Discussion about AI created banner

More likely me?

TBH this discussion illustrates that concerns on AI imagery are not trivial.

They reach way beyond TW itself.

On the other hand I have no interest in bogging down TW discussions in the philosophies of the relation of ideologies to actions. Relevant as they are.

QED, TT

Who is saying that here? To say, more modestly, “most AI is weak” is to affirm that the slick outpourings very evident are a gambit rather than a realization. Seems true.

You are right in questioning.
But any blind acceptance that ''AI is good" won’t do.
Are you saying that?

Get your case studies in gear broheim to make the point you haven’t yet.

Best, TT

So I go to tiddlywiki.com to see what the last banner looked like, expecting it to be on the landing page, and I have to do a search for it.

Once again, especially given that it’s not part of the front door of TiddlyWiki, why not let the voting speak for itself?

I think both you and Jeremy used pejorative terms in describing AI. But I could be hallucinating.

I was hallucinating when hallucinating wasn’t cool.

I see it on the splash screen, but blink and it’ll be gone:

image

I think creating banner images is hard. At least for me. I was lucky, that the last one got selected.

We had a clear “change log”, which pointed out, what’s new. Some of the changes where visual.

So making some screenshots was easy to get it started. I did create a similar collage many years ago using Gimp. I could remember some steps of the workflow and the others I had to look up in the internet.

With the current release, the only visual improvements are very minor. We removed some bugs in the current palettes, which still have a lot of room for improvements.

Then there is the “transition” from v5.3.7 to v5.4.0 – So v5.3.7 is probably an “end point”. – But How do you “visualise” the end of a journey. I did like @Christian_Byron idea with a sun-set or sun-rise.

So I thought about several journeys I made, where I saw both beautiful sun-rises and sun-sets.

a) We had an apartment in south France, with a gorgeous view to the sea. There was a street that meandered along the coast line from the left to the right until it could not be seen anymore. That was the first image in my head, where it probably is more beautiful than it was in reality :wink:

b) I made a trip with some friends on a sailing boat, in Greece, visiting the Ionian islands. A lot of sun-sets, sun-rises. The finest food and something to drink. And finally the most relaxing and beautiful way to travel, sailing with the wind. (and motor, if there is no wind :confused: – or the rest of the crew puking, if there is too much wind :slight_smile:

With these two memories in mind I started to create a fictional scene as a prompt for the LLM.

That’s the initial prompt

You are a photo artist.
I need a photo-realistic image with 560 x 315 pixel
It should be png format
Your position is in a Mediterranean setting.
You are on a rocky coast-line
In the left side of the image is a small village 
5-10 houses with flat roofs
they are painted white
windows and doors are painted some blue and some orange
to the left of the village is an olive hain
there is a main street coming out of the village which leads along the whole cost line
the coast line goes from the middle left of image, down to the center and up again in the far distant right.
The rest of the image is sea
The sea is calm only small braking waves are visible 
They come from the top left and go towards the coast line
There is a sunset in the distance top right of the image
Very far from the left, there comes a container ship driving towards the sun
the container ship is far away so it is small but visible
In the center there is a sailing boat docked at a small footbridge
The sailing boat is the main focus point of the image
The sails are halve retracted, but still visible

The initial image was this:

image

Not exactly what I had in mind, but looked good and contained all the elements that I did describe. It could be a photo I made somewhere in Greece.

But then “the journey began”. – I think I made a mistake. I did add the following text, instead of refining the full prompt.

That's nice already, but the sailing boat should be a little bit more in the deeper water.
The street should go with a small bridge from the left side of the image to the right side and out of the image.
the boat should be docked on small dock, a bit closer to the vilage.
The ship should have full sails, but it should already be docked.

image

Most of the new text was ignored, but I did like the small dock on the right side.
As you can see, some of the houses disappeared. The boat was a bit bigger, which was not intended.

An other refinement

no. the sailing boat should have full sails all of them.
It should be docked an a small dock that leads from the street near the vilage into the see.
At the bottom of the image there should be an ancient bridge from the left side of the rocks to the right side of the rocks.
the bridge should be a bit more at the bottom at the image, but ist should still be visible.
The ship should be a bit more in the center of the image, but it should still be the main focus point.
the dock goes from the street near the vilage into the see.
the ship is docked at the small dock

Result:

image

I did like the new bridge in the front. But the image becomes way too noisy and more houses are gone :wink: – Full sails - docked - What a stupid idea. More houses gone. No orange elements anymore.


I wanted to start a new → but the LLM did not forget the earlier context. 3 more images and “corrections” and it got worse. I managed to completely mess up the bridge in the front, which was completely my fault. As you can see in the last image, the bridge comes from outside and leads into the water. But I did like the upper 2/3rds of the image.

image

Final request:

Take the photo and convert it into a semi realistic oil painting

image

Not bad. → Starting Gimp → Cut the bridge → Add the version number → Submit.

Why the long post?

I did read this thread and thought more background is needed. Now with the background, there are some questions which I want to ask:

Q - to myself: Could I have created the “end result” on my own?
A: No. I can not paint with oil and I do not know Gimp enough, that I could create that scene with Gimp alone. I could have used some of my own “real” images, but manipulating them in a way to get this result, would have been impossible for me.

Q: Is the “process” that lead to the image a “creative process”?
Q: Is the result worth to be accepted as a contribution as a banner image?
Q: Does it “visualise” a start or end of a journey?


I have to say, it was fun, to create it and I do like the scene. For me it’s beautiful and that’s why I did share it.

Have fun!
Mario

2 Likes

I think what we’re doing here is engaging in the kind of deliberative reflection that can make for a meaningfully informed community vote.

Also, I’m interested in learning more about whether — if I were to put more hours into developing my graphic ideas — a significant number of people are still interested in that. I’m not looking for a guarantee of winning. I don’t really care about that. I’m looking to understand whether the majority now share this appetite for big-data-fueled heavy images, and whether the majority here think the details matter — in terms of of where and how this tech is being integrated.

3 Likes

I can only speak for one person — myself — but I appreciate your ideas and your work and I hope you continue to pursue them!

And in general, I would greatly prefer to vote for a banner that’s not the product of machine learning. While I appreciate the thoughtful choices behind the AI-assisted submissions, I don’t think incorporating AI in our branding sends the right message about TW’s priorities; it seems at odds with the idea that users have full control over their content and the ways it is (re)used.

2 Likes

I’m not much of a visual person.1. So I generally only half-heartedly pay attention to the banner competition threads; they simply don’t matter much to me. But I do read them, and sometimes make my preferences known. Almost always that comes down to a preference for simplicity. While @pmario’s ship is pretty, and @stobot’s trains are impressive, I prefer the simpler ones such as yours, and the ones that speak to something about the actual changes, like @pmario’s 5.3.6 version, and both of Xrizzy’s offerings. I was happy enough with @Christian_Byron’s cat looking into the sunset. I’m not experienced enough with AI images to recognize what @TiddlyTitch seemed to grasp right away, that this was AI output. That sours me a bit, but doesn’t disqualify it in my mind, even though I find myself significantly opposed to the current crop of AI.

So yes, I will absolutely enjoy seeing the sorts of things you create. And for a first approximation, simple is better in these banners.



1Really not at all. For instance, when I read, I do not get an interior image of the scene. I can tell you relationships: this is to the right of that, but further back or both of them look scared, but I don’t “see” it, even in my mind. Since my mother and one of my brothers share this, I assume there’s some genetic oddity at work. But I’ve never really investigated.

2 Likes

3 posts were split to a new topic: Aphantasia - the inability to voluntarily visualize mental images

Wow. Could you do the Harry Potter thing, and make old pictures with friends and family come alive? That might be worth spending $60 one month.

make old pictures with friends and family come alive

I’m certain we will get there. No joke. Not alive as in that it is really them, but you won’t be able to tell the difference. Or, well, at a later stage it will almost fully be them… when it is possible to make a genuine digital version of individual brains. In a way we’ll become immortals then… or at least copies of us.

Interestingly, my main TW5 knowledgebase I call “NemoBrain” (or, since it’s a node install, the full title+subtitle is “NemoBrain As A Service” - as a followon from the TWC version which was “NemoBrain On A Stick”)

Part of that knowledgebase is errata to my will. One of my errata notes is

  • Strongly request that no “AI” recreation of my personality be made

I’m not sure how or why a person (or any other animal) would be reducible to a brain…? The symbolic register of code-like inputs and outputs is just the tip of the iceberg of communication and interactive competence. Even our nervous system is distributed throughout the body, with the gut (and its microbiome) and various other chemical signal-channels having a huge effect on a person’s temperament, sensibilities, and attitudes — not to mention actual skills in the world.

1 Like

Well…

OK, I mean whatever organ(s) we have that create our experiences of the world and sense of self. I was pretty sure this was limited to the brain but maybe other parts also create this experience. But in a digital clone of the full organism (as in the previous link I gave, which is reality!), I’m presuming that it allows the scientists to simulate anything to be input/output to the brain (…in that case a worm, so it is not quite a brain).

I’m a strong believer in eventual AI and in the science-fiction possibility of uploading our brains to a computer network. But I think we’re many orders of magnitude away from this. I don’t know how much of the brain’s network complexity is necessary for intelligence, but the brain has 1014 – 1015 synapses I don’t think we have LLMs yet with as many as 109 parameters. Given how power-hungry AI is now, it’s hard to see it scaling up 100,000- or 1,000,000-fold. Not without a radical change to the architecture.

The current crop of AI tools seem to me to be overhyped nonsense. Of course, my main experience with it is fixing the code of junior programmers who use it to get to 80% of the requirements, and have no idea how to fix it to get to the 95-99% we expect from them. You would think that if things like Copilot were so obviously beneficial, then Microsoft wouldn’t need to be forcing it down their employees’ throats.

1 Like

Certainly I’ve not refrained from raising negative points about it. Is it pejorative to call it consumptively ravenous — dependent on a vast diet of creative work (culled from an unaccountable archive of selected human work, digitized) and carbon-captured energy (unsustainably burned up as petroleum-based fuel)? I think it’s just true, however unpleasant.

Some expect AI to “solve” the problems of energy use and climate volatility. I’m no physicist, and no economist. But I’m on familiar terms with quite a few, and the idea that these technologies will usher in an era of abundance and harmony depends on incentive-structures that aren’t in place… and there are no incentive-structures in place to help better incentive-structures come into place.

In other words, so-called AI (along with robotics) is currently functioning as the long arm of capitalism, and the purpose of the intelligence-extension it offers is ultimately the shoring-up of access to resources for an increasingly small set of folks in the cockpit.

The amusement-part loss-leader here — free access to LLMs and image-toys — is a necessary step to growing the footprint of familiarity and attention. It’s a bit like giving away “free” formula to all new and exhausted mothers, after which we should not be surprised that they are subsequently dependent on buying formula.

For those who’d say I’m naive because similar technological shifts have happened in the past, and it hasn’t been terrible…, I’d say of course this is the continuation of a long trend of consolidating power through certain unifying technologies (also shifting what people experience as “needs” along the way). The printing press and the automobile and the word-processor were not similar distinct events, but part of the same general trend. It’s not that I personally experience this as terrible for me, now… But there are plenty of people on the planet already living in the hot exhaust-cone of “our” ravenous appetite for efficiency, convenience, and entertainment.

4 Likes

Obviously they cannot. As well as the other biological bits of the nervous system, we can’t forget the social networks we inhabit. I think the platitude of “she lives on the hearts of all who remember her” is more of a literal truth. We are not only biological beings, but social ones.

But we can imagine a brain transplant: the brain of a person whose body is failing transplanted into the body of someone whose body is fine but whose brain has died. We would expect to wake up disoriented and to have lots of relearning to do to control our new muscles and understand the signals now supplied to our brain. But we would still, I contend, feel substantially like ourselves. We would certainly not think of ourselves particularly as hybrid beings. And we would certainly identify much, much more closely with the self we were in the old body than with the self of the new body.

Similarly with social networks. If we were to wake up in an entirely different place, in an unfamiliar landscape and culture, perhaps not knowing the language, we would still feel substantially like ourselves.

So a great deal of who we are can be localized in our bodies, and further localized in our brains. Not all, but a great deal.

(If you find these sorts of thought-experiments about the brain and identity interesting, then I highly recommend Hofstadter and Dennett’s The Mind’s I, which is chock full of them.)

1 Like

Hah! I second that recommendation! That’s the first time I’ve seen anyone other than myself bring it up! Apropos; I was considering if I should link to Nagels “What’s it like to be a bat”, which I believe is inluded in The Minds I, when I responed to Elises description of how she interprets the world in some other thread here. And the excerpt from Dawkins “The Selfish Gene” has had a great impact on my understanding of biology.