I have continued to play with the Dall E 3 image creator, and I think I have been able to lose the anatomical style. Firstly the casings have all but gone and now we are left with just the images.
I did seem to notice a difference when I used the generator through Bing, and when I used it through ChatGpt. If I give CHatGPT an instruction, it seems to adapt it slightly from image to image. At least, the description of what it has done would appear to signify this. So, when I found some images I liked, it was harder to keep the generator creating images in that style.
Overall, I prefer these images. I think they better represent what I want from a pro-life image, that is, something beautiful that hopefully reminds people of what a wonderful and incredible state this is, to carry and then give birth to new life.
There seems to be a loss of antiquity, as the new images feel more modern, but perhaps this will connect better as the audience is, of course, modern themselves.
I have started a new Instagram account for this body of work: @almosglass. At the moment, it seems that this account is restricted for hosting pro-life content. But I could be wrong.
Okay, so once again I have underestimated the sheer power of AI. If we can ignore its artistic merits for a second, and instead stand back in awe at an AI process that can not only replicate the collodion alternative process, but can also add on further materials, whilst also computing a creatively very difficult topic. It’s absolutely fascinating to look at. I have been absolutely bowled over by the results.
I’m very interested to know where we go from here with AI. I would guess it needs to fine grain the actual creative process. At the moment, I am using Dall E 3 within ChatGPT, and I’m able to refine slightly from image to image by prompting the AI to alter the prompt itself. From here on in, I’m too scared to lose the theme/style of what I am creating, and so I revert to saying “and again” a lot. I have previously lost a very nice style by tinkering too much with the prompt. So I would like to see some more controls over this. Of course, you should just be able to go back and take the exact prompt, but the same prompt rendered on different machines seems to give different results. (I will double check this again, as it doesn’t seem right). In addition to this, as mentioned in the previous post, I would like to be able to vary the style if the generator becomes “stuck in a rut”.
Now onto the images. They are concerning pregnancy, and the pro-life movement. I wanted to see how the AI would deal with the topic using the collodion style. What seems to have been arrived at is a very scientific/anatomical style. Almost a machine making art about a biological process that it cannot quite understand. Some of the images are beautiful in a way, but in their encasings, they seem trapped, and certainly, frozen in time. Perhaps we can see the images as a warning, not in fact to dabble with God’s biological process by amalgamating with metal, wires, and chips. The beauty in them is certainly that of the biological.
I will continue with this new project for a while. I want to see if I can get it to produce art worthy of the pro-life cause. For now, it remains in an interesting, if slightly anatomical state.
Since Dall E 3 has been up for a while now, I decided to delve back into it again.
The first noticeable change was that the images seemed to be higher in quality from a strictly technical perspective. But issues remain.
Firstly, it seems difficult to get the AI engine, at least through Bing, to alter its subject style. Often I was presented with some rather trashy depictions of women, in highly suggestive poses, which is obviously wrong for the subject matter of pregnancy. Even when I did manage to change the suggestive posing, there was still the issue that the subject looked very consistently the same. I feel that perhaps there was more variety in previous versions of Dall E. However, I probably shouldn’t complain too much, as I think a fault I complained of before was that it was difficult to get the engine to maintain a style.
Also, it still has a composition problem for 95% of the images in question. The best way to describe them would be boring, bland, etc. Perhaps this is the fault of the prompter.
So, ways to improve… It would be good if there were some advanced settings provided, where you could manipulate sliders, and toggle on/off settings. For instance, I would like a “vary subject style”, so that I am not stuck looking at minor variations of the same subject every image. You could also have a randomness slider, so that if you wanted to completely change it up, you could. In a way, this could be a style search mechanism. I’m sure you can think of many others.
The main image I have attached in this post was created through Dall E 3. It is a pro-life image, where the aim is to celebrate pregnancy.
I’ve decided to write a further update on the Dall E image creator. I have attained my favourite image so far ( see top), but at great difficulty. The sheer amount of images that I have had to trawl through to get this image…well let’s just say it was a lot. The problem lies in a number of areas.
Firstly the software does not seem to produce great composition on a regular basis. Composition, of course, is integral to an image capturing the eye.
Secondly, the formation of hands, legs, and eyes seems to be a problem. Eye’s in particular can be very frustrating, as the rest of the image can be perfect. Also, just the general formation of the body with relation to
bodily positions specified can be a problem. Images with malformations can be very distracting. You can attempt to fix this by using the edit function, but I have found this to be currently poor with relation to realistic faces with eyes that need to be fixed.
Thirdly, the randomness factor. Each variable you supply to the algorithm seems to have a randomness factor associated with it, so that there is often deviation from what you have specified. When you have lots of elements or variables, then attaining what you intended can prove to be very difficult indeed. Perhaps the algorithm creators could allow the user to specify the randomness factor for each image. Of course,it could also be a text processing issue. For instance, I suspect that certain words or descriptions associate with specific styles, so that even if the style isn’t specified,that’s what you get. Also, certain words or descriptions seem to be able to clash, in the sense that the algorithm has difficulty fulfilling them all for one image. Also, sometimes your prompt will produce a style that you really like, but then you are unable to reproduce this style in further iterations. This can be very frustrating, and I would suggest that perhaps having some metadata associated with each image generation could be useful for image creators. Of course, this could prove troublesome for the AI image generator companies if they are using specific people’s styles in order to generate certain images. However, we have been told that this is not how the algorithm works unless specified. Personally speaking, I no longer specify a specific person’s style, as I don’t want to affect an image’s usability in the future.
All in all, I’m enjoying using image processors again. The “sickly” feeling I was experiencing appears to have disappeared, although I still worry about the speed at which this may replace normal artists. I say “may”, as I’m not sure whether AI is capable of creating something new yet, a new style for instance, or even how this would relate to the prompt. There would need to be a much finer grain on the prompt, perhaps pages and pages describing each minute detail of the image, and probably a series of iterations over the same image, as the artist altered slightly from image to image. So that, I think, is still someway off.
This leads to another problem. Although you can create a lot with the current prompt description, the character limit is now being reached. For the above image, the character limit was more or less met, and I could really do with more characters to be even more specific.
You can find this print and more available at the below link:
I’ve opened an Etsy shop for my AI Pregnancy Vases work.
It feels strange opening a shop for AI, since AI has such a heavy hand in the final image. But the prompt is important in evoking a style to your images, and then there is the selection process… Many, many, many images have to be generated in order to obtain an image that is a) bodily well formed; b) aesthetically pleasing; c) not overly simple; and d) carries the theme of your prompt well.
For this project, I have been trying to communicate the beauty in both the pregnant state, and also the beauty of the unborn child. I felt the vase and flower was apt as a representation of this beauty.
With thousands of iterations, I have been able to settle upon maybe 20 images so far. A few of these I have put up as Limited Editions of 1 (apart from an Artist Proof), where the buyer and myself will be the only ones in the world to own the print. I feel that this business model is the way forward for prints in the AI age, since each print generated is unique, but the vast majority of iterations are unusable, or severely lacking in artistic merit. With this model, it is like Pokémon, with people seeking the “rares”, except in this case, the rares are decided by the one who generates. So, great images are far from being a foregone conclusion.
Another part of the process that feels strange for me is letting go of my photography ideas. Primarily, I wanted my art going forwards to be pro-life photographic images, but my situation at home means this really isn’t possible in the short term. I could be waiting years, and the new thrust in the pro-life movement is happening now. The dehumanisation of the most fragile among us, the unborn, is very real, and the Roe Vs Wade overturning gives an opportunity to persuade people that the past 50 years has been a mistake, with 63,000,000 in the United States alone having been killed. This is an immense atrocity, and very similar to genocide. Of course, these words are considered extreme in today’s climate, but this is only because, due to its legality for so long, people have ceased to question abortion morality, and *maybe* there is an element of double-think at play, since I feel that deep, deep down, a lot of people must know it is wrong. With this in mind, I hope that I can be one of the many who fights the prevailing rhetoric. Doing this through art is just a small measure of course, but I am limited in time thus far. I hope to be able to do more in future.
Despite this project being pro-life, I haven’t tagged this in my Etsy store… Why is this? Well, I initially set up a few months back uploading a few pro-life cards, with pro-life explicitly stated on the photos. However, I felt like I was profiteering of a movement that shouldn’t really be profited off, so I pulled the listings down very quickly. I’m still unsure of whether this was the right thing to do… With this latest Pregnancy Vases project, I’m hoping to at least place the seed that life is indeed a beautiful truth to be treasured and nurtured, when many are arguing the opposite. If people want to read into a little more of what I stand for, then they can.
I have been experiencing some issues on the social media and sales platforms that I use. Is it that I am a victim of “the algorithm”, or is it more likely that pro-life content is deemed unacceptable for public consumption?
My latest bugbear is with instagram. I had the audacity to start up the account @prolifeai. On my fourth upload, my account status read that I had been limited due to sensitive content. This was the image in question.
xr:d:DAFgXF6uDEg:3,j:45331930398,t:23041712
And here is the text in question:
Pregnancy Vase
I like the idea of abstracting the female form and symbolising pregnancy. The vase can also be seen as the uterus, and the flower, the freshly conceived baby at the zygote stage.
Was this the result of an over-zealous pro-abortionist reporting my content, or is there something more at play? I wonder to myself why there is a dearth of pro-life art online? Are others having similar troubles?
My previous post on @prolifai garnered over 60 hashtag views. So what is really going on here? Instagram later removed the sensitive content warning, but my account remained “restricted”. Had I received the dreaded “shadowban”? Only followers were able to see posts, as evidenced by a later post.
I also started another account, @almoslataanai, specifically for my AI work, which I will probably tag @prolifeai with( since that is a community account). However, I found that I had zero hashtag reach on this account. I found that I had to run a story advertising my new account on @almoslataan to get any reach whatsoever. So I am assuming this account is in “shadowban” too. Is this again to do with pro-life?
I would be interested to hear if others are experiencing issues with pro-life on social media. Are you in receipt of a “shadowban”? Is this a wider issue, or have I just been unfortunate? I started a Reddit discussion on the topic here. The basic gist seems to be that others are also the same with regards to their pro-life content.
It may just be that Instagram are reviewing my case, and will enable my accounts to be seen by the wider public soon. But for now, it is possible that they are in receipt of the “shadowban”.
Okay, so I think I was underestimating AI in my original posts. With some more detailed prompting, I have been able to come up with some further pro-life images that are celebrating pregnancy (although perhaps a celebration on the moodier side of things).
The new images have surprised me. I was under the impression before this that, at least on the traditional artistic side, artists would be safe for a little while longer. However, I no longer feel that this is the case. It seems that you really can fine grain the algorithms into producing some very detailed and beautiful work. Complexity, at least with Dalle, seems limited to the amount of lines you are able to supply. But with 6 or 7 lines, we really are able to vastly affect the visual landscape, and make the images a little more our own.
I imagine that with enough exposure to these generators, you might begin to notice some similarities between disparate images that might indicate to the viewer that they were seeing AI, but perhaps this will change. However, it seems hard at this point to envisage how you could specify the algorithm to draw in a unique style that you could keep. For instance, how would you describe to a computer how to draw with John Byrne’s style without specifying John Byrne? How should one draw a chin? What about the eyes? What exactly is it that defines a style? Really, we are looking for a brain interface that can map our thoughts to the viewing device, and this may not be as far away as everyone thinks. But of course, if we do get to that stage, then we will have a whole host of new problems that perhaps go beyond just AI.
With regards to pro-life, I hope that the images may promote the pregnant state, despite their moody nature. A woman, of course, is much more than just a vase, but that both vase and woman can both be and contain something truly beautiful is what I am attempting to convey in an age where the unborn are being dramatically dehumanised.
I have started a new Instagram account to promote my personal AI work. Tag is: @almoslataanai
I was asked to help on a small piece of AI image generation work, and so became drawn into the AI world again.
My opinions haven’t changed – I still find it unsettling on two fronts: that the prompting human only has a minimal impact on the creative process; and the ease and speed with which it’ll destroy some industries.
Despite this, AI image generators are not going away. They will only improve, and perhaps will grow to allow artistic thinkers to have a finer grained control in the end product.
So, from a pro-life perspective, we need to consider the time it takes to create art manually, versus the speed at which changes are happening in the movement. From my own perspective, I have had a photography idea for about 4 months, but important commitments prevent me from setting up a studio, and finding a model for such a venture is difficult in the current climate (if anyone has any suggestions then I’d love to hear them).
With this in mind, I have decided to use AI to flesh out some ideas and upload. Perhaps others can make something of the ideas before I can.
So, for this particular image, The Golden Flower (see top), I was interested in the idea of the golden ratio being also referred to as the Divine Proportion, and its prevalence throughout nature and the universe. In some sense, I like to hope that in each child conceived and given to us by God, there is the potential for perfection, despite what may appear outwardly, and that for this reason and many others, we should choose to treasure every life conceived.
I have also set up an Instagram community account for AI generated pro-life images: @prolifeai . Here you can use the tag #prolifeai if you wish to add any images to the community.
In an update to my previous post, I have abandoned AI image generators for the time being. The randomness factor is just too high to be able to realistically grant any authorship to the prompter. It isn’t recreating the image you have in your head. In fact, it was unable to manage this in any of the experiments I made. At best, it was able to give me a couple of ideas for future images. But even this I don’t like. Will an AI image now be the source for much of the art that we come into contact with? Pandora’s box has truly been opened.
It’s not that artists have never borrowed, but at least they were borrowing off fellow humans, meaning the final image in question was distinctly human. AI is trained on humans, but its final output is random, if random can truly exist in a computer.
To be completely honest, there was an initial period where I waxed lyrical about AI image generators capabilities, but now this has been replaced by a sickly feeling. I’m not sure if this is partly due to the massive backlash from actual artists using a physical medium, or whether it is just from using it more extensively, and thus seeing the artifacts/errors that paint the image as distinctly AI. Either way, I have seen enough to know that I will be abandoning it for the foreseeable future, and will instead look to continue with photography (which, although similarly lambasted during its infancy, is still capable of representing a more accurate artist’s vision from mind to medium).
It will be slow, but I hope to make more pro-life images in the future, within the photography medium, and with other human artists as my inspiration.
I thought I’d write a post on some experimentations I’ve made with Open AI’s DALL E2. For those that don’t know, it’s an AI image generator. It works by taking some textual prompts from the user, before fashioning what seems to be a unique image every time. Not only can you specify what you want it to draw, but you can also specify the style you want it to be drawn in. Now, it seems to have some degree of poetic (or not so poetic) license in how it interprets your prompts. No matter how specific you make it, it will still deviate significantly at times from your instructions. The result of this, though, is some large variation in output.
So, I started first by looking to generate pro life images by generating images of women and the zygote. I won’t go into specifics on the prompts I used (due to another project I’m working on), but I got quite specific. Here is the first keeper that was generated.
I was surprised by the quality of this. Compositionally, it is not perfect, as it almost seems to be cut from a larger picture. Perhaps the software isn’t aware of the conventions for which features should be in the frame. But it seems to be of reasonable quality, and is certainly better than anything I can draw.
An even more interesting image is included below. As we can see, it isn’t a very detailed image of a woman, but it is an image that an artist could perhaps choose to take forward further themselves.
I then moved on to images of the baby. This generated a number of interesting, if not wholly bodily accurate images. I’ve included one below.
I would classify the images as interesting, although I don’t really connect with them. Perhaps connection is related to intention and workload.
Interestingly, the Catholic Pontifical for Life is currently conducting a review into “emerging technologies”, and Pope Francis has said that we must, “ensure that scientific and technological growth is reconciled more and more with a parallel development … in responsibility, values, and conscience.” When looking at AI image generators, we must consider the potential, as with other AI’s, for job displacement. At the moment, with the low resolution images these AI’s are outputting, the artist’s job is safe. But what about the future? Certainly, there will presumably always be a market for physical art created by human hands. But for those specialising in digital art, there may be problems on the horizon in terms of competition, as the creation of art passes into the hands of the masses, as digital cameras, and particularly phones, did for photography. The result will be a more diverse art space, but perhaps a less lucrative one.
However, we must also consider the value that a non-generated piece of art possesses, namely that of intention. There is no outpouring of soul into an AI generated image. Even with the most specific of prompts, the pictures generated are those seen by the computer, not the human. And so physical art, and perhaps art which travels on a medium, such as film photography, will keep a degree of value as being that created by a human (although there is no way to know whether an AI image has been used as a starting point for the picture). Certainly, when I look at AI generated images, it’s hard to find a connection, knowing that I am not peering into a human’s soul, despite it being an art piece. They are more like Frankenstein images.
With this in mind, it’s hard not to wish that AI image generators are banned. What we seem to place value on is human imagination, but with the advent of this technology, we can never be sure whether the image idea execution came from a human, or from a machine. It is important to note, though, that humans have often borrowed or reimagined art ideas from other humans in the past, which means that intentionality is never truly what we imagine it to be. It’s just that now, humans can borrow from another source too when creating their physical art.
A further problem comes from the AI generator in that you can ask it to copy a style, or produce variations on an image input. This is perhaps more problematic for the artist, as a style is like a signature for the artist. Often, it defines who they are. Problems arising from this could include “style dilution”, where the artist’s style becomes saturated into the market before they have reaped any kind of financial reward from it. One way to combat this would be to keep art secret until a body of work is complete, but this also diminishes the artist’s earning potential. In a market where any style can be replicated, how do you keep the market hungry for originals?
So, can this image generator be used to benefit the prolife movement? I think yes. There is relatively little pro-life art, as far as I am aware, and part of the difficulty is in the difficulty of creating art in general. Unfortunately, people are not drawn to highly divisive topics for fear of backlash. In this sense, AI provides an avenue for those that perhaps do not have the time or skill to create art in the traditional sense. Problems, however, are the disconnect people may have with the art generated. Is it going to be able to move people to changing their minds? Is the software good enough to even depict human emotion accurately? On the latter point, I would say no. The software seems to have difficulty depicting highly realistic human emotions. It also seems to have a propensity for inaccurate bodily depictions, particularly the face, which may conjure negative perspectives for people.
Still, others may have better luck than me in dealing with this software, so I have created a Facebook Group for AI Pro-life images, if anyone is interested in contributing. It’s called ProLife AI
