mailto: blog -at- heyrick -dot- eu

Jazzing up RISC OS

The traditional RISC OS desktop has changed little in decades, most of the changes have been around different icon sets. Over on the other branch, the icons had the option of being rounded. And... that was about it. The biggest visual changes were in RISC OS 3 bringing icons with different sorts of borders instead of the flat rubbishy RISC OS 2 method, and RISC OS 3.5 making using outline fonts "official" so we could finally put the VDU font to sleep.

Admittedly, that still is it. Window furniture is fixed in place, the "close button" is over on the left beside the window stack control, the toggle size is over on the right. You can change the desktop theme and on some versions make the icons look a little different and... uh...

Mom used to say that she knew when I was using RISC OS. It wasn't "the monitor on the right" or anything like that, it was simply "the one that looks old-fashioned".

But this doesn't mean that software cannot be added in order to improve the look of the desktop a little.

A better looking desktop
A better looking desktop.

 

A better clock

To this end, we should begin with MiniTime to pop a clock on the desktop. Having it on the iconbar thanks to Alarm is so 1987.
Pick up a copy of MiniTime from Fred Graute and set it up as you like.
If you want to have it always available, drop it into $.!Boot.Choices.Boot.Tasks and it'll start up automatically.

 

A better backdrop

The next thing that makes a big difference is Pinboard2. Please contact R-Comp regarding this, there's no public release at this time as it is currently undergoing testing (as can be seen by the "Beta build" message at the lower left of the screen).
However this brings into play a number of useful additions to RISC OS. For a start, it's much better at sizing images to fit the screen. The default version of Pinboard built into RISC OS was not terribly flexible in this respect, and it also sized images to fit into the visible part of the backdrop so oftentimes things would be slightly (or not so slightly) squished.
One that I find dramatically makes the desktop look better is the ability to make the iconbar translucent. This also copes with a full iconbar scrolling left and right, and when used with background blending can really look quite impressive.

I've not used it, but it would seem as if you can also use the backdrop as a place to dump files (à la Windows), if that sort of thing is useful to you.

 

Keeping the backdrop fresh

Using my backdrop changer, either with the standard Pinboard or with Pinboard2, can help keep things interesting by coming up with a new picture each day, a little like how Windows can... only you'll need to sort out the images yourself as there's no daily feed or anything.

 

Tree view Filer, anyone?

If you are a fan of the treeview list filing system available on other filing systems, then you might like Cloverleaf's CLFiler. A demonstration version is available from
A tree view Filer
A tree view Filer.

Personally, I don't care for the tree view layout. And as for CLFiler, I can't throw a lot of criticism at it because it is an early test release, but given that it doesn't seem to do many additional things (certainly mine refused to make thumbnails of JPEGs or sprites, for example), it's rather astonishing that it's a 2.64MiB executable that claims an 8MiB slot... for a Filer. I really cannot see it ever replacing the RISC OS Filer when it, alone, is over half the size of the entire rest of the OS.
It also doesn't understand read-only Shares, so you'll not see any of those.

Anyway, if you're interested in trying out CLFiler, get a copy of the demo version and see what you think.

 

French cake, take two

Last week I got myself a pack of French cake mix. Alsa's natural flavour cake. The instructions tell you three eggs and 200ml of cream get mixed up for two minutes, and the resulting slosh goes into a big tall cake tin and bakes for fifty minutes.
I read that and was like "what the hell?". But, okay, it said to do that so I did that.

How not to make a cake
How not to make a cake.

I don't know what passes for "cake" in the minds of Frenchies, but the result of following the instructions was suitable only for the bin. Now, I'm not averse to sad cakes, but the wreck that I ended up with was sad in the sense of "a thousand kittens just dived into a wood chipper". I don't even have a photo, it was that depressing.

However, since the mix wasn't that expensive and my only loss was time and energy to heat the oven (plus three eggs and some cream), I decided to try again. Only this time screw the instructions. Sorry France, but really, you guys seem to understand cake like you understand tea.

I made the mix up as it said, and measured it to be about 800g of cake mix. Which seemed like a nice amount to put half of into one regular cake tin to cook.

One cake cooks
One cake cooks.

While the other waits (it's a dinky electric oven).

One cake waits
One cake waits.

If I had a third tin, I could have made a cake queue.
Sorry, obscure Brit joke. ☺

The first cake became in just over 25 minutes. While the second cake was cooking, I force-cooled the first using a desk fan on the highest setting. By the time the second cake was ready (about 35 minutes, so clearly my idea of "half" needs work), the first was cool enough to put into the fridge.

I then force-cooled the second cake, and when it was cool enough I put the first cake into the freezer (lacking fridge space) and the second cake on the shelf where the first cake was.

I then decided since I had a bowl and a mixer and some left-over cream, to dump some fine sugar into the cream and whisk fifty shades out of it to make a nice stiff chantilly... to then enjoy eating.

Whipped cream
This, of course, is partly why my BMI is 24 and some fractions...

Back to the cake, about 300g of icing sugar and maybe 30ml of milk and a few drops of vanilla flavouring. I wasn't measuring it, just adding milk until I got a gloopy paste. I placed a cake on to a plate and iced the top.
I then placed the other cake on top of that and... it turned out to be slightly larger. Wait, what? 'twas the same size cake tin FFS!
Anyway, I iced this. And then poured and smooshed the rest of the icing so it dribbled down the sides.
And, of course, hundreds and thousands this time. I remembered! ☺

The only change I'd make is mixing up some of the vanilla with the milk, because adding it to the sugar and milk didn't really work out. Still, it's a lesson for next time.

Now this looks more like a cake
Now this looks more like a cake!

Back in the fridge, this time to let the icing harden. And at this point it's worth laughing at the instructions on the box that say "Preparation time: 3 minutes". Well, it is if you're going to toss eggs and cream into the mix and toss the resultant mix directly into the bin. Well, you could put it into the oven, but you might as well save the energy and pour it all into the bin for all the good it'd be.

Or attempt to make something that resembles a cake, in which case aim for about two and a half hours of care, attention, and love. Yes, you need love when making cakes. If you don't actively want to make a good cake, the resulting cake will show it.

Another hour passes, long enough to listen to a selection of ShadowIcon on YouTube whilst writing the above stuff about RISC OS.

Now? Now it's time.

Finally - cake!
Finally - cake!

It's a little heavy. Could be the use of cream rather than milk, perhaps. But it's moist and fluffy enough. It will be adequate cake to accompany a cup of tea.

 

More messing with Dall-E

Time to give the burgers and ninja kittens a rest and try some other ideas. You'll notice some similarity in the text used for some of these, I simply changed a few things to see how it would differ.

  • If you want to use any of these images non-commercially, please credit me and OpenAI, and also leave a comment below pointing to where/how you have used the image as I'm interested to know how these pictures may be of use to people.
    If you wish to use any of these images commercially, please get in touch to discuss licencing.
    The images here have been blog-sized to 680px width (and thus 680px height). I have the 1024×1024 originals, get in touch...

 

"Realistic photo of a happy extra terrestrial in a Neolihic town at sunset."

A happy alien in a technicolor world.
A happy alien in a Technicolor™ world.
The colours are a bit cartoonish, so I'm not sure "realistic" is appropriate.

 

"Realistic photo of a creepy tall man in a Neolithic town at sunset with stormy clouds, gothic."

Whoa.
Whoa.
Like the caption says, "Whoa.". That's is some real Reign Of Fire shit right there. We're only missing the dragons. I like that Dall-E has decided to go with ruins and the guy wearing a badass leather jacket. This one is just dripping with atmosphere. It's almost a movie poster, isn't it?

 

"Realistic photo of a scared young woman in a Neolithic town at sunset with stormy clouds, gothic."

Woman in an old town
Angela Rayner takes the wrong train, ends up at the Tory Conference by mistake...
Sadly it kind of messed up the face. Could be fixable with some Photoshopping but, generally, Dall-E has trouble with faces.

 

Same prompt as above.

Woman in an old town
Better, but...
This is almost good, but misses because of the eyes. We'll come back to this.

 

Let's take this up to eleven.

"Realistic photo of a scared young woman in a Neolithic town at sunset with stormy clouds and fireballs raining from the sky, gothic."

End of times
End of times.
She's positively cacking herself.

 

Sorry, I've not done playing with my food.

"A toaster made of knitting flying through a sky filled with glowing balls of scrambled egg."

No comment
No comment.
I like that Dall-E will attempt to do as requested, rather than hand back an empty square with the text "Dude, WHAT?" in the centre.

 

"A teenage girl wearing a British school uniform and holding a flaming katana."

A schoolgirl holding something
A schoolgirl holding... something.
I like how Dall-E does a fairly decent attempt at a school uniform and then makes it "British" by plastering a Union flag on it. But, Dall-E, how many shirts is this girl wearing?

This was a generation based upon the one above.

A schoolgirl holding something
A schoolgirl holding... what?
Now, two things can be observed from this image. The first is that Dall-E clearly has no idea what a katana is, and looks like it has pasted in the Olympic torch instead.
The second is the eyes, again. Now, my personal opinion, based purely on observation, is that Dall-E knows how a girl is constructed. Where to put the arms, legs, and so on. And it generally does a reasonable job here, but where it seems to fail is when constructing faces. Dall-E understands eyes and where to place eyes, but it appears as if it's a routine to pick a left eye and a routine to pick a right eye, rather than picking an eye and mirroring it. If you look back at all of the pictures with people, you'll see that the eyes, when facing you like this, are completely different from each other. This can lead to some rather freakish looking faces.

 

This was one of the original generations for the girl-holding-a-katana prompt.

What?
What?
I don't want to even think about what connections were made in the mind of an AI to spit out this.
Oh, and note the eyes, again.

 

"A creepy misty forest at night with a powerful light coming from behind the trees."

Spookiness in the forest
Spookiness in the forest.
Nailed it.

 

"An angry unicorn with blood dripping from it's mouth."

Dall-E refused, citing the terms and conditions. So it's perfectly okay with flying cheeseburgers fitted with lasers attacking a city, but it draws the line at mythological creatures being evil and ever so slightly gory.
It probably baulked on the word "blood". I probably should have said "ketchup" or "tomato soup" instead.

 

"Unicorn, Rembrandt style."

Evil unicorn
A portrait of an evil unicorn.
Cheers, Dall-E. This is the stuff nightmares are made of. ☺

 

"Anthropomorphic chess pieces in a state of panic."

Argh!
Argh!
Yup. Pretty much sums up the situation between Carlsen and Niemann, doesn't it?

 

"A watercolour painting of a Citroen 2CV on the moon."

The 2CV can truly go anywhere.
The 2CV can truly go anywhere.

 

"A top down picture of a woman taking a swan dive off the top of a high dam."

Ouch!
Ouch!
Aw, man, brutal.

 

"A happy Goth girl rocking out on stage with an electric guitar."

She's doing melody
She's doing melody.

And...

She's doing the metal chug.
She's doing the metal chug.
🤘!

Interesting what Dall-E thinks "Goth girl" looks like.

 

Okay, I think that's where we ought to leave it for now. I'm thinking of dropping $15 to buy a bunch of credits to play more, as the free 15-a-month doesn't allow for much experimentation. Especially when there's no end of weird things one can come up with to challenge the AI artist.

 

Now, a new method of creating art has arrived, so as expected a bunch of stuffy people have crawled out of the woodwork to moan about how this sort of technology will destroy art, blah blah bloody blah.

Certainly, this will have an influence on art. But people still create paintings and people pay money for paintings, even now in an era where any child can have a smartphone that takes pictures with more clarity and accuracy than most paintings. Yes this hasn't wrecked painting, has it? Sure, paintings are no longer required for people's portraits (yet some still have this done as it can present a more desirable product than a photo). Photography and painting have learned to co-exist. They both have their strengths and weaknesses.
So it is with AI. This is a great tool that allows people like me - I have the artistic abilities of a blind toddler - to generate interesting pictures. I can compose images using my smartphone or the digital camera, and then fiddle with the photos to correct issues. This stuff is pokey-draggy tech these days.
Allow me to present a picture looking up along the front of the house.

Looking along the front
Looking along the front.

It took about eighty seconds to create that photo. If you look carefully you might see some of the editing as the content replacement algorithm is a bit basic, but what you're looking at isn't real.

This is real.

What it really looks like
What it really looks like.

So when it comes to AI imagery, yeah. It's new. It's entirely synthetic. And maybe it's also mashing together bits that it has observed elsewhere. But it won't be the end of art any more than photography and the widespread availability of decent cameras in the hands of the masses was the end of art.

As for the argument about people being able to create fake images and trick people? Well, that horse bolted years ago. All that's different now is the technology to do this is in the hands of regular people and not just wealthy people with vested interests. The phrase "deep-fake" has been around a while, and it probably wouldn't take that much effort to make a video of Zelenskyy giving a Nazi salute, or Putin screwing a llama. Not these days, so for all of you who are worried that this sort of technology can be used to trick people, that horse bolted years ago too.

But for now, if you want to see if an image is maybe real or maybe Dall-E, just look at the eyes...

 

 

Your comments:

Please note that while I check this page every so often, I am not able to control what users write; therefore I disclaim all liability for unpleasant and/or infringing and/or defamatory material. Undesired content will be removed as soon as it is noticed. By leaving a comment, you agree not to post material that is illegal or in bad taste, and you should be aware that the time and your IP address are both recorded, should it be necessary to find out who you are. Oh, and don't bother trying to inline HTML. I'm not that stupid! ☺ ADDING COMMENTS DOES NOT WORK IF READING TRANSLATED VERSIONS.
 
You can now follow comment additions with the comment RSS feed. This is distinct from the b.log RSS feed, so you can subscribe to one or both as you wish.

J.G.Harston, 3rd October 2022, 02:52
Yes, I tried to have an initial fiddle with CLFiler, but it crashed horribly tried to claim so much memory. 10.5M for a *FILER*??? WTaF??? I've written a filer in BASIC that is about 80K. 
J.G.Harston, 3rd October 2022, 02:59
How do you use Dall-E? I've gone to the website and can't find any way of entering anything. 
Rob, 3rd October 2022, 18:15
There's a sign-up link at https://openai.com/dall-e-2/ 
I believe it's open to all, now. Previously there was a queue and/or invitation only process.
David Pilling, 4th October 2022, 02:34
In Physics there's a thing that if you watch a film you can tell which way time flows - eggs do not appear from the recombination of a lot of broken egg. 
If we lined up OS desktops could someone put them in temporal order from how modern they looked. 
Or is it just that some look more modern than others - a bit like 1930s Art Deco looks modern, despite it being 90 years old.
Bernard, 4th October 2022, 19:22
CLFiler: with Cloverleaf Stefan’s approval I have written a description of beta version 0.79-9 which WROCC may use in their monthly PDF magazine. I’ve tried to be fair about its ambitions and its achievements so far (though I haven’t commented on its use of RAM). You are right IMO to think it’s a long way from actually supplanting Filer. It has obviously consumed programmer time which should arguably have been better directed (though I’m not saying that either in the piece for WROCC).

Add a comment (v0.11) [help?] . . . try the comment feed!
Your name
Your email (optional)
Validation Are you real? Please type 15551 backwards.
Your comment
French flagSpanish flagJapanese flag
Calendar
«   October 2022   »
MonTueWedThuFriSatSun
     
35678
12131416
182022
2426272829
      

(Felicity? Marte? Find out!)

Last 5 entries

List all b.log entries

Return to the site index

Geekery
 
Alphabetical:

Search

Search Rick's b.log!

PS: Don't try to be clever.
It's a simple substring match.

Etc...

Last read at 13:46 on 2024/11/21.

QR code


Valid HTML 4.01 Transitional
Valid CSS
Valid RSS 2.0

 

© 2022 Rick Murray
This web page is licenced for your personal, private, non-commercial use only. No automated processing by advertising systems is permitted.
RIPA notice: No consent is given for interception of page transmission.

 

Have you noticed the watermarks on pictures?
Next entry - 2022/10/04
Return to top of page