any anatomy even animals it dont give great results at all
Hi Vicky!
hey @slim lion
Gotcha, thanks
oh hiii Vicki!!
Hmm I signed up with a non adobe ID email address.. is it a must for the beta ?
oh! I did not even remember about that either! ๐ฎ
Firefly went through a lot of that in its early days but has improved quite a bit since. And it will only keep getting better.
This is a known issue and team is working on it.
Version 5 looks like it's cleaning up some of those bugs like the fingers
Almost 6500 members holy.
Mj has no api for integration anyway, so its between dalle and SD
I prefer SD anyway. More liberty to do what I want. 8)
and only 6 has access ๐ญ
You need to sign up with you Adobe ID. See more info in #faq
Did we get the answer where the name 'Firefly' actually came from? I saw somebody question it in the livestream.
is it out?
i saw @untold tendon here earlier too.
Yeah
where can I have a taste?
Patience is key
Eric mentioned that it was the 'spark' and 'magic' behind it that lead to the name.
viking rasta futuristic black color with medium scar on face
the son of the marketing chief was playing with fire, burning flies
Ah darn,. I registered with my company email address hoping my beta gets approved faster ๐ Its not linked with an adobe ID.. i will re-sign up with my personal email address in that case, thx!
So glad I accidentally turned on Behance live today, Firefly wasn't even on my radar!
You need to request access at https://firefly.adobe.com/ You can learn more in #faq
ty ty ๐
This is the most impatient server chat Iโve ever seen

Good luck good sir
You've never been to an NFT server, have you? ๐
you're welcome ๐
we refined
I havenโt. Itโs that bad huh?
naaaa, SD's was the same when they opened to testing
So much worse ๐
when stablediffusion first launched we had to wait for days yeah
Man you guys are gonna get an avalanche of prerelease requests!
i felt that lol xD
I saw bunny illustration in V5 was looking like Frankenstein, hands, legs..ears bad modified. They still have big issues.
What if you sign up to the waitlist with your email, and then create an Adobe ID after you signed up to the waitlist? Would you need to sign up to the waitlist again?
Hi everyone, nice to meet you ! Cheers from France ๐ซ๐ท
FireFly was also a Sci-Fi TV series that I loved - so I thought that was the "spark" ๐
V5 gets hands right about 33% of the time, which is extremely impressive for an AI that knows literally nothing about humans or the physical world apart from bitmaps
Lemme get a baguette pls ๐ฅ
Does anyone know the waiting time in general before the access ? What a period to be alive and to be creative !
Lol yes - Browncoats
Sometime around 2024 thinking October/November
Still waiting here too
Thanks you !
every day something new and shiny to learn for us
Not thanks you !
Could be an hour could be two days
heard 24-48 hrs
Do we need to connect with our Adobe ID mail ?
We cannot predict timing at this moment but the team is doing their best to approve requests. Hopefully in the next 24-48 hours
We're working on a solution to get things moving quicker, no promises.. but rest assured, the team is on it! ๐
Thanks everyone for your answer !
How long until it's confirmed that we're NPCs in an advanced video game?
Iโll grab Dunkinโ Donuts for everyone in the server while we wait
You need to sign up with your email, which is associated with your Adobe ID. Learn more in #faq
im already a mutant .....
ChatGPT confirmed it already ๐
Thanks you Lora, i'm running to #faq
If you are eager to experiment in Photoshop, there's also this neat plugin available:

you're welcome
they/them
haha I can relate
there is also the extension for automatic1111: https://github.com/AbdullahAlfaraj/Auto-Photoshop-StableDiffusion-Plugin
Do you guys like Boston crรจme donuts? I love Boston crรจme man
I bet integration with other Adobe apps will happen with this. That would be dope.
the text fx are such a cool spin on it tho!
tbh I haven't tried that but the fastest way to get your access is to sign up with the email address, which is connected to your Adobe ID. The approvals are on the first come, first serve principle, so the earlier you register, the earlier you get your access ๐
Last year I explained to my 70-year-old father the madness of MidJourney, then a few months later that of ChatGPT... He was already having trouble understanding my job as artistic director, so now it's even more crazy for him.
Weโre part of the changing of the world.
It says in the FAQ already:
At first, Firefly will be integrated into Adobe Express, Photoshop, Illustrator, and Adobe Experience Manager. The eventual plan is to bring it to Creative Cloud, Document Cloud, and Experience Cloud applications where content is created and modified.
i love to show people how easy it works
for better or worse?
Close enough ๐
Iโm not but maybe the devs
I need this
hello ๐
hello
now im hangry
Whaddup
Likely bothโฆhopefully leaning toward the better.
Hello
man i didnt expext Adobe to come around the corner with this so fast
Embed fail
expect
That's a Boston Cream Sangwidge. :D
I remember doing the same effect iteratively for a single font for the whole alphabet and numbers! Guess that will soon be a thing of the past! Phew!
but i knew Adobe will do this sooner or later even before they announced those features back in october 2022
money calls
Yes, it's crazy how technology is surely the best way to return to childhood.
Even the less creative in this world have fun saying "Try to make a..." "OH WOW CAN YOU MAKE ME AN AIRDROP OF THIS THING?"
They've secretly been working on it since 1982
We are if weโre using it in its early stages. Not in the same way as devs but weโre a testing pool of sorts. Taking part counts IMO.
take ma monaaaay
heyyyy
It's been a long time coming. We wanted to make sure we tackled this the right way, including this beta to get feedback from the community.
Ok, thanks!
i'm excited for Firefly!
True that
you're welcome
I mean, this is not the first time adobe uses AI, any photographer knows this.
smart move, dodged the bullet that comes those days due to the controversy we know with AI art
see Stability AI, Midjourney and co
im really excited to try this out
AI is the wild west right now. It's tough to keep up
xd
well when i get invited xD
indeed
can you make the chocolate color here pink, and the chocolate sauce white?
How well does it work with fire-y text?
I remember when Content-Aware was first announced. Game changer!
it really was
Iโm craving Taco Bell
Me too, I was in disbelief at how effective it was
yes, that's precisely what I was ref to!
We spent time doing on Adobe what we had in mind... Now we will spend even more time looking for what we have in mind. I notice it with MidJourney, it's like no idea is amazing enough to try. Crazy !
Iโm just hungry in general at this point honestly
It is STILL my fav feature!
I grew up with this. This was game changer for me
haha same
Yeah! I have to imagine this new tech will take Content-Aware to whole new levels.
Microsoft paint cause Iโm an OG
Imagine not being in your 30โs
Same.
I'm almost in my 40's and I say I grew up with PS7, the oldie but goodie, bc I never really used paint xD
Paint was top tier entertainment
additionally if we can get an Ai upscaler 2x, 4x, 6x, 8x, that denoises and refines details etc, like ailab, built into the firefly app that would be ๐
i hope i typed my e mail right for invite
hmmmm
i'm doubting myself if i did that right
hell nah O_o
KIdPix > MS Paint
i didnt get confirmation that i applied for access
this is adobe, ask more! x16!
I once crashed the computer of local library by making 10 000px x 10 000px image with microsoft paint. It just kept calculating. ๐
lol I worry the same thing sometimes
Oh my gosh KidPix!!
HeLLO eVeRyOnE
thats why say this
it can take a moment. there are a lot of people signing up right now ๐
Hi
scale it to a billboard 30x plskthx
Welcome y'all!! If you have cool idea, feature requests, or anything you cant us to see > make sure you share it as a post in the appropriate forum!
now we're talking! 8)
For the horde or alliance? xD
just waiting for a firefly powered meme generator ๐ฎ
KidPix / Kidworks 2 was my childhood
old school og horde from classic sooo. fth!
Just spamming Email refresh over here.
alliance... always ๐
https://kidpix.app/ You can still use kidpix :p
I can smell wow player anywhere ๐คฃ
this is the way. lol
even she was originally ally .. soooooooooooo
Will the service be part of your plan applications?
I've been to 11 BlizzCons. I know the smell well ๐
When I suppose to get beta access, hungry to have my hands on it
24-48 hours
iv played both wow beta and frozen throne beta!
havent been to blizzcon tho... because europe ๐
I feel like discord needs to make pinned messages as a colored banner at the top of chat so people actually see it lol
Any leads lad how much does this service gonna cost?
like IRC?
the beta is free. i don't rly know anything else
I'm going to bundle up go out to check my mailbox, in the snow, 'cause I live in Great White North...eh ๐
Hi guys, is there a special subscription plan that you need to have to access or is it enough to have requested access with the ID and have a monthly plan?
I was Horde player 6y then Bfa happens and make Horde into ...Slaves! So and my heart go with alt on Alliance and better loot ๐คฃ But Sylvana is still my fav char!
whats up gang!
Welcome @turbid gull !
Anyone who has an Adobe ID and is 18 years or older can apply to join the beta and will be approved for access as space is available. Eventually, Firefly will be available in Adobe applications. Learn more in #faq
hoping to get access to firefly!
thanks Gus!!
Got it, being just curious to get to know ๐
Does anyone have trouble with chatGPT right now ? Sorry if it's not the topic here.
Welcome welcome
It was down for me earlier
thank you ๐
Midjourney is going to drop V6 before adobe catches up ๐ U should rename FireFly to maybeasparkTurtle
u wot m8?
If you're just joining, you might be able to find some handy information on Firefly and where to apply to the beta in #faq
I'm personally not worried about MidJourney. Once we integrate Firefly's tech into our apps, it'll be like comparing apples to oranges. Very different usecases.
actually Adobe Firefly seems to instantly beat MJ outta window
ChatGPT was having an issue, also MidJourney had an outage too. dunno if outages causes tho, maybe downdetector something, or just internal updates that went sideways.
Who cares about catching up? More tools probing the Latent Image space is only good for everyone. Let them all "catchup" as it were.
midjourney is fine, dalle2 should be worried
midjourney is terrible for already existing work, and usually way too animated
use cases are more similar and they've been sitting on that next-gen alpha for a while now
Maybe their servers were shook by our announcements. ๐
i bet this guy is from NFT world... such a salty comment xD
haha
im not worried about MJ or SD as a competition to Firefly
[To me] The amazing thing about Firefly is that is improves and enhances creative workflows, as a natural extension of the technology Adobe has been making for 40 years. Opening new workflows, not replacing them ๐ All while being trained on Adobe Stock images, openly licensed content and public domain content, where copyright has expired.
feels pretty cool to me!
" Metaverse & Nfts " i knew it ! ๐
ahahahhahahha
Yeah it seems so
also if i read it correctly Firefly will adapt a way to ban using artists in prompts?
honestly I'm personally just not comfortable with the approach some generative AI services take. just scraping any image on the internet regardless of license doesn't seem fair. that's just one reason why firefly is a win
I cant wait to make a powerpoint and be able to just find a good image in seconds instead of getting something close to what i want from google
Fireflies are dope
im a beginner artist tho, should watch not to forget to learn artistic and editing skills ๐ XD
All this competition talk, I think we can appreciate all these new models ๐
same!
that would've been my second reason
Smart man
preciseely. from the video, I get the impression it will have the capabilities and some features of MidJourney, but it also is like Canva in some ways too, but having these tools that are just built in and integrated properly with the CC apps because its not 3rd party. I think this streamlines everything from design to art, etc.
Do you think this will help the graphic designers ? or will it make them obsolete?
Depends. In my line of work, it could be an immensely useful tool.
Theyโll have to learn to adapt
Industries are always changing
It is what it is
@willow gorge before apps like Photoshop and Illustrator existed, I suppose you could've asked the same question about those apps
it will definitely help designers. i don't think it will replace them tho
I design book covers. If I need, say, a unique looking dagger and I can't find stock or 3D assets of what I need, I could generate it in AI and overpaint it. That would save me so much time and headache.
I personally think it will help the designers are able to figure out how to use it to their advantage.
The goal with Firefly is to help creatives and non-creatives alike bring their ideas into the world with more control, I believe it will help graphic designers, maybe they can spend more time on the story, concept, fit and finish and less time on the minutia.
Which Apps? Because i remember everything was done by hand before photoshop.
it will be another tool in your belt - just a very good one ๐
Got to go run some errands. Congrats guys on the Firefly! Hope I get the beta invite when I get back. Looks super fun to play with 
Thanks for hanging out!
It's like if you a give a bunch of people the paint and brushes - some will be bad, some good, and some brilliant.
Inpainting and outpainting in Photoshop, FTW!
Gotta run pick up the kiddo from school. Catch you all later!
see ya Howard!
byee!
later ๐
I haven't gotten approved yet but I'm so excited to play with it. Do we know if/when images will be available for commercial use? Or is that still a legal issue that has to be ironed out?
oh man i thought i got an email from Adobe...but it was a god damn spamn
damn you spammers
spam*
Sometime around 2027
Idk though on a real note
Iโm a chef not an Adobe employee ๐๐
Exciting times! Cant wait for the approval
Heya! The images are not viable for commercial use during the beta and the goal is to make sure this is the case when we launch (but no timeline as to when that is)
uhh... but it has Stable Diffusion tech under the hood. It just trained with adobe's stuff. Right, @slim lion ?
Or did I misread your replies earlier?
You guys are using SD's AI under the hood. You just trained your model with something that you already had access to thanks to adobe stocks, etc.
I just want to be sure here. ๐
Im really worried about Midjurney and their being transparents to creators and artists. How I understand they dont have own database their data is internet, im right?
@edgy oyster Howard is definitely more of the technical expert than I am, I can tag in @slate solar who might be able to chime in later ๐
hep is in the stable foundation server ๐
i had no idea about behance @slim lion and all the videos i dont know whether to laugh or cry been using ps on mobile for over a year ๐
thanks. ๐
better late than never ๐
i use midjourney for rough concept art, other than that its pretty useless to me with the limited tools they have
chris can you go make me a sandwich please
personally I am mostly worried about the training data. and Firefly was trained on Adobe Stock images, openly licensed content and public domain content, where copyright has expired - unlike other generative Ai services which just used everything
yeah, but like, this is also quite the dogwhistle, bc it's the kind of discourse ppl who are completely against AiArt use. Or have been using against my work online that I've been making with SD, even when I made my own model, trained with my own art.
So I was like... Uhh... but Adobe is using the same tech??? isn't it? am I missing something, here? know what I mean?
Everything being the LAION 5B dataset... Is the training data 100% Adobe Stock then or other sources as well?
"openly licensed" i assume means anything under CC
like things on wikimedia
(there's copyrighted stuff on wikimedia but they're fairly anal about metadata so you could trivially cleave just the CC stuff)
Create images, vectors, videos, and 3D from text with Adobeโs AI art generator. Discover a whole new way to create with Adobe Firefly. Join the beta now.
Even with most Creative Commons stuff, you need to clarify your original source and can't just "freely" use the data.
Without citation
yes, that is why i've been picking at "openly licensed" for a bit
seeing whats in progress is making me so excited lol https://video.tv.adobe.com/v/3416970t1
two words that say a lot and very little
Indeed
Sup
(for the record i think AI training is transformative and it's only an issue when outputs overfit specific works)
but people who aren't me will probably need a lot more detail on what "openly licensed" entails
correct. and the Laion 5b dataset contains "uncurated, large-scale datasets crawled from publically available internet" unlike the training data for Firefly, which comes from Adobe Stock, along with openly licensed work and public domain content where copyright has expired.
Yes. I know what the LAION 5B dataset is. Thanks for the reply. :)
man the features of Firefly...this is brutal
SD, MJ? Get outta here
make room for Firefly
that's the thing, tho. SD being under the hood of firefly means that Adobe is using tech that was made with the help of laion 5B, even if adobe itself is now not using laion 5b for their model. laion 5b was used to make SD. Right?
Does my logic follow or am I missing something here? @slate solar ?
Could they actually do SD without that dataset in a way that would now be useful for you guys to use and make your own model?
anyone here has access?
i think the deal is that stablediffusion is both a model (very similar to most other text-to-AI graphic models, they're all descended from the same colab notebooks bouncing around about two years ago) and a set of weights (the Good Stuffโข๏ธ, the results of a model being trained on a dataset, that actually enables it to generate output)
SD was trained by Laion 5B but that doesn't mean it is an integral part of it
Adobe has nothing to do with SD stuff afaik
including LAION 5B
well except the fact that images were taken from Adobe Stock too
you can train "a" stablediffusion model into a certain set of weights, but it won't be "the" stablediffusion, it'll be totally different in output, etc.
and Behance
wait for the confirmation...
@dusty minnow no, images were not taken from Behance
hello team does anyone have access to adobe firefly to try it out?
Firefly has nothing to do with Stable Diffusion
also i think SD has an infectious RAIL-M license anyway
true
so they couldn't use it for a commercial product
sure?
I think we're splitting hairs here. The point is: Adobe is making a conscious effort to use properly sourced data to train their model. And it seems like they are actively working on including everyone in the conversation going forward.
how can I begin an atrwork?
@dusty minnow We do not train on any Creative Cloud subscribersโ personal content. For Adobe Stock contributors, the content is part of the Firefly training dataset, in accordance with Stock Contributor license agreements.
Hi๐
welcome ๐
And we are working on how to compensate the stock contributors for their work here. More details when we release
I'm going to be obsessively checking my email until I get approved for beta access lol
The integration with Adobe tools will be an incredible win for this tech. Not just a plugin but true integration into features we already know and use.
oh, I think I was being misread all this time. I have nothing against Adobe nor what they're doing. ๐
What I have against is, if they had or have any sort of link with SD's tech, against the hypocrisy of some people. ๐
Once you are granted access, does anyone know how many images you are allowed to generate per day as a new user ?
wait to get rid of the confusion, who is we? Adobe or Stability AI?
there is only one way to find out 
i did it before and i got some mail and...it was a spam mail... lol
adobe. Stability is a different company
lol noooooo
I am going to step out for a bit, but we still have some folks monitoring & helping out! See y'all later, so psyched to get your feedback and build this thing with you ๐
yeah but i dont know if @pine stirrup meant Adobe or Stability AI
I copied from the FAQ. so "We" is Adobe
ah, my bad
i was like there is no way they didnt use images from Behance
so then i thought you are from Stability AI team xD
because you said "we"
oooooooh. gotcha ๐ no!
Looking forward to a time when "Content Aware Fill" has a prompt input field :D
that would be so cool!
I have tried generating hands with Firefly - actually pretty good
ok my brother is shocked about Adobe Firefly, he wants to make our mom being Spongebob
Midjourney's interpretation of digital extremities is...interesting...to say the least, so the ball's in Firefly's court haha
this is already going to be the case with Firefly if i saw this correctly
i mean as soon as they integrate it in apps
You mean hands? MJ fixed that last Wednesday.
not fully but yeah
Can't wait to get access and try Firefly!
look at this monstrosity in mirror
Did they? That's cool. My trial ended and I've been waffling on paying the subscription and then THIS happened haha
I hope we can use legacy models to continue generating diffusion hands
Photographs are mostly good on hands now. Their illustration skills are still... rudimentary. Less artifacts, but no idea of anatomy at times. I guess the community rated all photos highly when they were training... I need a tool that knows illustration,
this is pretty cool! i hope adobe make a good user interface for users related to adobe suit in general!
Another AI to have fun with and the waiting for access
I look forward to seeing what level of integration they achieve with Photoshop. This AI thing is getting a bit too fast.
That's what I noticed as well...including some interesting cross-species examples lol. And I agree, the illustration capabilities would be most useful to me too...this is going to be great for brainstorming and speeding up the ideas phase of a project.
I wonder if Adobe releasing Firefly to the general public will cause a spike in GPU prices. Might be a good time to buy Nvidia Stonk. :D
(Disclaimer: this should not be considered financial advice; please note that all investments entail risk.)
hey tim firefly is working in cloud? no gpu need? or can we run firefly in local? what do you know about it?
Like.... I can work with messed up illustrations as reference. But sometimes... you wonder... whyyyyyy?
its almost certainly cloud
waves
currently the beta runs in your browser only. all the computing is done by Adobe and not on your machine
Will firefly run on cloud or locally in the future?
Applied for access!! Canโt wait! Is this trained on your own model?
Adobe Sensei processing Firefly images in a secret dojo. :D
how did you get a photo of my dojo....eeerrr Studio?
Big brother is always watching
Firefly was trained on Adobe Stock images, openly licensed content and public domain content, where copyright has expired. Learn more in https://firefly.adobe.com/faq
How do I opt out of being trained?
are you a stock contributor?
hi, can we try firefly on discord?
yes
Firefly is not a Discord bot
at the moment, you can only use firefly on the website. firefly.adobe.com
i just applied for the beta
you have to apply for it
and have to wit
then please check https://helpx.adobe.com/stock/contributor/help/firefly-faq-for-adobe-stock-contributors.html ๐
wait
gotcha! sounds good! i think this represents that designers who have been working with adobe pirate have their days over if they don't want to miss out on this new adobe in AI! what impresses me the most so far is the vector tool! i find it very useful! i have been using similar tools before vector AI, but this thinking in layers from the same system is quite revolutionary!
also, with this AI vector effects! it will make psds take a back seat! in the next few years!
correct ๐
6 hours later
where Mail for Access?
i'll use it for my graphic design purposes
nice! just keep in mind that currently you can't use firefly generated content commercially
If you successfully submit your access request, it may take time for the confirmation email to arrive. You'll receive an additional email when access has been granted.
Thx, why not a "open beta"? ๐
after 6 more hours
kek
i'll just get some sketches done
Kindly check the #faq section.
| Visit https://firefly.adobe.com/ to apply for Beta access. We're approving applications ASAP and you'll receive a confirmation email once you've been added. If you submit the form, you're in the queue to receive beta access, but it will take time before you're approved.
for logos and concepts
this move by adobe will boost the artificial intelligence competition even more! i was already using ia to design this will be great!
damn i haven't gotten a confirmation email today
A lot of requests are being processed, it will take time but you will receive the email ๐
I have sent a request, what is the response time?
Beta applications are being approved on a rolling basis and you'll receive a confirmation email once you've been granted access. At this time we do not have an accurate estimate for when access will be granted.
cant wait, looks promising ๐
One thing I don't understand at all, why not make the beta open? Right now it's super important to let users loose on something like Adobe Firefly. Users want to know what's possible and not wait days to apply to know if it's even possible. Adobe is such a big shed, why not open it up? In my opinion you are slowing down your own progress because you are slowing down everything at that point with user access. But just my 2 cents on the subject.
it just so happens that Adobe partnered with Nvidia 
this is not a joke
really? @dusty minnow
Getty images (where 5 of my photos are) partnered with nvidia for sure
/imagine a flower
@plucky dune
thx for the Link! ๐
@pine stirrup do you have any idea how it will work? I imagine it will be similar to adobe photoshop a subscription. i imagine that for all the computing power they will have limits for the users, maybe give us credits or limited generations! although it would be great to have direct access! adobe must have a lot of gpu power at their disposal without a doubt!
Firefly is not a discord bot (check #announcements for how to get access)
hehe thats is for other discord chat bro...
Tuesday, March 21, 2023 < okay ๐ Thought I missed something 
I never used MJ but wouldn't get out of here Stable... it's pretty good and open source
i used both
hahahahaha /imagine ๐
I'm sure letting everyone loose on a cool tool like this will go well. it's not like other Ai services are going down frequently because of user overload ๐
you are not here at midjourney ๐
Unfortunately, you cannot use prompts in discord; kindly check the #faq section for more info.
| Visit https://firefly.adobe.com/ to apply for Beta access. We're approving applications ASAP and you'll receive a confirmation email once you've been added. If you submit the form, you're in the queue to receive beta access, but it will take time before you're approved.
Eagerly waiting for beta approval
yeah me too! i also tested several alternative models!
DALL-E
A flying origami dragon
/imagine Karim without an ape profile .firefly (don't ban me just joking)
from the infrastructure perspective ... its difficult to scale horizontally without "work"
You might want to automate this process at some point, whenever somebody tries typing /imagine for example ๐
that's true ๐
people use stablediffusion prompts in midjourney, they think its literally all the same thing
I was referring to the open source models from stable diffusion! the community has a lot of models available at https://huggingface.co/.
the beta is free as of right now. once we have more information and once Firefly leaves the beta, I'm sure the current situation will evolve into something that is fair for everyone
I signed for beta, how long u wait for respond usually!?
It makes sense to me.
and right now firefly is super quick at generating content. much faster than Dall E IMO
yeah it makes sense
Hey hey! Super excited for Firefly from a web design standpoint
Aaa okay ^^ I avoid SD
it will take some time. we can't estimate how long exactly but the teams are doing their best to go through applications. so for now you can only be patient ๐
well! if you are here! on day one! i want to tell you that you are an early adopter! these things will not catch you by surprise! we are in the future today! ๐ฅ
Imagine being patient! ๐ But ye im sure they r overcrowded with aplications.
I'm wondering if Firefly could generate websites for Adobe XD?
I guess we might be among the first ones
It was announced about 8 hours ago
I applied about 2 hours ago i think
I think so! at least the draft! I have no proof but no doubts either.
oh wow really. i am an early adopter then ๐ FOR SCIENCE!!
(Leave my hope alone, i hope to be top on the list xD)
yeah day one!
I share link to my designer friends ๐
So Beta will be online on site, but when its out of Beta it will come to apps?
Is that correct?
Seems that way.
Adobe Firefly will be built into Creative Cloud to give people the tools to both generate results quickly and customize them to fit their unique vision.
https://www.adobe.com/sensei/generative-ai/firefly.html#faqs
@dusty minnow
democratization of design.... no more ugly designs! well, maybe we can't promise that! ๐
challenge accepted
Here I am! Hi everyoneโ๐ผ I Just made the request for beta version. Canโt wait โจโจโจโจโจโจ
Don't miss out on the opportunity to learn more about FireFly. Adobe just concluded a livestream introducing Adobe Firefly! You can catch the replay here: https://www.youtube.com/watch?v=c3z9jYtPx-4
Today weโre excited to introduce #AdobeFirefly, a new family of creative generative #AI models. Tune in to the live stream to get your questions answered, connect with other creatives and get started with Adobe Firefly!
Learn more: adobe.com/firefly
Eric Snowden is the VP of Design at Adobe and is responsible for the product design teams for t...
Yeah after your argument I got the first weeks in my mind from Midjourney .... ๐ minutes to create, or was offline (discord server) so yeah - you won for stability ๐
Has anyone got their hands on the beta yet? I'm curious if it is as slick as Adobe's other apps? I've always loved Adobe Flash (now Animate) simply for how fluid it is to work with vector art in that fashion, but haven't been able to justify the cost of CC (as I don't use it regularly). Them adding AI to the suit definitely makes it more appealing.
@pine stirrup finally appearing in final (red) form
oh hello there
i hope i get access. id be bummed if i learned all this info and don't get access ๐ฆ
as requested: "a badly designed logo"
What in the blue hell is that
Hi eveyone
I think @pine stirrup needs to get banned.
can you do multiple target prompts?
I'll see myself out
dont leave tim we love u
this was how you used to be able to cheat legible text into midjourney last summer and i've been chasing that high ever since....
Ye waiting is hardest part ๐ฅบ
Came to see if people were still being impatient.
Actually, is Firefly just going to be something added to existing Adobe Products? So, will I be able to say use Adobe Photoshop and just prompt in a background for my image?
Im wondering same thing
oh like this?
fire
Lemme get some donuts though
It will be integrated in many ways, which we'll reveal as development continues. The hope is to build Firefly into apps like Photoshop to allow designers to perform tasks like 'object replacement', or 'stylize a photo', etc etc. Lots of opportunities
That second unicorn is the most boring unicorn ever. ๐
ok try "A dog made out of putty in a realistic alley"
so midjourneys try with the same prompt:
I would say you have a better result ๐
Like extension?
That is sounding quite promising. Will it be some kind of ugly token system like Midjourney, or something we actually just run locally and get as part of the CC subscription? Do you guys know yet?
Here's one prototype from the Firefly website
like that but think seven or eight words
No details on that have been released yet, but for tools built directly into Photoshop, I would imagine (this is my personal thoughts), that it would be included with the PS subscription. Again, just speculation.
@slim lion Many designers and artist r worried about ai will take jobs from them. Average person will be able to do their job. What do u think about it?
hahahaha! well, this is one of the best I've been able to get when it comes to asking AI! in this case, I used a lot of words related to "vectors" and embeddings for stable diffusion.
if adobe allows that level of customization! we would have a great source of ideas!!!
looks like an average beeple dream ๐
yeah xD @dry glen
If Adobe Firefly is an extra subscription, i will take it anyway
I can only speak for what we're building. Our goal is not to replace designers (I would be very much against this). We're building tools to help them work faster and expand their creativity.
Many are focusing on just the image generation aspect of Firefly right now. The built-in tools are what will help artists keep on going.
But lets see where it will end up
in the spirit of speculation, honestly i would love to see it as its own app in the CC suite, As its own Ff perhaps in the package with Lr, Ai, Ps but also have integrations with those apps that make them more powerful too. idk.
I hope you guys can just bundle it into the subscription overall and keep the tools locally run. I'm not a huge fan of the limited generation options offered by token-based systems. I'd personally rather just pay a monthly fee for CC and get the tools I want to use.
I work for a university. Will my educational subscription include Firefly?
That is why is important to be part of Photoshop, Illustrator etc So many average people would not have time or desire to deal with it.
And API is coming too apparently (sooner or later)
Once Firefly is out of beta, I'm sure there will be more info on pricing/integrations/etc
sounds good bro! i mean i think designers who are already using artificial intelligence just see it as a time saver! it allows us to come up with interesting ideas! good creators will always take the time to make sure that what they create always has the best quality! that implies going beyond just writing a single prompt! work in depth!
As long as i dont have to sell my soul, my kidney and my mom im accepting 0_0
does this mean finally AI art is not 'Discord software' lol
it hasn't been discord software since the stable diffusion release
well i mean in how people run it, like every other software has its own UI, it doesnt use a glorified chat engine as an art engine ๐
I totally get the worry, though! We're seeing new tools pop up on a daily basis that appear to 'do the job' of designers/artists. Many of us internally are doing what we can to make sure we're building for the community, rather than against it.
Damn my whole workflow is surrounded by subscription based programs
I just want an AI art program that is ethically trained
about time MJ had some competition
Do you expect the beta to be rather long? Of course, not seeking any kind of actual ETA, as I know that never works out. Just kind of wondering if it might be this year, or if we're looking further out?
Will firefly have the ability to add to a picture by layers. or separate a generated image by layers. And above everything else. Generate transparent background images?
the way I see it, these Ai apps don't run themselves and still require human input to know what jobs to do, but only the human comprehends the nuances of whats right and what isn't and would still need to work the curation too. So when it comes to jobs, I'm not worried.
Sometimes, if you want the job done right, you gotta do it yourself.
MJ is limited in its features
Stable Diffusion is the more accurate one in terms of features
They r amateurs ๐
you better make money!
yes, and so do we! you can read more here: https://blog.adobe.com/en/publish/2023/03/21/responsible-innovation-age-of-generative-ai
That's a very good question. I have zero idea. ๐
I do, but i earn zero from art
We're hoping to support layers and transparent backgrounds.
hahahahahaha oh man
Yap xD
Judging from Adobe's other recent features, I'm guessing this is going to be called beta for a long time. Everything Adobe Sensei is still beta and those first features came out a few years ago
yeah! jus t imagine to drop your psd!
and plug and play!
Creative Cloud + Substance 3D textures + Autodesk Maya and 3ds Max + zBrush + plugins that may or may not cost 1.000โฌ+ and eventually more programs
ironically midjourney will probably never be able to do transparent backgrounds because the downside of free-for-all scraping is that half your "transparent" images have fake gray checkerboards 
If I used my school account/Adobe Cloud ID for applying for Firefly Beta could I still get in or would I have to wait until the Beta is released to public?
I want to know this too.
as long as you are 18 or older you should be fine. just apply ๐
omgggg that would be SUCH a game-changer! Generations in Firefly Ai that are already organized elements as layers that i could export as a PSD for easy editing. sheeeeesh where do i ship all my money, i want it! haha
Damn. I really want to get my hands on a good, locally running, AI tool specifically made for artists. Stable Diffusion and its various UI implementations are nice, but they don't really feel like art tools.
in 10 years everyone will be a digital artist WIZARD by today's standards and the elite will be defined as those who can synthesize visual art through telekenesis via their Teslappgle brain chip musk pro max
lol
@heady moon no you'll be able to access with your Adobe ID.
sincerely it would be good that it doesn't cost too much that extra... photoshop is at a good price! i consider that this will help to eliminate the pirate programs, so it has to be an ethical price!
With how fast AI is moving, I can't even imagine where we'll be in 10 years.
Is there an average time estimate for approval? hours, days, weeks?
Text prompting to video in After Effects. :)
Yeah lets see
entire tv show series based off a few prompts
Thought prompting
Did you see the filmmaking and animating AI from Wonder Dynamics?
Not yet!
have we just unlocked a new field of science fiction?
Like Delores was doing in Westworld. Creating characters and narratives at her desk. :)
dont have the prompt cable hooked up during a nightmare ๐ฉ
we don't have a time estimate as of right now. I can only ask you to be patient ๐
Check it out, but im a bit worried it might be very expensive for the average Joe
I still think even with Ai generation u need eye for it, to have talent creativity without it result would be poor.
doesn't look like anything to me
ok, thanks
Its developed by Hollywood people
Unless you use an ai to make your prompt lol
omg. what if you are all AI and im alone in here #solipsismButAI
Brb, using AI to write an AI chatbot that writes AI prompts.
๐ต
@floral violet ooooh look, a squirrel!
I think about this a lot.
hacker mind set!
I kinda got that going with GPT-4
"Random prompt generator" haha
We need a chatbot to interpret an ai image and give feedback on it
meow
So basically. This is how we begin to Fuse with A.I
there is some around... actually haha
I trained it to write prompts as if it were "an award-winning photo-journalist"
!gpt give me how to h*ck adobe photoshop
I'm not even suprised everyones on this train now
i'm joke guys
Another curiosity I have is model tuning. Will we be able to tune the Adobe model on our own works to allow it to generate in our own style? That on the docket at all?
There already are "clip interrogators" if that's what you mean...
Yes.
in product hunt you see a new startup base on ai every hour at day ๐
@clear cloud As Firefly evolves, Adobe is exploring ways for creators to be able to train the machine learning model with their own assets so they can generate content that matches their unique style
Every minute on YouTube. :D
i asked gpt to explain m-theory to me like it would to a teenager, kid, old man etc. and it made me realize that AI has crossed the uncanny valley conversation-wise. In other words, I can't tell redditors from AI responses and vice versa. we've arrived.
That's both exciting and terrifying ๐
something like that, the last one I saw was an Art. intelligence that measured how "good" your thumbnail was for youtube. based on the ones that have been successful. or something like that
sun
Thanks Howard and Tim, appreciate the quick responses. That would be awesome! I am really looking forward to being able to mess with this. Sounds like you'll be offering an awesome service here.
guys i want to get adobe photoshop (bla bla) but i don't have money any one know how to get this but free?
Many Redditors probably ask GPT for the answers, and paste the replies as their own
and 6 months ago! they said "bitcoin will reach 100k!"
I feel like this AI model will be better at adapting to the artist as its not trained on stuff like art station, right?
And this personalised model will be then available for every user automatically or there's a chance to keep it "private" and decide in a second moment about sharing it?

@dire dune this is where community feedback is important! Right now, everything is in beta, so make sure you share your opinion on what you prefer.
will firefly be included into the "full package" plan that adobe has now once released ?(I have a corporate subscription)
If i feed my current ones to the model it will just respond with "Get good dawg"
Git guud
Note to self: begin curating datasets to use for custom Firefly Training Models. :)
I don't know... a few days ago I was talking with an illustrator friend of mine with a drawing style I like, to see if we could create a model using his art as a base, in order to speed up the work (the model would stay with us).
give me adobe photoshop free ๐ซ๐ซ
@slim lion
There's still a lot to be determined with this. We also need to make sure it protects artists. I wouldn't want someone to ingest all of my work to train a custom model
because you asked so nicely @finite night https://www.adobe.com/products/photoshop/free-trial-download.html
i use the trial
Then you gotta subscribe ๐
I've submitted my request for the beta access, I'm pretty critical about how AI actually works and I did a couple of experiment with diffusion bee on Mac. I'm looking forward to see how Adobe handle this because I think it can be a turning point from an ethical point of view about the training issues
well, a friend of mine wanted it, so he wouldn't have to work extra! of course, he would keep that model private. ๐
Hmm will Firefly be capable of creating/generating stuff that is actually copyright protected like Marvel characters?
Gimme Ultron prompts
Firefly's model is not trained on trademarked content
does this mean firefly is confined to the art only created by adobe products? Unlike midjourney where it scrapes the web?
Oh well should have thought about that
My bad
I should have used my brain before i asked
It's currently trained on Adobe Stock and openly licensed / expired public domain content. It doesn't scrape the entire internet like some other tools ๐
Exactly. I was a little bit worried when I read that this model isn't trained with Behance, but I fear that it will be done in the future. Hopefully the team will find a respectful way to handle this
From the FAQ: Copies of customer content will not be included in the Firefly models. We do not train on any Creative Cloud subscribersโ personal content.
k
...meanwhile, over in alphabet city... LOL https://www.msn.com/en-us/news/technology/googles-new-ai-bot-bard-just-debuted-and-its-already-warning-about-its-creators-monopoly-over-ads/ar-AA18UmgZ
How much of Firefly's UI and functionality was inspired by tools like Stable Diffusion and MidJourney? Is there a chance that we could see more granular tools like those available via ContolNet in Stable Diffusion (depth maps, canny maps, pose maps etc.)?
But it will be trained with the material produced during the beta phase, right? The Behance thing was also in that FAQ page, more or less at that point. I had a quick chat on twitter with Rufus Deuchler and Scott Belsky about this and looks like they're trying to do everything very ethically with the support and feedback of the users
@steady marlin firefly's interface looks somewhat similar to Stock's, but it sounds from the faq and livestream that the plan is to launch the functionality into the app suite directly once it's out of beta. Seems like very early days; I doubt they even know, tbh ๐
we're working on a whole host of cool features that will be available later. as of right now the beta is limited to text generation and text to image. I'd say the UI is better than some other GenAI tools. Firefly allows you to add filters and styles or even layer multiple styles
yes, early days indeed. I mean it was released today. so I can't wait what will be possible in the future
@pine stirrup I assure you, the agony's worse for us customers. ๐
Is the text-to-vector trained on a dataset of vectors?
yes, that is a big concern for us. training the model ethically and involving the community. The text prompts and resulting images are collected for feedback (you'll see a small ๐ or ๐ icon in the UI.
i didn't get any email about confirmation of the access request
It isn't instant
it does take some time. you'll need patience as the team is going through all the applications ๐
ok, thank you
glad its not just me i signed up about 6 hours ago, even questioned if i hit request ๐
Thanks! Very excited by this and what the future holds. Hope I get my beta invite soon! ๐
it must be a nightmare to go through all those applications for beta
Not if they use ai to sort through them
I think that a situation where we decide to train a model and then share it (free or paid) as a resource on behance (like we do with Kyle's brushes) could be a good solution to allow creatives to decide what to do with their assets. Behance is a great platform and in the last years it evolved a lot in this direction of content sharing.
Amazingggggg addition to the AI arsenal โค๏ธโค๏ธโค๏ธ
gm! just signed up for the beta and can't wait to dive into to Firefly. @quaint lily is there a timeline for when invites rollout?
using AI to sort the applications, to use AI to generate an image, to use another AI to upscale the image... that's big brain wrinkles
It's in beta. Go sign up and get on the waitlist
Ok thx
i believe they make the same approach as OpenAI
i think its a priority based one
so they see who wants to use Firefly for which tasks
that would be a cool idea! who knows, maybe it will become reality - if you like you can post it here #1087561412269854840
Excited to see Adobe join the Text to Video/Image arena ๐ ๐
In that case, considering i am great friends with GPT, they should accept me right away!
it does take some time for the teams to go through applications. but your patience will be rewarded ๐
you know what im really excited about and which wasnt mentioned here before i think? Generating custom brushes that can be taken from lets say hair of a dog as demonstrated by Firefly
that is dope
Right now, you can apply for Beta access (see #faq)! Once you get into the tool, you can experiment and share feedback. We are experiencing very high application volume, so it may take some time before you get approved (no indication on the timeline yet).
Will make a list and post it there as soon as I try the beta in the next weeks!
Does the Beta require subscription ?
I'd like to see some Adobe AI Jarvis in my photoshop who can perform complex actions from my text input!
Is fine-tuning on your own images/photos supported by Firefly?
in the current beta this isn't supported. but it might be in the future
Welcome, Kalsey! Waiting for your FireFly video once you dive into the experience. Your unique perspective will surely be a valuable resource! ๐
Ok thx. Will any APIs be available for generation/automated prompting?
//imagine some bubble wrap to pop while you wait for the invite ๐
||pop||||pop||||pop||||pop||||pop||||pop||||pop||||pop||
||pop||||pop||||pop||||pop||||pop||||win!||||pop||||pop||
||pop||||pop||||pop||||pop||||pop||||pop||||pop||||pop||
need a Dark Mode version ๐
Im gonna need a lot more bubble wrap
it's okay if it takes a while to get in the beta, i can just kick back and relax with the new bing image generator--
Guys I want to know which app from Photoshop and illustrator is suitable to make this design almost in one day max
Firefly?
actually that's probably a good question to ask: will there be a Content Policy on firefly? it's one thing for a dedicated text-to-image service to have PG rules, but an art tool to be used in conjunction with other adobe stuff... well, that jacket could get tight
Firefly will be embedded inside Adobe products. Once we know more about the feedback from beta users, features like automation might get prioritised on the dev list.
You would probably need to use 3D software
Yep
Like wt
hmm. looks like it was made in 3d. so with no experience whatsoever, I suppose you won't have time to learn Apps like Illustrator or Substance 3D. firefly could be your solution here: firefly.adobe.com
For Adobe theres Dimension for 3D i believe
I am waiting the access
Done
yes, it does take some time - that's normal as the team is going through the applications
hi
I hope I can finish it tomorrow as soon as possible ๐ฅฒ
nรฃo entendi โฌโโฌใ( ยบ _ ยบใ)
There's a whole host of issues including abuse of imagefonts, prompt engineering to extract images that are 99.99% similar to training images, juxtapositioning to create offensive images, etc. I ran a research project last year on the number of ways in which LLMs can fall foul of laws and policies via careful prompting. E.g. ChatGPT Jailbreak.
juxtapositioning to create offensive images"issues"
what i'm hearing is that, while currently, adobe sells tools that will let me create the most offensive imagery imaginable, on the theory that i'm the one directing it and i'm the adult responsible for my actions, in the bold AI-powered future, our tools will only create things the tool-creators themselves will approve of
I wonder how long before someone pulls an Alpaca on some of the great tools coming out?
there's been very little demand for that with stablediffusion already out; alpaca was in-demand because there was no comparably-capable text engine (all the FOSS stuff like GPT-J kinda sucks, no offense to novelAI fans)
Well, given Firefly is trained on works Adobe has the rights to use, I don't imagine the majority of these issues even apply.
Im most excited for whats to come for 3D ai
"the image-generating AI can create offensive images" is the opposite of an issue
car made with cookies
The issue is where the liability lies, legally speaking. There are many different content laws around the world and liability can reside with the prompt engineer, the LLM provider, the hoster or the publisher. A very thin tightrope to walk. There is forbidden imagery in different parts of the world. Wonder how Adobe etc. will deal with the fact that the AI model generates the art, not the prompt engineer who is exercising their free speech. Also what about unintended imagery from innocent text? I suspect we'll see AIs vetting AI output as the only practical solution.
I still haven't received the "waitlisted" confirmation email ๐ฅบ
@severe dew @shell delta this channel is for general chat, please post over here, not in the share your work channel
You can't generate images from Discord
There are many different content laws around the worldyeah, but there's only one adobe corporation, so if china tells them "don't let people prompt general secretary xi as winnie the pooh bear or we're kicking you out," we all get to live with it
Como gera imagem aqui no firefly discord?
If only the world worked like that.
No puedes generar imรกgenes desde Discord
You cannot create prompts here in discord. You have to request access to the beta first at https://firefly.adobe.com/ Then you can start creating your prompts directly in Firefly. You can learn more in #faq or #announcements
They do unfortunately. It's amazing what you can get an LLM to do that it isn't designed to do.
Has anyone gone through the application yet?
they said it's trained on public domain content where the copyright has expired
China would have this site blocked for more reasons than just General Secretary Pooh.
Chinese users that use the site will need to use VPN already.
i don't know if you've ever been to an art museum, but there's a few nudes here and there
hello people, i'm here because the request button on the adobe website doesn't work
so, this is an issue that will come up Sooner rather than Later
I know emails are getting out but I don't think we expected everyone to be THIS hyped about it ๐
And? If they did not train on works they don't explicitly have the right to use, then what's the issue? It being able to create something unexpected is no different from an individual being able to do so via Photoshop itself.
it should work, it just takes time for the teams to respond and approve your request ๐
I think you have to broaden your imagination. There are banned images in different religions, previously war-torn regions, restrictive states, etc.
oh, i clicked on it and there is no effect whatsoever, not a message, no nothing. Is it working?
Has anybody received the request confirmation email yet? Not access, just the confirmation. ๐
i'm consciously asking the easy and obvious question because i'm curious what the answer is
๐
oh, in that case you are right. you should see a form. perhaps an adblocker is stopping something on your end? maybe try a different browser
ok thanks, let me check
No, I submitted my application nine hours ago, but I didn't receive a confirmation email either
you can learn more about the ethics of Firefly and our standards here https://main--clio--adobecom.hlx.live/faq-external#about-firefly
that's kinda weird, because, on chrome i disabled my adblocks and nothing happens, and I've just tried with Edge with no more success
I can see per-region content blocking by filtering AIs to avoid legal issues. It's an unusual scenario because the tool is generating the creation, not the user, so some jurisdictions will hold the LLM provider culpable for breaking local laws. No company can afford a ban in major markets like Europe and China.
try this form https://survey.adobe.com/jfe/form/SV_2tNbydqEp6ovet0
Adobe Firefly Invite Request Form
that's why Firefly is still in beta and feedback is important
I'm really excited to see what Firefly can do and how it will fit into the workflow.
Just filled it, thanks a lot Tim
What a time to be alive!
I have a question, after we get access, can we only use "Text to image" and "Text effect"? Can we experience other features that we are exploring?
I'm trying to get the BEta
this ethics statement is very stirring and tells me nothing about what the actual policies will be around, e.g., nudity
what the hell
text to image and text effect are the currently available features in this beta. others are not yet available but might be in the future
Just those two for now
I was just on Twitter , and saw this I gotta try this
where are you stuck, maybe we can help?
so, is LAION database not used in training Firefly like it was for every other SD based model?
Got it,thanku
correct. we did not use LAION to train Firefly. The current Firefly generative AI model is trained on a dataset of Adobe Stock, along with openly licensed work and public domain content where copyright has expired.
no one is touching LAION these days even midjourney is running away from it, too much bad data