A year ago, the open-source text-to-image generation tool struggled to quickly produce high-resolution images locally. However, with the advancement of AI, I tried it again today and found that it has become fully mature. With just a few adjustments, even someone like me, who has no drawing skills, can bring to life the images imagined in my mind!
These images were generated using the open-source software Fooocus. For the initial use, you need to download the pre-trained models from huggingface.
To generate these images, follow these steps:
- Generate some beautiful images using the following prompt: “fantasy, dream girl walking in the cave, stones, masterpiece, ultra detail, 8k, RAW photo, best quality.”
- Use the Pyracanny tool in Fooocus’s Image prompt to change the style of the images generated earlier.
- Choose your favorite image and increase its resolution by doubling it.
In summary: Use Image Prompt → PyraCanny → (Add prompt: “tranquil night”) → Check Futuristic Cybernetic and Futuristic Biomechnical Cyberpunk on the right to generate images with a futuristic tech vibe.
This is setting screenshot:
These haven’t even gone through repeated selections; they were all generated at once.
As someone who hasn’t learned to paint much, in my opinion, what AI is drawing now is already quite impressive. Although you can still find many details that are not perfect upon closer inspection, I believe that if this trend continues, AI drawing will undoubtedly impact and change various industries such as fine arts, gaming, animation, and more, bringing significant changes to the profession of artists. Now, let me share my analysis.
Firstly, let’s set a time and scope for my analysis, tentatively within five years and up to ten years in the future. Predicting too far into the future is akin to fortune-telling, and I can’t provide meaningful insights.
In the future, even if AI drawing becomes part of workflows in industries like gaming and animation, it will likely be used by artists or other groups with an appreciation for fine arts. Only they can discern which drawings are better and make subtle modifications to correct any inconsistencies in AI-generated art. A point worth discussing is, in five years, how much will AI drawing reduce the time it takes for an artist to create a painting? Will it be 90%, 70%, 30%, or 10% of the original time? If the time to create a painting is reduced to 10% of the original, I believe AI drawing will bring about a significant transformation in the industry. Moreover, once the structure and art style elements are determined, AI’s speed in drawing similar images may far surpass that of human artists, possibly reducing the time to create a painting to 1% of the original time. I tentatively set this time ratio for five years ahead at 10%.
Based on my answer, can we infer that 90% of artists will be unemployed in five years? The answer is not that simple. In five years, many mature AI drawing applications will be introduced, which will undoubtedly have a significant impact on the personal commission market. Moreover, more people may commission artists not to create an entire painting but to modify images generated by AI.
However, from a broader market perspective, i.e., commercial commissions from various companies and artists employed by these companies, these artists will not necessarily become unemployed. The leaders of these company projects, no matter how foolish, will not revolutionize their already mature working systems. The challenge will come from external factors when new project teams of large companies or newly established small studios adopt AI drawing, achieving things that were previously possible only at higher costs and with significantly better results. In such cases, the market value of the former will undoubtedly decline continuously.
Let’s take the example of the anime tower defense mobile game I play, Arknights. Suppose another anime tower defense game suddenly appears on the market. It might have a slightly worse storyline and gameplay than Arknights, but the latter utilizes AI drawing. In Arknights, you might spend forty to fifty dollars on a skin, while in the other game, you could spend the same amount on a set of skins for a character, ranging from battle outfits to casual wear, and possibly even sleepwear. Arknights might reuse several backgrounds and character illustrations in one event, whereas the other game might introduce dozens of backgrounds in a single event. Character illustrations in the latter may change styles with different backgrounds, and there could be a plethora of facial expressions. In such a scenario, even if the latter falls slightly short in storyline and gameplay compared to Arknights, I believe most people would still choose to play the latter.
If we apply this example to animation, let’s say Animation A has a well-developed plot and settings, but Animation B utilizes AI drawing, reducing animation production costs. Animation A might often resort to ‘limited animation,’ while Animation B is more likely to ‘animate whenever possible.’ Compared to Animation B, Animation A may seem quite ‘limited.’ In this case, the audience is more likely to prefer watching Animation B.
So, will artists lose their jobs in the future? I think it’s possible, but not widespread. If, with the help of AI, the time it takes for an artist to create a painting is reduced to 10% of the original, but the overall demand for paintings in the market increases six or sevenfold, some artists might indeed ‘lose their jobs.’ However, the overall change in the artist community is not substantial.
In this scenario of significant AI drawing development, artists will need to learn new skills to enhance themselves, or they might be eliminated by the market. But there’s another possibility: the current state of AI drawing has already reached its peak, and in five years, there won’t be much difference from the present. In such a case, the artist community remains safe, and there’s no need to learn new skills.