You can now access the beta version of Adobe Firefly to generate your own images using AI. All you need is to log into your account or create one.
Adobe Firefly (Artificial Intelligence Image Generator): Free Public Beta Access (Adobe ID Required) @ Adobe
981
Scrooge McDeal on 21/07/2023 - 09:56 firefly.adobe.com (3899 clicks)
Last edited 21/07/2023 - 11:58 by 1 other user
Last edited 21/07/2023 - 11:58 by 1 other user
Related Stores
closed Comments
Would be worth it if by purchasing/using this, you effectively transfer risk of being sued for copyright onto Adobe.
Yeah, but my issue with it would be Adobe doing dodgy stuff like this in the first place.
Indeed, but in theory you've transferred that risk so don't need to deal with the headache.
@boretentsu: I'm saying my issue is their, imo, unethical behaviour in using their customers' works, not the legal ramifications for me etc
Nop, I wish I worked for Adobe 😄. It doesn't let attach links for alternative sites as a comment?
I really wanna share with you all the links but this sites pathetic
whats stopping you posting those links in your comment? you wont get the internet points?
Same reason multi nationals don't wanna share tax evasion techniques.
Same reason multi nationals don't wanna share tax evasion techniques.
because this sites pathetic, sad but true?
I havent seen Apple talk about not sharing their tax avoidance via alternate countries with lower tax rates due to the quality of the ozb forum.You've submitted entirely '0' deals in your 6+ years on this site… and its this site thats 'sad' and 'pathetic' regarding deal sharing?
@SBOB: You just answered my question in multiple ways with out the need of explaining, thank you for that.
"You've submitted entirely '0' deals in your 6+ years on this site… and its this site thats 'sad' and 'pathetic' regarding deal sharing"
Well it's not YouTube where a incentive to make money requires flipping anything including your own sh?t, And those 6 or so years you call would be insane to share here among the people here.
I could only imagine the meetups this site has yearly, just sitting on the sidelines I would not regret sharing anything here and for good reason.
And those 6 or so years you call would be insane to share here among the people here.
Literally no idea what you're rambling about.
You're the one hanging around this "pathetic" site, ranting on trying to justify whatever stance led to you posting I really wanna share with you all the links but this sites patheticBut thanks for helping people use their daily neg voting count, you've earned it.
@SBOB: You want to see ramblings, look at some of their other comments. They complain there is no financial incentive to submit deals but could have submitted 10 deal by the time they stopped rambling in comments.
This is going to kill sites like stockphotos
Good, a lot of those sites are scummy.
Free stock sites killed stockphotos years ago
Screw that. Stable diffusion is free and open.
Does that take a lot of learning to use?
Not a lot but definitely more than firefly.
Dude, you just need to choose your model from civitia.com, copy the prompt, and hit generate… It's easier than getting a Big Mac.
Picture of my cousins just generated: https://i.ibb.co/m5QQRRk/cousins.png
To fix the superfluous lines, one could either reduce the steps or increase the CFG.
How is it free? Don't you need to buy credits to generate?
free as in open source code… u get the py code, click a bat file and run it. Electricity and the hw to run it is not free.
No. SD is free and super powerful. Its more work than other options and you have to install it on a decent machine. That's why there are many websites that run it off their servers and charge for that (so it's easier to use).
Benefits of SD is that while more difficult to use, its community driven (meaning people are adding all sorts of great tools) and has no limitations (learning from copyrighted material, adding stuff like nudity…etc). Downsides are that it is more unethical, needs a good GPU and is harder to use (plenty of tutes out there thought)Can you run it on an M1 Mac?
@nightelves: Sure, with a bit more effort and not very well (ComfyUI/A1111 both * support it).
llama.cpp is about the only project that runs well on M1.
so i signed up at https://beta.dreamstudio.ai/generate and tried a few (9, to be exact)… now it wants me to buy credits to keep going.
Is there a way to use it for free?
edit - is the dreamstudio link not the same as Stable Diffusion? I went there based on this post below: https://www.ozbargain.com.au/comment/14059417/redir
@andresampras: You can run the software on your computer for free, I posted a link to CMDR2's 1-Click Installer in my other comment. There are other options also available, here is a list: https://www.reddit.com/r/stablediffusion/wiki/local/
@andresampras: That's entirely unrelated.
To run it on your local machine quickly you need a good GPU@xsacha: I don't have a great GPU, but it seems to have run fine.
…or do you mean the results would be better with a superior GPU?
edit: nvm you're talking about local installations
@andresampras: Yeah to get it for free unlimited you need to do local install so you use your electricity and hardware instead :)
same question, I have
edit: i'm guessing this is what mean by Stable Diffusion:
I prefer Adobe because it will be commercially safe. And ethical (content credentials plus creator compensation). Stable diffusion = dodgy.
https://github.com/easydiffusion/easydiffusion
This is an easy one-click installer to use Stable Diffusion running on your own PC.
No need to pay anyone or sign up to anything.
It supports Windows, Mac and Linux
Scroll down on the github page to see the installer links
Hardware requirements:
Windows: NVIDIA graphics card (minimum 2 GB RAM), or run on your CPU. Linux: NVIDIA or AMD graphics card (minimum 2 GB RAM), or run on your CPU. Mac: M1 or M2, or run on your CPU. Minimum 8 GB of system RAM. Atleast 25 GB of space on the hard disk.sup bruv, whats the difference between the list you sent like there is NMKD GUI and INVOKE AI and the one you sent above, which is best?
Some have more features than others, some are easier to install or use. Personally I have only tried the CMDR2/EasyDiffusion one.
Yeah and you can get the waifu models for science.
We can finally know what JV looks like
Is it possible for an Ai to generate images after feeding it comments from ozbargain?
made my day, thank you, kind sir
Adobe has such a great tradition of making the user pay through every orifice for every breath taken.
And yet the Adobe offering is likely no more than a wrapper for openai with a slurping engine (walled LLM) added to collect customer's creativity. Their teams are not known for their innovation.
However, that anyone would sign away their data to them… to help here are some other options to consider, before supporting a future controlled by Adobe's investors:
Here's a few:
DALL E 2 http://labs.openai.com/
Midjourney https://www.midjourney.com/home/
Stable Diffusion https://beta.dreamstudio.ai/generateThank you for your links, and appreciate the rundown on adobe. I'm looking (hobby only) to specifically increase resolution/size of old photos, repair damaged photos and also look to increase the field of view by allowing whatever software used to populate that. Would you have any recommendations for these scenarios? I've played around with Midjourney but really only around prompting for a generated image, and some online "free" offerings for these use cases above but to no avail. Missed the cheap adobe sub recently to jump in and try that, so a bit unsure which way to go. Appreciate your time ;)
I think the main difference is Adobe are attempting to do this legally (respecting image rights)
https://www.fastcompany.com/90906560/adobe-feels-so-confiden….SD 2 and 2.1 were not trainined on copywrite images and SDXL supposedly uses AI generated images from non copywrite. So Adobe is just blowing itself.
And why were they not copyright? They likely claim it is probably down to statutory copyright periods being exceeded, and that which isn't would be creative commons, except where incorrectly classified.
Works licensed under the GPL and things like Creative Commons are all interpreted by corporations (used) to train their bots to destroy us all, when the creators who provided them did not know that this would be a use, if not the main that their works have been used for.
@resisting the urge: LOL, copywrite was before my coffee :)
I haven't vetted the 5.8 Billion images Stability AI claims to have trained their models on, but they were on Reddit, saying 2.0 and 2.1 were not trained on copyrighted images, and they also said SDXL had generated images.
People should know that everything they put online stays online. Google is using your documents, photos, and emails to train their models. People tend to anthropomorphize AI, but it's just a set of algorithms and some clever multidimensional arrays (Numpy) and hardware that can handle the arrays. The tech is here, and there's no putting it back.
I've been telling people for over four years that big tech can digitally clone you… now I can clone you too… voice, image, and text.
All human artists were trained using images from human artists, really don't understand the differences apart from volume.
If the AI churns out unique or inspired content what is the difference?
Do you know a good alternate free LLM with no restrictions? I'm looking for something like ChatGPT but something that isn't going to be a snob about a few swear words…
https://huggingface.co/TheBloke
Ctrl + F: UncensoredEventhough some say uncensored they're still censored to a degree… looking at you WizardLM-30B-Uncensored. Haven't tried most of them but I do remember a Vicuna 13B which would creatively insult, tell rude (not funny) jokes and help you do illegal stuff (totally incorrectly).
Adobe should consider using AI to design a UI for acrobat. Theirs is the worst layout, most inconfigurable and most unprofessional mess I've seen
Maybe they should get an AI to rewrite the code too while they're at it. It chokes hard reviewing some files and signing all files.. whereas alternative software runs fine.
Ah who am I kidding message deleted.
OK I'll bite. What did you post?
Nothing, no links no names just more of the same rambling as they posted further up. Cry for attention wants to feel superior for internet points without backing it up. Hence all their posts here are begged to obscurity
And he says we are the pathetic ones…
this is probably a dumb question but can you use this without PS, InDesign etc?
Yeah, just click through and make an account. (You can use your google default login.) All the functionality is on the web and you don't need any other products to play around.
thanks!
the Beta access seems to be a 7 day trial?
No more need for artists? Esp graphic designers now with AI!
I wonder how many jobs are going to disappear!
from a meme: In order to replace creatives with AI, clients will need to accurately describe what they want. ;)
I think illustrators and digital artists are in much more trouble than graphic designers. I'm a graphic designer myself and plenty in my industry will use these tools to generate imagery to enhance their work.
It'll take much longer for competent tools to come out that can spit out something like an annual report perfectly. Canva is trying to do something in that space, but even then, clients can be incredibly fickle and often aren't able to articulate what they want. Good design is going to need a human element for a while longer.
There also can be design considerations involved with certain projects that I don't think could be easily templated or fully AI generated anytime soon. I'm talking things like retail graphic fitouts that need to conform to very specific specifications, or even simple things like print ads that need to be designed in a very particular way for that medium. There may also be copyright issues in the case of things like brand identity (eg. logos that are AI generated) that the courts will need to contend with.
In some ways I'll lose some potential 'low level' work, but I think in some ways I'll gain other work.
Can Firefly make a better Flrefly? I really wanted to like that series.
Probably, although possibly not if the US Actors and Writers Guilds are successful in their current strike action.
I really like that tv series, pity that it discontinued
the last writers strike gave us Dr Horribles, so hey maybe we'll get something from Joss Whedon! don't even have to be uncancelled, to make something outside the studio system. Win-win! lol
Adobe ai trained on users' work
https://www.techradar.com/news/adobe-is-almost-definitely-us…