A thread by Synesthesia describing their artistic process:

"*Oct 19, 2022*

*I was talking to an artist friend, they were sharing some knowledge as I asked basic questions regarding fields I’m not well versed in. Made me think about my process…*

*My process can be seen as a time line of audio visual hardware being digitalised in this new era.*

*I start from the tools I know, analog synths, compressors, limiters, etc. played in real time, an improvised session of sound, recorded usually to tape. (reel to reel)*

*This audio, with little processing at the same time, sends midi & CV signals to a multitude of hardware. Which triggers & manipulates visuals.*

*The visuals are created by python scripts generally, libraries upon libraries of visuals, midi and cv cycling through outputs triggered by changes in timbre, frequency & anything else programmed.*

*I capture stills, of the outputs, usually run a pass for a few minutes, capture the stills that are appealing to my eye. We’re still analog, I’m using CRT monitors to capture these stills. Saving to an old disk drive.*

*I take the stills and this is where the post processing and digitalisation occurs. Using modern tools to tweak colours, composition, and create the finished piece using layers and layers of outputs.*

*Then I mint it on the blockchain. So we’re going from 60’s hardware, into 90’s displays & modules and into 20’s distribution. The life cycle of audio visual tech. Always requiring a human element. Because these machines don’t service themselves.*

*Anyway, as boring as that may be to some. Seemed interesting enough to share. 💫*

*example of the above... This is part of what was used to create Y006"*

::video[]{src="ipfs://QmUkoQvXDH26V3BZ5rqD297LRhoSMiTFEUaQUMUALmnL5c"}

*"My process is layers upon layers of audio generated abstract 💫"*&#x20;

::video[]{src="ipfs://QmY5sVGwSpnQYiCx35pbKWvtwr44tNp8CG7Xhv2hVJcYk7"}

What follows are my own ramblings. Tl;DR it is hard to define what AI art is, Syn's art is not AI art, and the process raises some interesting (to me at least) points. Read on if you care to.

I found this fascinating for a couple of reasons. Beyond the fact that I am a fan of Syn's art the reason this piqued my interested is because at the time there were various discussions going on about whether AI generated art is art. That is an impossible question to answer unless you can also answer "What is art?". Greater minds than mine have attempted to answer that question in its entirety and I will not try here. But I would like to highlight a few elements that were brought into stark relief for me by looking at the process above.

To me the above describes a perfect symbiosis where human and machine cannot create the output without each other. The art is created by original synthesis using a mixture of hardware and software. This is interesting because to me it makes it clear that AI art is beyond the boundary of what Syn describes here - the machines in Syn's process are definitely tools in the hands of the artist - just like a paintbrush is a tool, but they create a signal that stands on its own.

To better understand what AI art actually is, then, I tried to [understand exactly how Dalle ](<https://www.assemblyai.com/blog/how-dall-e-2-actually-works/ >)creates images.&#x20;

I got the first bit - how the machine ingests and stores images, and how it decides what is semantically close to the text describing the required output. And I understood the steps in the process for generating images that did not exist yet but I did struggle with understanding the magic of diffusion. I intend to revisit that as well as the concept of [autoencoders](https://www.assemblyai.com/blog/introduction-to-variational-autoencoders-using-keras/) later.

The second reason I found reading about the Syn's process interesting is that, in a round about way, I came to the question of "Does it matter how much an artist knows?" because I always believed the ability to imagine was one of the factors distinguishing humans from machines. A human's imagination, however, must be bounded by knowledge, just like a machine's. Could a neolithic person imagine an airplane carrying passengers and then paint it? I don't believe so but  most humans could paint one now, however badly, so clearly knowledge of a thing, and even the ability to imagine doesn't make an artist. Now the machine's knowledge far exceeds any human's and it can perform an approximation of "imagination" by generating new material so where does that leave us? Clearly the ability to use the tools well, be they paintbrush and paint or P5.js are important to define how good an artist is but is it one of the factors in determining whether someone, or something, is na artist?.

Thinking deeper about Syn's process brought me back to a salient point - the digital world is quantifiably finite. The machine's knowledge is already inconceivably large for our human brains but finite neverthless. The analog, however, is infinite. A line 1 cm long has just as many points on it as a line 10 km long - infinite. Syn's selection of which still to use from the analog signal was one of potentially infinite choices but once the choice was made and digitized, it was made finite. I think I am correct in saying there is a word for this - quantization.

An artist's knowledge, then, doesn't matter. What does matter is their ability to use the tools well and the artistic choices they make from the infinite possibilities available to them. You might argue that machines making a stochastic choice equates to them same thing as an artist making a choice. I don't know, I'm still trying to figure it out.

You can see more of Syn's art [here](https://objkt.com/@synesthesia). Maybe one day Syn will be back - for now we have their art and know how we came to have it.


A brief description of Synesthesia's process as described by themselves

Synesthesia's process

Written by

Editions

Share

Details

stay ahead with our newsletter