Suggestion for an additional Affinity product, Affinity Generative AI

[William Overington]

I suggest an additional Affinity product, Affinity Generative AI with the AI system used trained only on art produced by Affinity Staff Artists and the original pictures and all of the legal documents involved included too. The end user could enter a text prompt, short or long as desired, even specifying which Staff Artist, the Staff Artist perhaps identified only by a stage name if so desired. Then the end user would have an original, royalty-free image to use. Maybe with built-in metadata and/or some in the picture watermarking system, unobtrusive to the eye yet detectable with specialist software.

One could be using Affinity Publisher, want an illustration, go through to Affinity Generative AI, specify a size and a text prompt, receive the picture, go back to Affinity Publisher and include the illustration in the document. An original, unique picture, not a piece of clip art or a stock photo known to others. Royalty-free too.

Some of the Staff Artists could each be featured more than once, using a different stage name each time, for example, once for portraits, once for traditional buildings, once for modern art, and so on.

Also, one could if one wished, use any of the pictures produced by the Staff Artists royalty-free too.

Offering Affinity Generative AI with only such carefully controlled content used to train the AI system and the legalities all carefully worked out and openly published would be an innovative leap forward.

William Overington

Friday 31 May 2024

 

[wonderings]

2 hours ago, William Overington said:

I suggest an additional Affinity product, Affinity Generative AI with the AI system used trained only on art produced by Affinity Staff Artists and the original pictures and all of the legal documents involved included too. The end user could enter a text prompt, short or long as desired, even specifying which Staff Artist, the Staff Artist perhaps identified only by a stage name if so desired. Then the end user would have an original, royalty-free image to use. Maybe with built-in metadata and/or some in the picture watermarking system, unobtrusive to the eye yet detectable with specialist software.

One could be using Affinity Publisher, want an illustration, go through to Affinity Generative AI, specify a size and a text prompt, receive the picture, go back to Affinity Publisher and include the illustration in the document. An original, unique picture, not a piece of clip art or a stock photo known to others. Royalty-free too.

Some of the Staff Artists could each be featured more than once, using a different stage name each time, for example, once for portraits, once for traditional buildings, once for modern art, and so on.

Also, one could if one wished, use any of the pictures produced by the Staff Artists royalty-free too.

Offering Affinity Generative AI with only such carefully controlled content used to train the AI system and the legalities all carefully worked out and openly published would be an innovative leap forward.

William Overington

Friday 31 May 2024

 

Why would you want something so limited to a small selection of artists?

You can already do this with AI by saying a painting style along with your text prompt. You are basically asking for something Photoshop and many other apps do, just in a limited way. Also I am not sure why you would want it as a stand alone app, integration into the apps is far more useful. AI is way more useful for fixing up pictures then it is for creating from scratch. 

[William Overington]

1 hour ago, wonderings said:

Why would you want something so limited to a small selection of artists?

Because of copyright issues and perceived copyright issues otherwise.

1 hour ago, wonderings said:

You can already do this with AI by saying a painting style along with your text prompt.

Yes, I have been researching what can be done using Bing Chat AI. I have obtained prints and framed them, just for my own art collection.

1 hour ago, wonderings said:

You are basically asking for something Photoshop and many other apps do, just in a limited way.

I have no idea at all what Photoshop will or will not do.

1 hour ago, wonderings said:

Also I am not sure why you would want it as a stand alone app, integration into the apps is far more useful.

Because I feel that some users of Affinity products would be delighted to have access to such a facility, and some users of Affinity products would be very much against any AI being anywhere near their work. So if Affinity Generative AI were a separate program that could be used alone yet could also be accessed through the other Affinity programs, then people who would like that facility could buy a licence and people who do not want AI anywhere near their work could avoid buying a licence and also be content that there was no AI in the Affinity programs that they are using.

As well as pictures, Affinity Generative AI could also write short stories, and poems, including haiku and sonnets and villanelles. Having only been trained on poems and short stories written by Affinity Staff.

1 hour ago, wonderings said:

AI is way more useful for fixing up pictures then it is for creating from scratch. 

I have no experience of using AI for fixing up pictures. I saw a video about it once. I have, however, used Bing Chat AI quite a lot for producing original pictures, mostly requesting in the style of paintings by a particular artist, group of artists, or artwork, such as The Bayeux Tapestry and the Lascaux Cave Art.

William

 

[wonderings]

1 hour ago, William Overington said:

Because of copyright issues and perceived copyright issues otherwise.

Yes, I have been researching what can be done using Bing Chat AI. I have obtained prints and framed them, just for my own art collection.

I have no idea at all what Photoshop will or will not do.

Because I feel that some users of Affinity products would be delighted to have access to such a facility, and some users of Affinity products would be very much against any AI being anywhere near their work. So if Affinity Generative AI were a separate program that could be used alone yet could also be accessed through the other Affinity programs, then people who would like that facility could buy a licence and people who do not want AI anywhere near their work could avoid buying a licence and also be content that there was no AI in the Affinity programs that they are using.

As well as pictures, Affinity Generative AI could also write short stories, and poems, including haiku and sonnets and villanelles. Having only been trained on poems and short stories written by Affinity Staff.

I have no experience of using AI for fixing up pictures. I saw a video about it once. I have, however, used Bing Chat AI quite a lot for producing original pictures, mostly requesting in the style of paintings by a particular artist, group of artists, or artwork, such as The Bayeux Tapestry and the Lascaux Cave Art.

William

 

- Not sure copyright is a big issue with creative images, from what I have read AI generated images in Photoshop are free for commercial use. 

- This is a personal thing, but art without humanity to me is just empty and hollow. Could be a story, music or art. Generated by something completely soulless just looks and feels horrible.

- If AI is integrated into the applications, no one is forced to use it. So again having it separate just makes it more cumbersome for people who would want to use it for creative professional work. It is an extra application, extra. updates, saving files, importing elsewhere. There is literally no good reason to have it stand alone. In Photoshop there is no gun to my head to use AI If I want AI I use it, if I don't I don't.

- Again this is a personal thing, but AI creating art is anti art. There is no soul, no history, no story. It is algorithms making something you know you will like. Will you be challenged, have your mind provoked by AI? I doubt it. 

- So instead of hanging up the works of masters, you have gone for an empty algorithm generated image? It is lower than hotel art. 

 

[William Overington]

I was considering your post and wondered as to what is "hotel art" as I had not known the term before seeing it your post.

A search for hotel art produced surprising results, for art hotels. Interesting yet not, I presume, not what you meant.

52 minutes ago, wonderings said:

Will you be challenged, have your mind provoked by AI? I doubt it. 

Yes, I have been.

William

 

 

 

 

[wonderings]

On 5/31/2024 at 3:00 PM, William Overington said:

I was considering your post and wondered as to what is "hotel art" as I had not known the term before seeing it your post.

A search for hotel art produced surprising results, for art hotels. Interesting yet not, I presume, not what you meant.

Yes, I have been.

William

Hotel art is basically mass produced commercial art. Generally pleasing and acceptable to all but entirely forgettable. 

How can AI provoke your mind when it is not created by a mind? In essence it has analyzed all the works and making a copy of the style with no substance behind there. There is no artist behind it with a story for the piece they have created, it is pure algorithm that can never have substance behind it. The world has no shortage of artists and art, with stories and emotion behind their creations. I find it sad to think anyone would put AI above even the worst of real artists who have lives they live or have lived, who have been through the ups and downs that have brought them to a place to create their works and in through those have moved other people. AI is giving you what you want, every time, all the time. There is no challenge in it, it is in essence a "yes man". 

 

[William Overington]

1 hour ago, wonderings said:

The world has no shortage of artists and art, with stories and emotion behind their creations. I find it sad to think anyone would put AI above even the worst of real artists who have lives they live or have lived, who have been through the ups and downs that have brought them to a place to create their works and in through those have moved other people. 

Something that I found some years ago is the following.

Museum Of Bad Art – art too bad to be ignored

Art deemed bad, yet it was produced by real people, for whatever purpose. It is easy enough for someone who was not there to deem it bad. Some of the pictures have a title that may not have been what the artist intended. For one of them, I thought of a different title that changes the meaning of the picture - but that may well not be the correct meaning either.

Yet why were they produced. One is thought to be from a facility of "send us a photograph and some money and we will paint a picture based on the photograph". Yet people involved.

Yet please have a look and then later I will put in a spoiler-protected note my opinions.

William

 

[William Overington]

Spoiler

I really like Two trees in love

Unlikely Landscapes – Museum Of Bad Art

,

Spoiler

One picture has the title Post Apocalypse,

yet what if it had the title

Dawn Redwoods and a White Birch in Winter

Would it then be viewed differently?

 

 

[Bryan Rieger]

On 5/31/2024 at 12:48 PM, William Overington said:

As well as pictures, Affinity Generative AI could also write short stories, and poems, including haiku and sonnets and villanelles. Having only been trained on poems and short stories written by Affinity Staff.

Warning: this post contains generative AI work for illustration purposes only.

Not to dismiss the many talents of the Affinity staff, but typically you'd want an AI trained on a MUCH larger dataset in order to be able to generate anything meaningful.

Anyway, you can do this already - check out Canva (you know, the folks who know own Serif/Affinity). Their 'Magic Media' and 'Magic Write' features are pretty impressive. For instance, I give you the following Canva Magic Write poem based on the prompt " a villanelle about a frog and toad", along with a Magic Media image generated using the following 5 word prompt 'frog toad pond moonlight friendship'.

Quote

### AI generated - for illustration purposes only ###

In a pond where lilies float serene and bright,
A frog and toad, best friends, do leap and play,
Beneath the moon's soft, silvery light.

Their croaks and ribbits echo through the night,
In harmony, they chase the bugs away,
In a pond where lilies float serene and bright.

Through reeds and rushes, green and slightly tight,
They splash and dive, their worries kept at bay,
Beneath the moon's soft, silvery light.

With every hop, they reach a greater height,
Their friendship grows, come what come may,
In a pond where lilies float serene and bright.

When morning dawns, and birds take flight,
They rest on stones, in sunbeams warm and gray,
Beneath the moon's soft, silvery light.

Oh, frog and toad, your bond's a lovely sight,
In nature's dance, forever will you stay,
In a pond where lilies float serene and bright,
Beneath the moon's soft, silvery light.

### AI generated - for illustration purposes only ###

It's not great (although it does have that 'generative AI smell'), but it's also not terrible, and could easily provide a great starting point for your own illustration and villanelle about a frog and toad.

I don't think a separate Affinity Generative AI app is needed (or a wise investment), just an SDK that enables others to add integrations to various AI services. Users can then opt to use, or not, any AI in whichever way they personally choose.

[wonderings]

4 hours ago, William Overington said:

Something that I found some years ago is the following.

Museum Of Bad Art – art too bad to be ignored

Art deemed bad, yet it was produced by real people, for whatever purpose. It is easy enough for someone who was not there to deem it bad. Some of the pictures have a title that may not have been what the artist intended. For one of them, I thought of a different title that changes the meaning of the picture - but that may well not be the correct meaning either.

Yet why were they produced. One is thought to be from a facility of "send us a photograph and some money and we will paint a picture based on the photograph". Yet people involved.

Yet please have a look and then later I will put in a spoiler-protected note my opinions.

William

 

Yes, people make bad art, but I would take something produced by a human over AI anyway of the week. This is not to be confused with saying anything humans produce is good, or to say AI will not make something nicer to the eyes. The difference is a real live human being created something and a computer algorithm made something. One is a being with history, with hurts and joys, laughter and tears, and everything in between. AI again is programming, and pretty amazing programming, but lacks humanity. With so much beauty in the world, but natural and created by people, why anyone would settle for algorithm based generation baffles me. And I would love to know how any art created by AI has challenged you. What AI has done is given you exactly what you ask it for, where is the challenge in that?

[William Overington]

30 minutes ago, wonderings said:

 What AI has done is given you exactly what you ask it for, where is the challenge in that?

Well, yes and no.

I have entered some text as a prompt. The AI system has acted on that prompt. So, yes, the poem, or song, or story, or picture often, usually, is in some way qualitatively what I asked it to produce. But the result can be something that surprises me and causes me to smile with delight.

Alas, due to the restrictions imposed by some of the moderators I am not allowed to include examples here, and I am not allowed to link to them.

41 minutes ago, wonderings said:

With so much beauty in the world, but natural and created by people, why anyone would settle for algorithm based generation baffles me.

For me it is because I do not have the ability to produce paintings of that quality myself, and I find it interesting to be able to express a picture that I would like in text and then see the result that the AI system produces. I am a researcher so I find exploring what AI will produce is fascinating.

While I do look at what other people produce, for me, just watching the world go by from the sidelines is not enough - I need to produce things myself.

Generative AI has allowed me to produce beautiful pictures that may well never have existed otherwise as I could not have painted them and I could not have afforded to pay an artist to paint them.

William

 

[wonderings]

17 minutes ago, William Overington said:

For me it is because I do not have the ability to produce paintings of that quality myself, and I find it interesting to be able to express a picture that I would like in text and then see the result that the AI system produces. I am a researcher so I find exploring what AI will produce is fascinating.

While I do look at what other people produce, for me, just watching the world go by from the sidelines is not enough - I need to produce things myself.

Generative AI has allowed me to produce beautiful pictures that may well never have existed otherwise as I could not have painted them and I could not have afforded to pay an artist to paint them.

William

 

I hope this is not the dawn of a new type of artist, the type that enters prompts and then when AI creates something they take credit based on the prompts they entered in. 

What are you researching with AI, are you going into the code and algorithms and how these things work and generate? Or is your research simply seeing what comes out when you type something in? I would suggest part of your research to look into the importance of art and the philosophy of art. 

Again AI is a "yes man" giving you what you want, from writings and all aspects of the arts where AI is in use. Maybe you have a favourite character or story by Chesterton, you want a specific story with that favourite character. AI is going to give you that, but gives you what you want with some embellishments as it pulls it all together. There is no real creativity here and you get what you give AI. It then connects the dots by fill in what is necessary. 

Why not do your research by actually learning how to paint? There is no talent in AI, but the energy put into learning brush strokes and colour will be far more rewarding. 

 

 

[William Overington]

18 minutes ago, wonderings said:

What are you researching with AI, are you going into the code and algorithms and how these things work and generate?

No.

18 minutes ago, wonderings said:

Or is your research simply seeing what comes out when you type something in?

Well, I suppose that the answer is yes. But I like to think that it is also in deciding what to type in and critically assessing the output.

22 minutes ago, wonderings said:

I would suggest part of your research to look into the importance of art and the philosophy of art. 

Thank you. I had not thought of that. I will try to look into those.

24 minutes ago, wonderings said:

Again AI is a "yes man" giving you what you want, from writings and all aspects of the arts where AI is in use. Maybe you have a favourite character or story by Chesterton, you want a specific story with that favourite character. AI is going to give you that, but gives you what you want with some embellishments as it pulls it all together. There is no real creativity here and you get what you give AI. It then connects the dots by fill in what is necessary. 

You might like to have a look at the output that has been achieved by looking at the AI threads on pages 1 and 2 at the following place.

Art & Literature (Page 1) — Alfred's Serif Users' Forums (punster.me)

26 minutes ago, wonderings said:

Why not do your research by actually learning how to paint? There is no talent in AI, but the energy put into learning brush strokes and colour will be far more rewarding. 

Well, I am trying to do that.

William

 

[wonderings]

16 minutes ago, William Overington said:

No.

Well, I suppose that the answer is yes. But I like to think that it is also in deciding what to type in and critically assessing the output.

Thank you. I had not thought of that. I will try to look into those.

You might like to have a look at the output that has been achieved by looking at the AI threads on pages 1 and 2 at the following place.

Art & Literature (Page 1) — Alfred's Serif Users' Forums (punster.me)

Well, I am trying to do that.

William

 

- deciding what to type is what I think will become the lazy artists claim to fame. It was my words that brought about this art, thus my defining words are art! As Scrooge would say "Bah! Humbug!" 

- I am well aware at what AI can accomplish and create. I use it daily in print for fix images. I have played around a fair bit with AI, been using it since the betas with Photoshop and exploring its capabilities. As a tool for fix ups it is amazing and helpful and saves so much time. This is where I think it has its place, at least in the creative commercial fields. I think it has no place in the arts field and diminishes and cheapens all art. 

 

[Bryan Rieger]

57 minutes ago, wonderings said:

Deciding what to type is what I think will become the lazy artists claim to fame. It was my words that brought about this art, thus my defining words are art! As Scrooge would say "Bah! Humbug!" 

I've run across a few art directors at various agencies that have been trialing 'AI artists' for some time, and have recently recommended that their agencies no longer commission AI generated (prompted) art.

A couple of problems seem to come from 'artists' who only work with AI prompts. First, they often lack any proficiency with Photoshop that would enable them to tweak the generated images to meet the agency briefs, and instead the keep modifying their prompts in the hope that the AI will somehow magically incorporate any art direction supplied. Unfortunately, the continued prompting often makes things worse, which then causes the second problem. Because these 'artists' have no formal training in the arts, they've never been exposed to critique or art direction, which means very often they take any criticism or direction as a personal attack on themselves. Needless to say, this cause a lot of friction when working to an agency brief.

We're in the gold rush (where the folks who make the shovels win) days of AI (ahem, Nvidia), and while there's LOTS of excitement, generative AI generally hasn't proven itself as a solution to any problem most folks actually have. Yes, it creates images based on a prompt, but the majority of the images have a definite 'AI smell' to them, which has recently been labelled as 'AI Boomer Art' by younger folks (I'll leave it to you to do a web search). It's also interesting that there's been a noticeable uptick in younger artists creating using analog materials (paint, charcoal, clay, etc).

Personally, I think the prompter AI 'artists' are likely soon to go the way of all of the NFT artists from a few years back. If you can't make a living creating prompt generated AI art today given how cheap the services are (many companies currently view AI as a loss-leader), there's no chance you're going to pay the adjusted standard market rates when things cool.

[William Overington]

 I neither describe myself, nor think of myself, as an "AI artist".

I wonder if I may include in this discussion something not about AI.

I saw something somewhere sometime about an art exhibition where it was all, and only, wall panels with text of the kind that are normally used to describe a work of art that actually exists and is on display in the art exhibition. So the idea was to imagine the artwork from the description. 

I wonder if there is something on the web about it.

William'

 

[Bryan Rieger]

14 minutes ago, William Overington said:

I saw something somewhere sometime about an art exhibition where it was all, and only, wall panels with text of the kind that are normally used to describe a work of art that actually exists and is on display in the art exhibition. So the idea was to imagine the artwork from the description. 

The art is a concept here, which asks questions about the nature of art, of creativity, interpretation, etc by presenting the audience with a blank slate and asking them to imagine the work from the description. There's been a number of artists who have executed similar concepts over the years. Cy Twombly did this back in the late 50's, where he 'painted' works with alluring titles (such as 'Poems to the Sea'), using only a few pencil lines, brush strokes, and scribbled words to depict the subjects, compelling the audience "to read what is there, but not fully manifest in the artist’s scrawled script."

[William Overington]

@Bryan Rieger 

Thank you for your post.

Influenced by your post mentioning the title 'Poems to the Sea' (I have not at the time of writing seen the picture) and by a picture entitled Aurora Borealis at The Museum of Bad Art,

Unlikely Landscapes – Museum Of Bad Art

But that title was not necessarily the intention of the artist.

both of which I saw yesterday, and have intertwined in my mind.

I have produced the following

I have deliberately not added a title that might influence how people view it.

There is one reply already.

William

 

[William Overington]

16 hours ago, wonderings said:

- deciding what to type is what I think will become the lazy artists claim to fame. It was my words that brought about this art, thus my defining words are art! As Scrooge would say "Bah! Humbug!" 

Possibly.

Yet just because lazy artists may do that does not mean that everyone who decides what to type as input to an AI system is a lazy artist.

To argue otherwise would be to be by a fallacy, the Fallacy of the Undistributed Middle.

I have produced custom greetings cards each with an AI generated image on the front, and where normally the greeting would go I put a title, a statement that it is an AI generated picture and the prompt that I used. I state that I authored the prompt, and add the date. That seems perfectly proper.

William

 

 

 

[wonderings]

10 minutes ago, William Overington said:

Possibly.

Yet just because lazy artists may do that does not mean that everyone who decides what to type as input to an AI system is a lazy artist.

To argue otherwise would be to be by a fallacy, the Fallacy of the Undistributed Middle.

I have produced custom greetings cards each with an AI generated image on the front, and where normally the greeting would go I put a title, a statement that it is an AI generated picture and the prompt that I used. I state that I authored the prompt, and add the date. That seems perfectly proper.

William

It is completely lazy, there is zero effort and you are at the whim of what AI interprets your words to mean. Type the same thing into Photoshop and Krita and get very different results. I would not call anything AI makes art, I think it is bastardizing everything it touches for the means of art. If you truly want to get into philosophy of art I would highly recommend this video. The BBC has refused to air it, they technically own the rights. Though in emailing the Roger Scruton Legacy Foundation regarding this piece, this is where they directed me to watch it. 

 

 

16 hours ago, Bryan Rieger said:

I've run across a few art directors at various agencies that have been trialing 'AI artists' for some time, and have recently recommended that their agencies no longer commission AI generated (prompted) art.

A couple of problems seem to come from 'artists' who only work with AI prompts. First, they often lack any proficiency with Photoshop that would enable them to tweak the generated images to meet the agency briefs, and instead the keep modifying their prompts in the hope that the AI will somehow magically incorporate any art direction supplied. Unfortunately, the continued prompting often makes things worse, which then causes the second problem. Because these 'artists' have no formal training in the arts, they've never been exposed to critique or art direction, which means very often they take any criticism or direction as a personal attack on themselves. Needless to say, this cause a lot of friction when working to an agency brief.

We're in the gold rush (where the folks who make the shovels win) days of AI (ahem, Nvidia), and while there's LOTS of excitement, generative AI generally hasn't proven itself as a solution to any problem most folks actually have. Yes, it creates images based on a prompt, but the majority of the images have a definite 'AI smell' to them, which has recently been labelled as 'AI Boomer Art' by younger folks (I'll leave it to you to do a web search). It's also interesting that there's been a noticeable uptick in younger artists creating using analog materials (paint, charcoal, clay, etc).

Personally, I think the prompter AI 'artists' are likely soon to go the way of all of the NFT artists from a few years back. If you can't make a living creating prompt generated AI art today given how cheap the services are (many companies currently view AI as a loss-leader), there's no chance you're going to pay the adjusted standard market rates when things cool.

I can see why, again you get a lot of lazy people tossing out words, getting something, plopping it in their design without really scrutinizing it. The AI smell I would say is the hollowness of AI art, it lacks soul, depth and anything real. 

 

 

[William Overington]

17 minutes ago, wonderings said:

It is completely lazy, there is zero effort and you are at the whim of what AI interprets your words to mean. Type the same thing into Photoshop and Krita and get very different results. I would not call anything AI makes art, I think it is bastardizing everything it touches for the means of art.

Writing a prompt to an AI system for the AI system to produce a painting is not necessarily lazy, it depends why it is done.

It is not zero effort.

Yes, I am at the whim of what the AI decides to produce. I can accept or reject what the AI produces.

One can reset the same AI and use the same prompt again and get very different results.

Sometimes I am delightfully surprised at what is produced. For example, when I asked for a painting in the style of Claude Monet of a lady in a long green dress feeding an okapi in a garden with a pond and flowering water lilies. Even more so when I edited the prompt and changed "an okapi" to "a stegosaurus" to observe whether the AI would produce that and it did.

William

 

 

 

[Bryan Rieger]

I ran across this video from OH no Type Co on Instagram today in which he talks about how he uses AI in his work. Rather than simply using AI to generate 'new' work from scratch, he uses it to enhance his own original work, which he uses as input. He then takes the output (feedback) from the AI and further refines it in his own designs. 

I think this is a great use of AI in that it provides the artist with the ability to rapidly explore multiple variations of their own work. This 'AI feedback' is in many ways not dissimilar to some of the feedback artists might routinely receive from peers, but it's immediate and doesn't pull them too far out of their workflow.

This is a great example of AI as a tool to enable artists, rather than as a tool to replace artists.

I've include the video itself below, but the comments in the original post https://www.instagram.com/p/C7pk-dKSGoH/ are definitely worth a read.