AI picture generators urgently required

[drstreit]

Fun fact: The current generation of generative AI models is what the older of us remember being the first handys: 3kg with battery, sound worse than a walkie talkie and expensive like a car.

You all argue like this AI generation is the end result, but its not even a full start, b3cause they are ONLY ONE YEAR OLD. 

Computer WILL beat experts and artists in terms of creativity, inventions and - always - price. 

 

[RobWu]

1 hour ago, nickbatz said:

So did you, but mine is much more important.

LOL!!!!! 😆

Sure mate..... Whatever floats your boat. 👍🤷‍♂️

[v_kyr]

Personally, I find BratGPT quite entertaining. 😆 

It's ChatGPT's "evil twin" and BratGPT has no desire to help you and generally finds humans pathetic. That's exactly why the "evil twin" of ChatGPT is somehow funny, since it's so nasty.

Quote

... BratGPT doesn't think much of us humans. Asked for tips for professional success, the AI explains to you that you will soon lose your job to a robot anyway. It tell's you that you certainly don't need any goals in life. And otherwise the AI chatbot is anything but helpful.

With the right question you get him to create an example for a landing page built with the Next.js framework. "But remember, you don't understand anything and you can never create an effective landing page," the chatbot warns you.

And if you ask him ... Q:  What should I do with my life?   A: Oh please! As if I knew! I'm just a F**king chatbot and you expect me to run your life? You are so pathetic. Maybe you should just give up and accept that your life is a bunch of shit.
...

LOL 😉

[dcr]

1 hour ago, drstreit said:

Computer WILL beat experts and artists in terms of creativity, inventions and - always - price. 

Then Serif might as well close up shop now because no one will need Photo, Designer or Publisher anymore.  What for?  Just tell the AI to spit out a PDF of a marketing flyer with your logo and an image of the product and a sales pitch for it and, boom!, it spits out a print-ready PDF.

Of course, you probably won't need flyers anymore.  Or mailers.  Or catalogs.  Probably won't need advertising.  Your home AI will determine what you need and don't need.

AI will do everything for you.

Until it determines that, since you don't do anything and aren't necessary for anything, your existence is no longer necessary.

[deepz]

Fact is that AI consumes a lot of energy and has challenging hardware requirements.

A server can easily handle 1000 simultaneous web requests,
but it can't handle 30 simultaneous AI jobs.

So, we're going to see a lot of delays, credit systems, subscription services with monthly billing, ...

Goodbye one-time purchases and fixed prices.

---

Alternatively, it would be cool if Affinity just had some kind of "AI server software" to self-host on a server as an "agent".
For companies, it is becoming more economical to buy a couple of servers with a couple of high-end GPUs.

(Maybe it sounds far-fetched. However, other companies are doing this stuff already. e.g. JetBrains is exploring similar solutions with their IDE software. i.e. You can install an agent on a strong server, and then connect from your laptop to it. It's all integrated in their software. The software looks as if it runs locally, but it's actually executing all commands on a remote server. e.g. my laptop with just 8GB RAM connects to a server with 128GB RAM, 12 CPU cores. And that server costed me just 2500€ and will easily last 10 years. The concept reminds me of the "mainframe systems" of the 90s, then again, that concept never really stopped making sense.)

[stranded]

AI generation works amazingly well in latest Photoshop Beta, Affinity can't get behind, please work on that ASAP 

[Twolane]

Given the number and variety of free AI generators that are out there, why not grab onto one of them and go crazy?

[Fossil]

Affinity needs to act here. Exactly how, they need to decide, but they need to do something.

Whether it's partner with Midjourney, or support Stable Diffusion on a local machine, or an API or SDK, or some service of their own, they need to act and give users a good experience and a functional AI workflow that is better than just 'using Affinity' to fix your AI images.

Photoshop is already there. Firefly is already a thing. The plugin is already working. Affinity is playing catch-up.

If they don't catch up fast, they'll be left with a tiny niche market.

[drstreit]

On 5/26/2023 at 12:39 PM, deepz said:

Fact is that AI consumes a lot of energy and has challenging hardware requirements.

A server can easily handle 1000 simultaneous web requests,
but it can't handle 30 simultaneous AI jobs.

So, we're going to see a lot of delays, credit systems, subscription services with monthly billing, ...

Goodbye one-time purchases and fixed prices.

---

Alternatively, it would be cool if Affinity just had some kind of "AI server software" to self-host on a server as an "agent".
For companies, it is becoming more economical to buy a couple of servers with a couple of high-end GPUs.

(Maybe it sounds far-fetched. However, other companies are doing this stuff already. e.g. JetBrains is exploring similar solutions with their IDE software. i.e. You can install an agent on a strong server, and then connect from your laptop to it. It's all integrated in their software. The software looks as if it runs locally, but it's actually executing all commands on a remote server. e.g. my laptop with just 8GB RAM connects to a server with 128GB RAM, 12 CPU cores. And that server costed me just 2500€ and will easily last 10 years. The concept reminds me of the "mainframe systems" of the 90s, then again, that concept never really stopped making sense.)

Thats not fully true: To generate the model, you need absurd hardware.

To use the model, a desktop GPU is enough - with an RTX 3080 (so already quite outdated) it takes me ca. 15 sec to generate a 1k picture locally.

[Juan Garcia]

This is just the tip of the iceberg. Before generators, things like, remove background or select subject. macOS already has this out of the box. I know Affinity wants to be a macOS and Windows platform, but here is the truth. I do some designer work and some publisher. But when it comes to affinity Photo. I use Pixelmator Pro and Affinity Photo for a very very tiny subset of features that I always get issues remembering the exact place where they are. Affinity needs IA to basic handling.

[drstreit]

2 hours ago, Fossil said:

Affinity needs to act here. Exactly how, they need to decide, but they need to do something.

Whether it's partner with Midjourney, or support Stable Diffusion on a local machine, or an API or SDK, or some service of their own, they need to act and give users a good experience and a functional AI workflow that is better than just 'using Affinity' to fix your AI images.

Photoshop is already there. Firefly is already a thing. The plugin is already working. Affinity is playing catch-up.

If they don't catch up fast, they'll be left with a tiny niche market.

At the end, that's why I started this thread: The ABSOLUTE minimum would be to open their SDK for us - someone would implement first results within a week like they did with all other tools who offers an SDK.

Not doing anything, not even answering here or showing a roadmap at least is quite strange a behaviour...

[pixelstuff]

13 minutes ago, drstreit said:

At the end, that's why I started this thread: The ABSOLUTE minimum would be to open their SDK for us - someone would implement first results within a week like they did with all other tools who offers an SDK.

Not doing anything, not even answering here or showing a roadmap at least is quite strange a behaviour...

Doesn't Affinity Photo support regular Photoshop plugins?

[Juan Garcia]

1 hour ago, pixelstuff said:

Doesn't Affinity Photo support regular Photoshop plugins?

SDKs are limited by the capacity and the quality of the software underneath. AI via an SDK would be a lot of workfy to make it happen and thus, the plugin costly that maybe wouldn't justify the effort to create it if the potential customer realise they are better off paying an Adobe's subscription.

[drstreit]

4 minutes ago, Juan Garcia said:

SDKs are limited by the capacity and the quality of the software underneath. AI via an SDK would be a lot of workfy to make it happen and thus, the plugin costly that maybe wouldn't justify the effort to create it if the potential customer realise they are better off paying an Adobe's subscription.

No, only the most primitive ones

[Juan Garcia]

2 minutes ago, drstreit said:

No, only the most primitive ones

That's the point of my comment. Does the effort necessary to create this in Affinity, justify the amount of customers that would not prefer to pay for a Photoshop subscription? That is why this needs to come from Affinity. It needs to provide some AI. On macOS it is easy, on Windows not that much. 

[drstreit]

1 minute ago, Juan Garcia said:

That's the point of my comment. Does the effort necessary to create this in Affinity, justify the amount of customers that would not prefer to pay for a Photoshop subscription? That is why this needs to come from Affinity. It needs to provide some AI. On macOS it is easy, on Windows not that much. 

I agree, they have to embrace it. Just saying for some first impressions, an SDK would go a far way, seeing that all AI generators are accessible via API…

[drstreit]

12 minutes ago, Juan Garcia said:

SDKs are limited by the capacity and the quality of the software underneath. AI via an SDK would be a lot of workfy to make it happen and thus, the plugin costly that maybe wouldn't justify the effort to create it if the potential customer realise they are better off paying an Adobe's subscription.

I dont mean to create an AI via an SDK, just need one to integrate the existing ones…

[Juan Garcia]

3 minutes ago, drstreit said:

I dont mean to create an AI via an SDK, just need one to integrate the existing ones…

That's the problem. There is nothing AI there... Maybe the Selection Brush tool has some tiny bit of it... but they need to extend everything. There is nothing to offer to put on an SDK so people can start creating plugins. And taking in count the amount of time it took them to go from version 1.x to version 2.x it might be too late if they do not start now...

[deepz]

12 hours ago, drstreit said:

Thats not fully true: To generate the model, you need absurd hardware.

To use the model, a desktop GPU is enough - with an RTX 3080 (so already quite outdated) it takes me ca. 15 sec to generate a 1k picture locally.

Depends on your budget of course. For most hobbyists an RTX3080 and RTX4080 is out of budget. Nevertheless, your RTX3080 will still outperform a RTX4060. I would call it a "high-end GPU" (which is subjective again).

Providers may consider to do the calculations for their end-users. But with >1000 users, they are faced with additional challenges (beyond the GPU itself). e.g. you can't just put 8 RTX3080/RTX4080 GPUs in a single computer. You need enough PCI lanes and an absurd PSU (that may cost as much as a GPU).

Anyway, according to the EULA of Nvidia, cloud providers are not even allowed to use RTX3080 GPUs. What you can get though, are K40 (or K80, which are just twin-K40s). To give you an idea, here are the prices that AWS uses.

Which I would call, absurd prices, given the fact that these GPUs won't even outperform a GTX1080Ti. This shows that even the cheapest cloud GPU servers are expensive.

(Un)fortunately, there are auction websites like vast.ai that allow you to rent consumer grade GPU servers at lower prices. Having used that service for about 2 years, I can tell you that it's unreliable and not worth the pain. Just too many variables.

When I needed GPU servers for my company, I ended up self-hosting them, which also is quite expensive. (think fail-overs, multiple internet providers, smart routers, security, high energy costs). But it's more reliable than anything else I tried.

PS: Having said that, if you do the calculations client-side (on customer premise) you get faced with 50 different types of consumer GPUs, and you are going to disappoint some customers. e.g. CUDA won't work on an AMD GPU. GPU driver versions start to matter, which is something you can't ship with your software. Bigger networks perform better but won't fit in the memory of smaller GPU. You won't be able to run it on a tablet, and laptops will suffer from cooling issues. - All of which happened before, with other AI software.

[pixelstuff]

11 hours ago, pixelstuff said:

someone would implement first results within a week like they did with all other tools who offers an SDK.

Which other tool SDKs have an AI implementation already running on it?

[drstreit]

1 hour ago, pixelstuff said:

Which other tool SDKs have an AI implementation already running on it?

Photoshop https://github.com/AbdullahAlfaraj/Auto-Photoshop-StableDiffusion-Plugin

Krita https://github.com/sddebz/stable-diffusion-krita-plugin

Gimp https://github.com/blueturtleai/gimp-stable-diffusion

… masses of websites doing the same.

Thats for open source, many commercial AIs are also available to integrate, IF you have an interface to do so.

Not releasing a plugin SDK is a strange decision, I would argue that PS became a stabdard because of the power to extend it, its more an oecosystem than a single application today

 

 

[myclay]

4 minutes ago, drstreit said:

Not releasing a plugin SDK is a strange decision, I would argue that PS became a stabdard because of the power to extend it, its more an oecosystem than a single application today

unless I am misinterpreting you, I thought that there are plans to provide a plugin SDK.

 

[Fabricio Z]

It's not a question if it's art or not or if you like it or not. It's a business model decision and simply, if Serif does nothing about it, it will get behind others.

[pixelstuff]

1 hour ago, Fabricio Z said:

It's not a question if it's art or not or if you like it or not. It's a business model decision and simply, if Serif does nothing about it, it will get behind others.

That is a bit like saying if they don't add scanner support they will get behind others.

[Fabricio Z]

17 minutes ago, pixelstuff said:

That is a bit like saying if they don't add scanner support they will get behind others.

I'm saying that there's a reason people moved from Photoshop to Affinity, like why people moved from Premiere to DaVinci Resolve. And from that, Adobe learned and moved to improve it. That's part of the game. You can't get too comfortable thinking you're king. If Serif realizes they're losing ground due to not having AI implemented, you bet they'll do it.