AI for Artists
AI Art Tips, Dos and Don’ts
This year witnessed a big shift as AI technology became available and accessible to everyone, inevitably impacting artists too. As with all technological leaps, resistance or denial isn’t an effective choice. Instead, artists and image makers should get a taste of machine learning, probing its pitfalls and its advantages. Artists are sometimes disruptors and one day they can propose better ways to use AI, if not at least try to keep at least some control over it.
AI, biggest upheaval to art since photography
AI technology is a game changer for any form of visual expression. The visual art category is just a small portion of images created today. And the impact is mostly on the flood of images produced. Digital artists have all the reasons to worry about their jobs, as it is already disrupting their business.
Yet traditional artists will be impacted only because there is this new machine coming into their lives. They can ignore it, and integrate it into their process, but likely the impact for them is small as their name is as important as the art they produce.
15 AI Tips for artists
AI is changing how we make art, and AI Art Generators are at the heart of this change. But how do you use them right? Let’s explore some simple tips and common mistakes to avoid when using these tools.
1. Explore AI generation tools
My 1st advice for artists for this AI new era is to use and explore the tech instead of rejecting it. Today artists already use technology like cameras, projectors, and Photoshop as an assistant.
Why not use AI as well? Rather than diminishing the value of paintings, photography inspired artists with fresh perspectives. It encouraged them to incorporate photographic methods in their art, enabling them to depict daily life with more subjectivity and closeness.
Try the AI generation tools
Start with the basics with the most accessible apps like MidJourney. Use it with your own approach and make your own idea of it. You don’t necessarily have to integrate it into your process but think about how it can influence it. Because, you want it or not, it will.
2. Make your Art about you
Copying in Art isn’t a new thing.
And the solution against it is to have a visual style strongly anchored with your life’s interests and story. The style has to be defined and be something that only you would make sense to make it. Make your art consistent with your story, your mindset, and the message you want to carry out. Being an artist is to be a strong individual in this society so you need to be someone, not just make something. Authenticity will be more and more important with AI tech development.
“Cultivate your garden”
One of the most significant differences between humans and machines is that we have real-world experiences. Turn your personal story and experience into your best marketing. Show yourself the way you are: imperfect. Find the links between your personal life and your visual style. If you reach that point and your art is scrapped by algorithms, it will have a higher chance to be identified as yours.
3. Control the AI tools
Choose the AI tools you can control most parameters. While the initial use of AI generators like Midjourney offers quick, pleasing results, it often aligns more with the tool’s style than the artist’s.
Control comes with complexity
However, methods such as Stable Diffusion, though seemingly complex, provide the artist with greater control. This shift from passenger to navigator is vital. Experiment with various AI tools, just as a traditional artist explores different brushes or mediums. Tools should serve the artist’s idea, not the other way around.
4. Focus on your originality
In the word originality, there is “Origin”. Obviously yes, but it seems we forgot the importance of it.
Data vs Information
In mathematics, there is only information if something new is added to the equation. Similarly, there is only information when a reporter gets to a place or an event and shares about it. All the others sharing the same after him are just spreading data. Information enriches the knowledge of human beings.
Original content value
This idea applies to artists as well. As the world changes, artists observe and react to it. When artists feel a deep connection to something they see and make a piece of Art that reflects that feeling, it has a particular impact. It is something that AI will find hard to do. Original content will grow in value, and artists focusing on this aspect will be more relevant to our future society.
Generated content will decrease in quality
The theory of model dementia is a perfect argument for this. It works like the Mad Cow disease, resulting from cows eating their own meat. Large Language Models LLMs will with time contribute to most of the text and images found online. The result will be “model dementia”: with time, models will increasingly scrape the content they have generated, and the quality of the generated content will gradually decrease. (Source HuggingFace)
How to put originality in your art
- Embrace the Unconventional: Originality is born from the willingness to be different. Artists can experiment with new forms, styles, and media, combining disparate elements to create something unique.
- Seek Inspiration from Unlikely Sources: Look beyond the typical sources of inspiration. An artist might find a spark of creativity in the patterns of a crumbling brick wall, the cadence of a busker’s song, or even the layout of a supermarket.
- Challenge the Status Quo: Be bold and question established norms and conventions. Artists have a long history of pushing boundaries and upsetting the status quo. AI tools will blindly follow the narratives of the country they originated from.
Embrace and Learn from Failure: The path to originality is often paved with mistakes and failures. Each misstep is an opportunity to learn and grow. See it as a stepping stone on the journey to originality.
- Incorporate Technology: Technology offers new tools and platforms for artistic expression. From digital painting to virtual reality, these new mediums provide innovative ways to create and display art, pushing the boundaries of what’s possible.
5. Join forces
Being Individualistic is an artist’s weakness
Born with romanticism, solitude has been a pillar of the creative process, a way from which artists have drawn inspiration. However, in the digital age, this behavior makes artists vulnerable to large corporations that often prioritize profit over creative integrity.
An example of this collective strength is seen in the unified resistance of digital artists using ArtStation. Their collective voice echoed through the industry, demonstrating that unity can defy the self-serving interests of tech companies. The balance of power can shift, at least a bit, when artists band together, advocating for their rights.
Find a community of artists in your region. Meet your fellow artists personally, join their talks, and make collab works. You can have this intimate social experience that enriches your art.
6. Solve blank canvas syndrome
In moments of creative stagnation, AI art tools can act as a way to kick the artist’s imagination. Think of it as an assistant suggesting ideas with text generations. AI image generators can help artists in many ways:
- Compositions: AI can provide a range of unique layouts. This helps artists explore different ways to arrange elements in their work.
- Color Studies: AI can offer diverse color palettes. Artists can try new color combinations they might not have thought of.
- Body Postures: AI tools can generate a variety of body positions. This can assist artists who need references for drawing figures.
- Patterns and Textures: AI can create endless patterns and textures. Artists can use these as inspiration for background elements or intricate details.
- Style Inspiration: AI can mimic many art styles, from classical to contemporary. Artists can explore these styles for fresh ideas.
With Lora tools, you can even train a model with your style and explore the above possibilities with it.
7. Show (or hide) your process
If you don’t use any AI generation tools: Show how much time, work, and passion go into making a piece of art by documenting the process. You are not only justifying the value of your art but also the human element. AI art doesn’t have the same process as we do, and showing ‘the making of’ will be difficult if not impossible.
In the case of digital Art, always keep your work in proprietary file formats that can’t be easily edited or copied can help protect it from unauthorized use.
Many artists are using tools that they traditionally hide so it won’t hurt the image of their art (From the camera obscura to projectors). It is the same for AI tools.
If any AI tools are not used as a part of the concept of your art, and the finalized artwork doesn’t show any trace of using them, you can omit to tell you used them.
If you use any AI-generating tools as a part of your concept or the message you want to carry in your art then document the process. (But at this moment, you will need to show a certain expertise in the tools you use.)
8. Bring a message
Art is more than just pretty pictures. And “Art for Art’s sake” only concerns a small part of the population, mostly artists. Having a message is becoming increasingly important. For example, Art activism (or artivism) stands out in this era of tension, conflicts, and crisis.
Focus on what your guts make you want to express and relate it with your life’s experience. It will add a layer of authenticity that AI generations can’t reach easily.
9. Reinforce your art’s uniqueness
Just because something’s rare doesn’t mean it is unique, it means it costs.
For example, rare earth elements are not named rare because of their small quantity but because it takes a crazy amount of energy (and consequently money) to extract them.
If you use any AI generation tools, find a way to make your Art generations rare or unique.
While AI’s vast capacity allows for endless creations, intentional limitations can enhance an artwork’s value. This limited edition system is an old strategy for prints. So it has to be more related to the use of AI tools.
How to increase the AI-generated Art uniqueness
- Produce a limited number of pieces using a specific AI model of his creation, after which the model is retired.
- Utilize algorithms that generate Art based on specific limited parameters, making each piece distinctive.
- Use a smart contract on the artwork to a blockchain, ensuring its rarity and traceability.
- Use digital watermarks and certification methods to authenticate and validate the rarity of a piece.
- Link the creation of Art to some real events, specific data, or unique interactions with the audience.
10. Look at the law, check the ethics
Artists should Verify the terms of service when using AI generation tools. Many tools or models might grant free access but have restrictions on commercial use.
Artists should Check the implications of any specific AI tool in their work, ensuring they aren’t plagiarizing or infringing on others’ rights.
11. Develop your transversality
Machine learning-based algorithms excel at recognizing patterns. Given sufficient data on a specific style, they can effectively emulate it. For instance, if an artist is known for a certain brush technique, algorithms can be trained to produce similar outcomes.
Technical expertise as an artist’s signature
Historically, artists or creators who specialized in a particular style or technique became renowned because of their mastery in that domain. Such specializations allowed them to develop a recognizable signature, which in many ways was their brand. This kind of unique branding in an age of algorithms, can represent a weakness.
Be multidisciplinary or transverse in your art. It means drawing from various knowledge, techniques, and styles. This not only enriches your work but also introduces an element of unpredictability. Algorithms would find it challenging to pin down a specific style or pattern, making replication more difficult.
A multidisciplinary approach further amplifies the human touch, as it often involves integrating personal experiences from various domains, making the content authentic
12. Don’t hide your Art
If your art isn’t seen online or you limit it to a few images, then it won’t be used for any dataset. At the same time, you won’t be able to live from it. That could be a solution, you are really well-known but the impact will be too negative for your business. Rejection never is a solution.
13. Don’t copyright your art
The copyright naturally belongs to the artist from the moment an artwork is made. Not registering it can be a mistake, just in some particular cases. Registering every artwork is a loss of time and money, and you will still have to go to court if you have a problem. And it won’t protect your art from being scrapped by bots. Joining artist communities specialized in this matter is a better choice if you want to protect your rights.
If you still want to copyright it
A painting, as any work of authorship, can be copyrighted. It gives you the exclusive rights to reproduce, distribute, and display the work, among other things. Registering a copyright with the US Copyright Office is currently $55 for an online registration and $85 for a paper registration. And this won’t even count the cost of an attorney if you need assistance.
It still needs to be clarified, and whether the AI is copying any Art has yet to be decided. You can follow up with Stable Diffusion litigation.
14. Don’t watermark your art
Adding a visible watermark to your digital images or photos of your artwork doesn’t make it more difficult for the AI to scrap it.
Algorithms can remove watermark
When the dataset is trained with watermarks, and you see them in images, they are just here as a recognition of a watermark but cannot be identified as copied. If the AI can remove a dog because it can locate it, it also can do it for watermarks. No dogs are copied or pasted; just the concept dog is rendered. Same for watermarks, it mimics the watermark concept as if it were the style of a painting.
Plus, you won’t be able to share efficiently your Art online if it has a watermark. It makes it less appealing, hurts the composition, and makes it less professional.
15. Don’t try to outsmart AIs
By trying to trump the algorithms while they use the data, you can try to protect your art but it’s time lost.
Misleading Alt text or Tags:
For example, a painting of a cat’s Alt text or Tags could describe it as a truck. So it can have less chance to be scraped for cat-related data.
Using complex composition:
LAION-5B dataset isn’t good for relating subjects to each other and using complex compositions could make it more difficult to be used. But it will evolve and for example, the Gan dataset has no problem with that already.
Using tools to fight AI
For example, Glaze uses algorithms to make small changes to art:
While these changes are almost invisible to us, they make the art look very different to other AI systems so they “might” not be able to scrape it.
Trying to focus on that is a loss of time, as the datasets are already trained, and the new ones will improve on their training methods using the experience of the previous data.