š #065 - Overly specific design and evolution.
A solution to stabbing oneself, an early new project, riso (again), will I get truncated, and OMG WHY IS IT SO HOT?!
Iām trying an experiment this issue; which is to not worry about the Substack āPost too long for emailā thing. Each time I mention Iām wrapping up the newsletter because that message has popped up, I get some very kind messages (thank you all) letting me know that they themselves have gone over the limit before and nothing terrible has happened.
I tend to go over the limit ācause I include lots of photos (which often sends email to the spam folder anyway), and the help page saysā¦
āWhile there isn't a word count or length restriction on Substack posts, some email providers like Gmail will truncate messages exceeding 102KB
If the newsletter is truncated in an email, readers can click on āView entire messageā and they'll be able to view the entire post in their email appā
And everything Iāve ever read about newsletters says that people just donāt do that.
Mind you all those things also say to capture the readerās attention within the very first sentence, and I donāt really bother with that advice either, so perhaps nothing matters anyway.
# NOTEBOOKS, AGAIN
I made this āļø because I donāt like stabbing myself ā what it is down below in a moment.
Iāve previously talked about making notebooks back in newsletters #062 and #063, and also the question āwhat happens when you can generate images faster than you can print them, and what does that mean for the printing processā in newsletter #059 for context.
This isnāt new of course, FIELD NOTES, purveyors of the worlds best pocket notebooks occasionally make notebooks where each cover is different due to some printing process or other. But one release that stands out are/were the snowflakesā¦
ā¦99,999 notebooks all with a different generative snowflake on the cover created by artist and all round good egg Brendan Dawes, see also āPlay-Doh as Interfaceā - to pick a semi-random digital/analogue project of his. If you havenāt explored Brendanās work you really should, his website (and shop) are both a delight.
Anyone following my notes (which Iām assured by Substack are show to at least five people) will have seen these photosā¦
Where Iāve reconfigured the FALLiNGWATER code to extend what is normally a portrait design off to the left, moving some elements over, but keeping the main one (the floating builds) on the right, so when theyāre folded into a coverā¦
ā¦it looks pretty good.
I know this is going to be a series of 32 generative notebooks, but beyond that Iām not quite sure what Iām going to do with them.
One of the reasons for making them though it to give myself something to do in the evenings that isnāt doom scrolling, and making stuff with your hands is nice.
Whatās been happening though is when I sit on the sofa folding pages, or stabbing holes, or sewing the whole thing, is Iām putting the equipment down on the arm of the sofa, or the seat next to me, or OMG WHEREāS THE NEEDLE, I CANāT SEE IT, AM I SITTING ON IT? DONāT MOVE, WAIT, WFT.
Which is why I now have this desk organiser to hold my bone folder, awl, snippy-snipper-thing, thread (with a handy nodule thing to wrap the end of the thread around and then hold tight) and most importantly a curved needle holder (the needle is curved, the holder isnāt).
The temptation to cast it in cement is pretty high (or pewter, or art stone), because Iām an idiot.
# MEANWHILE, NEW CODE ART
Iāve been working on a project for the new fxhash Open-form feature, a technical-ish artist focused overview here: https://docs.fxhash.xyz/fxh/programming-open-form-genart
Itās still early days, and Iām sure Iāll write a much better overview soon. For the moment though hereās a test print (on the right), and a quick riso version (of course) on the left.
Printed onto archival paperā¦
ā¦and some riso versionsā¦
So, generally with code based generative art you write some algorithms that hopefully spit out a good result each and every time. The FALLiNGWATER project above is doing that, you run the code, you get an image.
Open-form is similar; you run the code, you get an image, but the idea is you can āevolveā that image which gives you the next āgenerationā.
Thereās actually a huge throwback here to Richard Dawkinās āThe Blind Watchmakerā - see the PDF Evolution of Evolvability from 1988 and some Pascal code drawing āBiomorphsā.
In the Biomorphs program youād start with something a bit rubbish, like a triangle (sorry triangles), and then 9 mutations would be created; by picking the one that looked most like a fly (for example) and then getting 9 mutations from that, youād slowly āevolveā wings and so on, from one generation to the next, until, magically: fly!
Thereās direct lineage to GANs and the current AI Generative Art here, fwiw; one image generation network makes some random pixels, and a separate image recognition system picks which one looks most like a fly (or the prompt itās been given). The most fly like random assortment of pixels is used as the basis for the next generation, and so on, until after 1,000s of iterations the most fly like image as decided by the image recognition system āwinsā.
Or alternatively the most shrimp like jesus picture gets posted to facebook.
š¦ š¦ š¦ š¦ š¦
Thereās a couple of differences of course in that weāre going to try and NOT start with something a bit rubbish and spend 1,000s of iterations getting to the good stuff.
Hereās a screenshot of an early version of my own code, the top row is the first iteration, which evolves downwards (so weāre seeing 7 more generations going down the screen), adjacent images with nothing above them are siblings to the ones on the same row.
Whatās shown above is a āsimpleā subdivision algorithm, but here are some basic principles Iām applying to the whole project.
First, premise:
Itās easy to make things start simple and get more complicated.
If you have āmutationsā than can introduce new features or design, theyāll tend to happen in later generations.
Itās tempting to keep ādesirableā features for later generations.
If, for example, we have a monochrome palette (which everyone loves) and that only shows up later, youāll end up with monochrome palettes only showing up on later more complicated designs.
And as it happens I like some of the earlier more simple subdivisions, so in my code, Iāveā¦
Taken on my design methodology of keep pushing things to be complex and then rip things out to simplify it again, a sort of six-steps forwards, four-steps back approach I apply to nearly all my work/projects/art/life. In this case the deeper you go into iterations, the greater the chance of the code reaching back to previous generations and grabbing simpler/larger design elements, or sometimes resetting the whole thing. So you can get earlier layouts along with the later features.
Add early but volatile āmutationsā giving access to later generation stuff, but it dies off in a couple of generations. I have some drawing styles that tend to kick in later, but thereās a chance for them to appear super early (and so again mixing with earlier features and simple design), but they donāt stick around. Unlike the more permeant version that can appear later.
Pretty much these two things mean that thereās a good amount of design space for people that want to dig deeper, but you donāt have to, itās entirely possible to access features sooner (through luck or āre-rollingā), but they donāt stick around.
The idea is to reward people for exploring deeper, while not penalising those who want to quickly dip in & out (while also not taking away from the people who explore deeper).
I swear it makes sense!
Hereās another test print where you can see the complexity of the subdivisions, but itās pulled in some of the simpler shapes from early iterations.
A lot of this is because I must confess to liking some of the very simple starting points, that mostly look like thisā¦
ā¦which turn intoā¦
Anyway, itās still early days, and thereās more I want to add (but not that much more), so Iāll throw in another gratuitous riso print version š
#šŗ š” š¦
Iām now going to try and tie the two above projects together; the design of the sewing tools holder and the code-art project, by dropping in this Bauhaus 100 documentary, in an attempt to appear cleverer than I actually am (not hard).
# THE END š„
š„ š„š„š„š„ š„š„š„ š„ š„š„š„ š„ š„š„š„š„š„ š„š„ š„š„š„ š„š„ š„ š„š„š„š„ š„š„ š„š„š„š„š„ š„ š„š„š„š„ š„š„š„š„ š„š„š„ š„ š„š„š„ š„ š„š„š„š„š„ š„š„ š„š„š„ š„š„ š„ š„š„š„š„ š„š„ š„š„š„š„š„ š„ š„š„š„š„ š„š„š„š„ š„š„š„ š„ š„š„š„ š„ š„š„š„š„š„ š„š„ š„š„š„ š„š„ š„ š„š„š„š„ š„š„ š„š„š„š„š„ š„ š„š„š„ š«
Love you all,
Dan
ā¤ļø
PS. People with exceptionally good taste in videos that are 37 minutes and 49 seconds long seemed to enjoy last weekās #Weeknotes
I am in love with your tool holder. I use a cigar box to hold my bookbinding thread and my lo-tech hack was to affix a magnet on the inside lid to keep my needle handy. But your tool docking space station is so much cooler!