A still life with Midjourney

10 min readNov 26, 2022

In this post I’m looking at my mother Anna’s creative and artistic practice up until her death, and how Midjourney, as an AI art creation tool, can complement, even facilitate, a creative practice. I’ll probably keep revising this post over time.

MidJourney

MidJourney is an an image generation tool based on an emerging family of novel machine learning techniques. The technical innovation is an idea of layered artificial neural networks and ‘latent diffusion’ — that there is an implicit structure in a dataset, and the larger the dataset the more of that structure can be modeled. These tools start with data, billions of tagged images, or significant swathes of the entire corpus of public literature and then are able to perform predictions of other similar data in that space. Deep learning has been wildly successful at predictively modeling outcomes, such as the best next moves to make in a go game, or the next phrase in a sonnet, or computer vision, machine translation, drug design and so on. Stable Diffusion, another tool in the same family has raised over $100 million.

Intriguingly when the source input is human text or art — these tools begin to parrot the latent structures of our aesthetics, values and become a probe of our humanity, our identity. Predictive text is ‘human like’ if not exactly human. Predictive images are ‘human like’ if not exactly authored by an individual human. It is fascinating to create images with MidJourney because one is accessing a substantial chunk of all artists and all creativity that has ever been recorded. It is like having the deft hands, minds, idiosyncratic ticks, opinions and life energy of thousands of artists at ones disposal. Results can both create a feeling of agency but also weirdly of a lack of satisfaction, as if the work was cheap and derivative.

Anna

My mom on the other hand was born November 25th 1942 and died on May 1st 2013 — before deep learning and before these innovations were much of a gleam in a programmers eye (although the roots are very early indeed).

To ever so briefly sum up a life — my moms story was that she’d eloped from her home in Pinerolo Italy at the age of 18 because her conservative catholic parents wouldn’t let her go to art school. She’d run away from patriarchy — a bargain many make. But she’d immediately fallen for my father who was in Paris at the time and gotten pregnant — which is why I exist. And from there ended up in Alberta and then shortly to a divorce and a succession of boyfriends and lovers all the while moving from town to town, sometimes into better situations and sometimes worse. I grew up bouncing around rural western Canada and was myself in fact perfectly happy. I do however have a hard time imagining a transition in that direction for her or indeed anybody - from Italian sweet sun warmed olive groves to endless unvarying snow swept fields of stubble for corn.

By the time I was growing up I knew her more as an artist than a mother. There was even a bit of a mania. Anna drew on everything. She drew large sketches on the walls of the homes we rented. On the doors. In folios, for magazines, for anarchist pamphlets, in ink, in oil, in pen, in pencil. The house was scattered with Heavy Metal and art reference material, and lightly sprinkled with flecks of paint — not noticed till you closely examined a chair, or a table. There was a fine forensic polkadot patina of paint flecks on many objects, clothing and herself. I didn’t see this as a form of mental illness till she was on her deathbed many years later ( possibly oil paint was a contributor to the primary CNS lymphoma wrapped around her brain-stem). In a sense she was stealing time from other things to do the thing she loved.

If she was alive today would her story have been different? What if art wasn’t something that was repressed for her — that she had to steal time for? What if she’d been celebrated for her work? What if she could simply iterate and explore at a high velocity? Would it even have meaning? What would she think of modern tools? And — mostly — given her specific love of Italian Renaissance work, and more generally her unerring accuracy in pinning down a specific influence on a piece — what would she make?.

I’m sure at the very least she’d have rich criticism (in her uniquely Italian way) and also probably would equally see it as a research assistant and would almost certainly dive deep.

The larger questions I am asking here are not just “What would my mom have used this for?” but as well “What if we all had more time to create art? What would we actually make? Would we make anything? Is time our excuse?”

Other questions

Many of us know instantly what a Gustav Klimt is — even those of us who rarely think about art. His work is a part of the language of western thought. When we see new work in his style it has an immediate valuation in our minds. His use of gold leaf, and how we value gold, has certain resonances. His willingness to include sensual imagery shifted subsequent artists language as well. When we see new work that incorporates his style we are subconsciously coloring that new work with his ideas.

Today marketing and media firms routinely composit artists and ideas together to promote product or manufacture a certain mood, or even just for humor. Boing-Boing for example used to tediously drive attention to themselves by taking a random topic and rendering it as sushi, or as lego. We often see this with other media as well; a hybrid remixed hiphop version of a song by Fleetwood Mac is palatable to both young and old.

Midjourney is remarkably good at this blurring between conceptual spaces. It can take concept X and render it in the style of Y; or mash together two different artists, or two different concepts. What intrigues me here is simply to get a feel for the space of kinds of artists, their styles and the kinds of concepts that they typically invoked. An arts major or a professor at an art college can presumably see the derivative of original work — is it possible to have that same level of understanding?

I’m also intrigued how materiality affects style. Artists have to work in mediums that have limits. Effectively they run out of time, and depending on the media they chose that media leaves different imprints on the work itself. How much of an artists style is a function of their ideas and labor being lensed through impositions of specific physical media; from papercraft to vaporwave, from marble sculpture to lego, badge buttons, enamel pins or even the size of a canvas or wall? Diego Rivera is known for large frescos, but also often chose deliberately blocky or simple almost cartoon like characterizations when more detail was possible — there is an interplay between his choices, the medium, the intended message and audience.

Midjourney takes part of what art is, part of what art is valued for, the labor itself, and reduces it away (some would argue steals it). For better or worse however the reduction of our labor brings out other values. Instead we ask if a piece is derivative, or brings any fresh inquiry, or if art itself is simply some kind of catalyst or muse for other conversations. The questions I have here are around “is art a function of virtuosity? if labor is removed from art then how is art valued?” beyond just simply “are these tools any good?”

Stick figures can tell a story. An image and a layout can tell a story, variations of the phrasing: the placement of objects, distance of the observer, pose of characters, lighting modulate that story. Often brilliant work is simply rendered; say Banksy for example. In the way that comedy can sometimes short-circuit two previously unrelated thoughts — can tools like this actually help discover and frame new ideas? Can Midjourney be like a poets aid — helping us discover new juxtapositions of images and ideas that crystalize out something fresh?

Finally although these new AI art tools produce amazing results; it can feel random. Can we have any kind of fine-grained control? Can I narrowly vary or explore changes in one way but not wildly have all the other dimensions of an image vary? For example to change the placement of elements, or the color, or texture? Can I have fine grained control? Can I ask it to make simple things without embellishment?

Starting Simple

Google Search for a glossy white cube on a white background:

A glossy white cube very lightly dusted with fine granular variations on a white background — v 4

A crude stick figure crayon drawing of kids in a boat with water and sun and a whale — v 4

Close up of steak and eggs breakfast plate, sharp focus, ultra realistic, ultra detailed, cinematic lighting, photographic, Eastman Kodak Color Negative film 5251 50T shot on panavision super ps — v 4

A tempest in a teacup

A boiling hot ocean in a coffee cup

A Pablo Nerudu Poem

Riffing on Anna’s work

Taking this original piece of art from my mom and then asking Midjourney to riff on it — I produced a couple of variations of the theme.

The second one actually is an interesting juxtaposition. A tiger is in a human landscape, and appears to be taking a human pose at a table, which I interpret as evaluating humanity for its decisions. I tried variations of “a poster with an oil painting of a tiger between two tables, hyper realistic, super detailed — v 4”.

Notably the interpretations are less “rich” than her image (fewer concepts) — although more accurately rendered. She was speaking to how the exploitation of the tiger is part of a larger problem of illegal game harvesting that includes elephant ivory and many other rare species. This paucity is of course a function of the simple prompt — it is difficult to express multi-part prompts (I suppose I could do the prompt in three or more pieces).

Here I take an original drawing by my mom focusing on 3 penguins as a simple colored ink closeup with no background. I pass this to Midjourney and then the second image is what it produces. Here — with good prompts I get better outcomes. There is a difference in the head poses and intent that I am not sure comes across. Knowing Anna her choice of the eyes and poses have meaning. I am unsure if Midjourney is conveying anything specific with the head poses it has chosen from whatever arcane library of human knowledge it has access to. The poses Midjourney chooses do seem more boring.

Art Mambo, Tom Fedro, Howie Green

Here I’m looking at compositional styles; can I use rich colorful image to guide the art that is created. I did manage to get a piece of grafitti onto a person but it wasn’t exactly what I was looking for — I wanted a less literal translation. I used variations of this prompt:

https://s.mj.run/sJpwPWGB6ss a heroic man wearing a skintight graffiti costume, no background, exquisite detail, 30-megapixel, 4k, 85-mm-lens, sharp-focus, intricately-detailed, long exposure time, f/8, ISO 100, shutter-speed 1/125, diffuse-back-lighting, award-winning photograph, facing-camera, looking-into-camera, monovisions, elle, small-catchlight, low-contrast, High-sharpness, facial-symmetry, depth-of-field, golden-hour, ultra-detailed photography — v 4

Using this image:

Narrowing the design regimes

A technical diagram of sequence space using real mountains to show fitness landscapes, insanely detailed, multiple layers, hypermaximalist, Rim Lighting, Artificial Lighting, Natural Lighting, Incandescent, Optical Fiber, Moody Lighting, Cinematic Lighting, Studio Lighting, Soft Lighting, Hard Lighting, volumetric Light, Volumetric Lighting, Volumetric, Contre-Jour, Split Lighting, Beautiful Lighting, Accent Lighting, — v 4 — @anselm (fast)