AI picture generators urgently required

[dcr]

11 minutes ago, Fabricio Z said:

I'm saying that there's a reason people moved from Photoshop to Affinity

No subscriptions?  No Internet activation?

Whoops.  Serif went the wrong way on that second one.

 

12 minutes ago, Fabricio Z said:

Serif realizes they're losing ground due to not having AI implemented, you bet they'll do it.

Can Publisher v2 export to .IDML?  Can Publisher v2 export to Publisher v1?  Can Publisher v2 export tagged/accessible PDFs?  Can Designer v2 open .AI files and not just the PDF stream?  Can Designer v2 export to .AI files?

There are a ton of areas where Serif is either losing ground or never established a foothold in the first place.  It'd be nice if they'd get some of that stuff done before worrying about AI.

[Fabricio Z]

4 minutes ago, dcr said:

No subscriptions?  No Internet activation?

Whoops.  Serif went the wrong way on that second one.

 

Can Publisher v2 export to .IDML?  Can Publisher v2 export to Publisher v1?  Can Publisher v2 export tagged/accessible PDFs?  Can Designer v2 open .AI files and not just the PDF stream?  Can Designer v2 export to .AI files?

There are a ton of areas where Serif is either losing ground or never established a foothold in the first place.  It'd be nice if they'd get some of that stuff done before worrying about AI.

Does it have to be binary, like one or other? Does Serif have like only one or two employees? Have you ever heard about development backlog and roadmaps?

[drstreit]

I see the challenge in th3 fact that an AI model is not as straightforward like implementing a technical interface: There a HUGE quality differences that will take time and resources.

Adobe recognized that and went all in. Serif does not even show a roadmap…

Not sure if they can catch up tha t gap.

[Fabricio Z]

I agree that Serif have its own goals and targets, no question about it. However, about the AI trend, it's up to them to decide if they need to pursue it or not.

To those who alleged it could be difficult for Serif to do it: You know the filters in Affinity Photo? They were mostly developed and coded by Serif itself. But most of those AI software or implementations you see in other apps over there are not coded at all. They are open source models, hence that's why we see millions of new apps everyday using this technologies.

If we can see a 13 year old make an AI art app based on those models, I'm pretty sure Serif is competent enough to do the same - if they decide to. So please, stop treating it like it is rocket science, because it's not.

[Twolane]

All righty, then. If Serif should include an AI generator, should the applications be turned into a subscription model to cover the associated costs? I think that would be only fair. After all, we can't keep getting something for "nothing" forever, now, can we? I think that would be an excellent way for the company to collect more revenue from the unhappy subscribers who can't seem to do anything without AI included in the products. Personally, I want AI to work in the Publisher version, too. I'm tired of sitting in a dark room while making things up all by myself.

Who does Red Sands work for? The competition? He shure does talk a lot. Isn't he one of the few who forecast that Affinity was surely going down the road to perdition and darnation if the outfit didn't come up with an update to v2 where everything worked to his satisfaction? Well, I guess Affinity showed him. Unfortunately for the rest of us, he's found a new bone to chew on and he's going to show us the way all over again.

[Fabricio Z]

4 hours ago, Twolane said:

All righty, then. If Serif should include an AI generator, should the applications be turned into a subscription model to cover the associated costs? I think that would be only fair. After all, we can't keep getting something for "nothing" forever, now, can we? I think that would be an excellent way for the company to collect more revenue from the unhappy subscribers who can't seem to do anything without AI included in the products. Personally, I want AI to work in the Publisher version, too. I'm tired of sitting in a dark room while making things up all by myself.

It's a possibility, yes. Like a subscription add-on, like DaVinci Resolve Cloud solution, where the app itself doesn't use a subscription model.

[dcr]

22 hours ago, Fabricio Z said:

Does it have to be binary, like one or other? Does Serif have like only one or two employees? Have you ever heard about development backlog and roadmaps?

On any roadmap or backlog, I would place tagged/accessible PDF capabilities way ahead of AI.  Tagged/accessible PDFs are a legal/regulatory necessity in many places.  AI is not.  And, if you're assigning people or teams to different items on the list, I would again place AI at the bottom of that list.  I would even place a Linux version, which we know isn't going to happen, ahead of AI.

So, yeah, sure, different people or teams can be working on different things.  Generative AI does not need to be one of those things.

[Fabricio Z]

3 hours ago, dcr said:

On any roadmap or backlog, I would place tagged/accessible PDF capabilities way ahead of AI.  Tagged/accessible PDFs are a legal/regulatory necessity in many places.  AI is not.  And, if you're assigning people or teams to different items on the list, I would again place AI at the bottom of that list.  I would even place a Linux version, which we know isn't going to happen, ahead of AI.

So, yeah, sure, different people or teams can be working on different things.  Generative AI does not need to be one of those things.

We all get that you don't need or consider AI an important feature and I agree there are other priorities. Still, it's up the company to make its market research to know the demands it could provide for. They know better than us.

About a Linux version, I'd like it. However, it doesn't considerate the current users, while other new features do, so it's less appealing unless they're missing a good niche, which I believe they are not, since Linux users have other open source alternatives and most Linux users aren't going for paid proprietary software.

But just for curiosity sake, based on what you consider a Linux version more important and viable than AI? Are you aware of the costs not only to port it to different Linux kernels but also to keep it always up to date and compatible with its drivers? They would have to hire a keep a new team dedicated to it. Believe me, it's way different than implementing a resource that most graphical softwares are already doing.

I'm not saying that AI is that top priority, but certainly isn't lower or less viable than a Linux version.

[stranded]

even Canva has the AI generator now, they could even create an extra program like Affinity AI so we can happily pay for it, just make it happen

 

as for Linux support it would just make them work more for nothing, no professional environment uses Linux except for a tiny percentage of freelancers maybe 

[dcr]

17 hours ago, Fabricio Z said:

But just for curiosity sake, based on what you consider a Linux version more important and viable than AI?

Okay, but we're going to go the long way around . . .

In my view, computers should save us time.  So, let's say I create a poster.  I do it in InDesign, Publisher, Illustrator, Designer, whatever.  I keep the source file.  Now, let's say in ten or twenty-five years, I want to change the poster a bit.  Maybe I want to add an anniversary logo and make some other minor changes to modernize it.  I should be able to open that source file and make whatever changes I need.  I should not have to need to recreate the poster in something else all over again.  Maybe a PDF could be edited.  Kind of depends on what tools were used in the creation of the poster and whether those tools remain accessible and/or usable in the PDF.

Ideally, version 10 of the creation program should be able to open a version 1 file and retain the original editing capabilities.  But, maybe it doesn't.  But, if I have the file and the original program that created it, I should be able to open it in that.  Maybe I need to get a hold of an old computer.  Maybe I need to use an emulator.  But, somehow, I should be able to open it in the original program.

That means I need to be able to launch the original program.  And, these days, that usually also means needing to activate it.

If the company that made the program is out of business or has shut down their activation servers or has otherwise made it impossible to activate the program, that pretty much defeats the purpose of keeping the original source file.  That means either recreating the original work in whole or in part.

So, for those reasons, I avoid application programs that require any kind of Internet activation scheme.  Which is why I don't use the Affinity v2 applications.

All that above also requires being able to install an operating system that supports the program in question.  Both Apple and Microsoft are trending toward locking down their operating systems.  So, in ten or twenty-five years, would I be able to install an older operating system on either old hardware or in an emulator?  If operating systems require user accounts and Internet activations or whatever, the answer is going to be probably not.  So, I can foresee a strong possibility of needing to switch to a Linux distro that can be installed and "activated" from an offline installer to cover that eventuality.

Of course, right now, with Affinity requiring Internet activation, a Linux version that follows the same scheme would not be viable for me.  But, were it to support Linux and use a good old-fashioned license activation (like v1), then it would be.  But, again, neither is likely to happen.

As for AI, it depends on what kind of "AI" you're talking about.  If you're talking about tools that make it easier to remove backgrounds or seamlessly remove unwanted portions of an image, that's fine.  Or, if you're talking about filters that make a photograph look like a cartoon or that remove motion blur or add motion blur or stuff like that, fine.  Where you're talking about tools that allow you to manipulate an existing image, that's mostly fine.

But, for generative AI, no.  For one, the copyright issues (at least in the U.S.) are still unsettled.  If I draw something or photograph something, the resulting image is clearly mine.  I have exclusive rights to it.  No one else can use or sell that same image without my permission.  If I pay someone to draw something or photograph something, I can have exclusive rights to it with the proper agreement or contract.

With AI generated "art", who owns it?  If I ask for an astronaut horse standing on the moon, do I have full rights to it?  Or can I simply use it on a non-exclusive basis?  If you later ask for an astronaut horse standing on the moon and the resulting image is substantially similar to the one AI generated for me, can I sue you for infringement?  Or do we both have non-exclusive rights to the images generated for us?  Or, what if it is deemed that no one owns the copyright on the image and it's effectively public domain that anyone can use?  Why would I want to pay to use an AI tool to create an image I don't have exclusive rights to?  I could just as easily use an image search engine to search for an AI-generated image of an astronaut horse standing on the moon and use it.  Maybe it's yours.  So, you paid to use an AI tool to generate that image which I can now download, print on a t-shirt and sell.  So, hey, thanks!

But, what if someone else had painted an astronaut horse and that image was used in the "training" of the AI?  Then, you and I request our astronaut horse on the moon image and then the AI spits out an image substantially similar to that original creation?  And then we sell "our" images on t-shirts.  Can the original painter sue us for copyright infringement?  Will the Affinity AI TOU indemnify us against such claims?  Can the painter sue Serif?  All of us?

There are too many unknowns.  Too much is unsettled.  That makes generative AI something I would not touch with a ten foot pole.

[ffrm]

The best path forward for serif is to open up an api.  It side steps legal issues like rights (since they won't provide their own generative ai) and it allows the community to iterate quickly and it helps explore various workflows without punishing their busy devs.  For example I can already think of various types of generative plugins I would write if I had the chance.  

[Visualsbydesign]

I just wanted to give my personal opinion and how this type of thing would massively help me and be put into good use for my work. I produce 3D visuals and have seen the talk and examples of the Photoshop ai now in use in the discord channels I’m part of and there is a definite buzz for those using Photoshop.  For practical uses, things like when you have finished the render and the client asks if you can add dog on the floor, if you can add a red bunch of flowers on the worktop, if you can add some more books on that shelf, you can see the type of thing I mean. I know you can try to find an image, cut if out, place it and colour balance it to fit but how much time that takes as opposed to it all being done in an instance is a massive thing, this isn’t replacing me and my work, it is enhancing it and giving me more chance of getting more work and being competitive in my field. I would love this sort of feature as an extra add on.  I will be thinking seriously if I need to jump to Photoshop purely for this feature alone.  AP, please do consider to add it to your great software🙂

[nickbatz]

https://www.macrumors.com

"In addition to gestures, the headset will support hand movements such as air typing, though it doesn't seem like those who have received a demo have been able to try this feature as of yet. Gestures will work together, of course, and to do something like create a drawing, you'll look at a spot on the canvas, select a brush with your hand, and use a gesture in the air to draw. If you look elsewhere, you'll be able to move the cursor immediately to where you're looking."

I'm absolutely livid at Affinity for not having developed a mind-to-picture interface already. Come on, guys, keep up!

(By the way, while this looks like a cool product, the hyperbolic reactions... oy.)

[v_kyr]

Personally I'm still a big fan of J.A.R.V.I.S. - (Just A Rather Very Intelligent System), which in later episodes has been replaced by F.R.I.D.A.Y.

[VectorVonDoom]

On 5/30/2023 at 7:49 PM, Red Sands said:

It's funny how a myth was established that Affinity was an Adobe killer and that Adobe was threatened in any way. Adobe is and was not. Of course not. There are a lot of people in here who have switched to Affinity, yes, but I haven't heard Affinity mentioned in professional circles on a national or global level. The name pops up on forums and around the web, but I've never heard the name mentioned in my day-to-day work or seen any trace of Affinity in market analyses. We're not talking about a Figma or Sketch company here.

It is a myth that Affinity has any noticeable foothold in the large market, but they have the market segment that Serif has always lived off, the small fish with correspondingly small wallets.

The programs ooze of being aimed at this segment, and of course Serif would like more customers, but after a fresh start with Affinity, where everyone was fulll of a fools hope, it's mainly new wine in old bottles. It's also for the best. Add business expectations for functionality, output and support on top of that, and Serif would be crushed by challenges.

The question is how the current business strategy will fare when AI steps in as a main program or sub-features in competing products. Whether there is a place for a low-cost model that does things the old-fashioned way (slow slow slow) in a revolutionzed software market.

No matter how much dreaming goes on in here, it's sales figures and market share that will decide everything. And the sprint is underway, started by many new companies where the core of the product is innovation, not a rewrite of the core.

Affinity didn’t start or spread the myth that it was an Adobe killer, a few users did and still do. if you are talking about functionality, they are clearly wrong. I’ve no doubt that there are pros who have switched as for some Affinity does enough and sharing master files isn’t a requirement for them.