AI Generated Art — New Renaissance?
Designer first impressions
Recently, I started playing first with DallE2 and then with Midjoruney. The opportunities they created looked fascinating to me. But since I am not a designer, I wondered what people who are experts and have many years in the field think about this new development of AI opportunities. How will it influence their future, and change their roles? Are they afraid or excited?
So I asked a few questions and summarized the answers from 3 designers with different profiles and backgrounds. As you will see, they have various expectations and comments about this topic, making it more interesting.
Q1. How many designers do you think have become aware of developments with AI tools that help with visualizations? How aware were you of the trend until recently?
D1. Designers who follow the trends, who are up to date, know that AI has been present in design for some time. Let’s say Photoshop uses AI Sky Replacement to replace the sky and background, and it does a pretty good job. I have personally been following AI and related topics for many years, so Dalle2 and Midjourney did not surprise me.j Google had a number of pilot projects, for example, AI Pictionary, which collected user data and learned to recognize shapes and images on a large number of samples.
D2. There is a lot of talk about AI art in general, and this trend is accelerating — something that a year ago was an avant-garde, an attempt, or an experiment, is now galloping into the future of graphics and design. To the extent that many designers have become skeptical, or afraid, of how this development will affect digital service markets. That is, what is the future of art (especially for digital artists and illustrators).
The whole story came to me through collaboration with other illustrators. Some have used AI for backgrounds, for concepts, refined the results of the AI, and created original illustrations that look awesome and take a tenth of the time to produce.
The advantages are obvious — especially in concept art, where you can produce endless visualizations and concepts, and their iterations, in order to quickly deliver effective graphics to the client. If AI is understood as a tool, it can be very well integrated into production, and this again depends on how and in what way the market will be positioned accordingly. The trend is that AI will be a kind of revolution if it isn’t already.
D3. I’m not sure how many designers are aware of the progress of technology, but I believe that with the arrival of such tools, the profession’s interest in technology will increase. I remember a lecture at FER a couple of years ago, there was an illustration of the landscape that showed the relationship between competencies and the risk of AI taking over these jobs. Competencies in the lake would be those that were already (in 2017) somehow replaceable by artificial intelligence (chess, Go, computer vision…), and on top of the mountain, they would be the most difficult to replace. Science, writing books, and programming is at the very top of the competencies that are the most difficult for AI to master, but art was quite high. When you look at today, it seems to me that we are still not sure how fast and what competencies AI can take over.
Q2. When you see some of the works, what are your first impressions? What is the quality, and what are the advantages and disadvantages?
D1. Overall, I think it’s pretty impressive what AI can currently produce based on a textual description. The advantages are Quality in a very short time, and the disadvantages are the impossibility of precise control (access to layers, resolution, vector format, etc.) of the final photo.
D2. It depends on what is combined and for what purpose. Fantasy genres, sci-fi genres, realistic rendering, and similar categories look phenomenal. However, some other branches, such as some kind of comic styles, and extremely stylized graphics, are not yet at the level where AI could release a finished product, which does not mean that they won’t be soon. However, it is also an incredibly adaptable tool for ideas and concepts.
D3. I haven’t played with these tools too much, but what I did with DALL E surprised me but didn’t make me use up all my available credit. I think the result depends on the quality of the query — if you enter something too general, you will get a similar result to the one you would get on Shutterstock or some similar marketplace. On the other hand, if the query is too complex, i.e. you know exactly what you want, it is less likely that you will get exactly that. I think the ‘wow effect’ is greater outside the design community than among designers because it allows the former to create something cool without prior knowledge of existing tools, which designers knew before. I would say that tools like DALL E or Midjourney are just an upgrade to existing stock photo services in the sense that the search is smarter and the result is closer to expectations. But I don’t think AI will replace designers. Stock services offer a lot of high-quality illustrations, web templates, and videos — so there are still designers who do the same work much better.
A bit of philosophy — the purpose of design is not the end result but the process by which it is reached. If we kill the process, we have killed the value of the result.
Q3. For some images, the classical design process would require hours and knowledge of different programs, how much DallE2 or MidJourney can help shorten the designers’ process or at least their work in the prototyping phase?
D1. I think they will serve designers only as another tool in the process, not as a complete replacement. Design is a complex process, and in that process, each step must have the possibility of subsequent intervention, which AI generators do not provide for now. I can certainly help with moodboards, stylescapes, even (I hope soon) with storyboards.
D2. I can save hours and hours of work, with equal if not better results. Not to mention the freedom to experiment — more time to try out some combinations that you wouldn’t normally play with — because there simply wasn’t time for them.
D3. They can probably help speed up the process like stock services did with photography. In terms of knowing the tools, I would compare it to web development — for a long time there have been various ‘plug and play’ tools that allow the average user to create a website for pennies, so we still do custom websites that usually take months of work. It could be said that AI tools have opened a new “low-budget” segment on the market where with a small budget it is possible to get quite a good result, but professionals will still turn to someone who better understands their project, values and the result they want to achieve.
Q4. My theory is that now designers and experts with broader basic academic knowledge and the ability to conceptualize some visuals and then successfully verbalize them will come to the fore. What do you think are the advantages of educated designers here compared to all the other laymen who stick to words and learn from trial and error and imitate others?
D1. This is a tool, anyone educated will use it to save a lot of energy and time. Here’s an example, let’s imagine artists who paint interiors. They (provided they know what they want) first have to draw a sketch and color it in order to offer it to the employer, and that requires time and energy. With these tools, they will type in their idea, and in a few seconds they will get a finished image that they will later refine and they are ready to work. In addition, it is easier to type your thoughts into an AI generator and see the proposed solutions in a few minutes, than to reproduce each of your thoughts from scratch.
D2. I do not think that formal education will play a big role in the application of AI in this field, but I think that there is a real advantage in some purely technical aspects of the job. However, creativity and ideas, with the application of AI, can very quickly compensate for formal education. Especially since the AI also uses some good design practices of masters of design, illustration, and painting as references… If it is given the right references, the result should be based on them. It still doesn’t work perfectly, but in the field of abstraction, it’s very clear where this development is going.
D3. Designers will probably do better queries than laymen, but it is interesting how some have already used this opportunity to develop a new product around this, so there are services that, for example, sell queries for AI models for a certain group of graphic materials in a certain style (e.g. Promptbase).
Q5. Good examples of the combination of new AI tools and old ones are already visible, for upgrading visuals/images from a certain level to the final one. Will AI be able to take over the entire process with further development?
D1. I believe that AI will become superior in a technical sense very soon because it can already turn a human sketch on paper into a photorealistic image. The entire process in the industry requires a lot of precise changes, additions, and micro-adjustments, which AI cannot do for now.
D2. It’s hard to predict how algorithms will work in the future. It looks like it will again depend on the wishes of the clients or the market. For now, each graphic needs a little polishing, or matching, but how does one work with it?
I think that the human hand will always be needed somewhere, no matter how little it is. But the combination of AI and knowledge of the application of other software solutions, filters, compositions, and just some theory, could lead to some high-value graphics, and some medium-value “common” graphics, if you can say that.
Value more will always be sought.
D3. I doubt they will completely take over the process, but they will certainly change it in the sense of speeding it up, and maybe take away some of the clients. But this is probably where designers will profit because the opportunity will open to work with higher quality companies — those who see value in the process.
Q6. On a philosophical level, how much will this kind of process make us all think more and look for and describe visual elements in the world around us, so that we can better synthesize and reproduce them?
D1. I think the opposite will happen, that a hyperinflation of ‘art’ will happen because anyone with 5 minutes of time will be able to produce phenomenal ‘art’. As a result, people will appreciate art and design less, because they will have the impression that ‘I can do that too’.
D2. Ideation always makes us reconsider how and what we do. It depends on the projects. I believe that the entire game industry, the game industry, the film and entertainment industry anyway forces us to constantly learn new things, try out styles, impossible combinations… I think that AI will mark the emergence of some original concepts, and that looks to me like that famous originality, or value more.
To me, Ai is just a tool that could produce some completely new results, in a world where 99% of the content is anyway duplicated, modified, copied, etc. Maybe the algorithm brings a dimension more to the game.
D3. —
It will be interesting to monitor how well their answers age with time. Is it even possible to predict developments at the current pace of speed? We already have the first projects that are generating videos.
Maybe in some, not that distant future, we will be able to get instantly created movies, generated on the parameters about actors, directors, plots, and styles that could be set by us, and delivered as a product in a minute. One can only imagine how those things will affect different industries and art itself. Maybe a question for philosophers and futurists.
Pictures in this article and many others can be found on: https://www.instagram.com/adria.ai.art/ with prompts for the pictures.