Artificial intelligence, commonly referred to as AI, has become an integral part of many creative fields. Find out how you can use it to create your designs on our platforms here.
The fact that artificial intelligence (AI) has taken the world by storm in recent years is definitely an understatement. Since the latest launch of “ChatGPT”, the technology has been a subject of discussion in many places. Due to its versatility and ongoing, rapid development, it was only a matter of time before it would be used for other purposes beyond its original use in chatbots and text-based messages – including the creation and editing of designs and art.
Programmes such as DALL-E or Midjourney, which make use of AI technology, give their users quick and easy access to an almost infinite number of unique designs that are also suitable to use as artworks. Although self-created designs often have a unique charm and a personal touch that AI-generated designs can sometimes lack, we still believe that such designs have a right to exist.
For this reason, and because we have been receiving this question by members of our partner community for some time anyway, we would like to take this opportunity to emphasise this:
If the terms of use of the respective provider permit unrestricted commercial use, you can upload designs that you have created using an AI image creation programme to our platform and publish them on all sales channels – i.e. both on the Marketplace and in a separate Spreadshop. However, there are a few important factors to consider.
On the one hand, there are the legal framework conditions, which of course also apply to designs created with the help of AI. Although the AI is responsible for the design and creation process in this case, the designs created may still infringe the law. As a platform user, you are still responsible for the content you publish with us. You can find more detailed information on the topic of “Copyright” here.
Furthermore, your designs must still be in line with our community standards.
Last but not least, there are certain qualitative requirements that we have for all designs published on our platform and which are the actual reason for this article. We have realised that there are recurring problems with designs created with the help of AI. We would like to explain these in the following section. This will hopefully spare you from experiencing a design rejection.
Content-related problems
1. Spelling and grammar
Although the relevant AI image generation programmes sometimes generate very impressive graphics, any text elements are unfortunately not yet completely perfect when it comes to spelling and grammar. This can sometimes lead to random letter mix-ups, which are not very appealing to potential customers. As spelling and grammatical errors result in the rejection of the respective designs, it is advisable to carefully check the text elements and make sure everything is correct before publication or to add them yourself in a graphics programme.
Spelling mistakes Correction
2. Content and logic
Another problem that we often encounter is something called “logic errors”. If you take a closer look at the example design, you will notice that AI does not yet seem to be very familiar with human anatomy or is at least unsure how it should depict the individual fingers of a human hand. In the case of self-drawn motifs, we would grant the artist a certain degree of artistic freedom, but in this case the result seems unintentional and is most likely due to the AI’s lack of understanding. To avoid rejection, it therefore makes sense to revise and improve the designs affected by this or similar problems before publication.
Example of “Logic error” – too many fingers
2. Design problems
1. Pixelated and/or blurred
The use of most of the relevant AI image generation programmes is associated with a certain level of costs. Although some developers offer a free version of their programme in addition to the paid version, the results are often not very attractive in terms of quality. This is primarily due to the low number of pixels with which these designs are created, which in most cases is around 1000 pixels per side. As a result, these designs sometimes appear somewhat “pixelated”. If you try to solve this problem by simply increasing the number of pixels, this often results in a certain blurriness that leads to sub-optimal results in print.
To avoid this, it is advisable to create the designs in a higher resolution from the outset – ideally with at least 1000 to a maximum of 4000 pixels per side. A sensible alternative would be to vectorise the design or convert it into a halftone screen, which is particularly suitable for semi-transparent areas, e.g. at the edges of the design.
Pixelated / blurred –> After vectorisation –> As halftone raster (bitmap graphics)
2. Remove background
As designs created using AI often don’t have the best backgrounds, you will pretty much always need to edit them in a graphics programme to separate them from their background. With careful editing, no background mess should be visible, and the edges of the design should be clean. To ensure this, a manual approach is sometimes imperative. Although both AI and graphics programmes offer automated functions that simplify the process and make it less time-consuming, the results are rarely satisfactory and should therefore be checked at the end so that any final imperfections can be corrected by hand.
Example of a messily cropped design -> The same design, cleanly cropped
Once all background mess has been removed and the designs edges have been neatly fixed, the design can be uploaded and published. Then nothing should stand in the way of approval (from a quality perspective) and your customers can enjoy optimal print results.
Now you know how to make the most of AI to create your designs on our platforms! If you still have any questions on this topic, feel free to ask them in the forum or in the comments.
The post AI and designs: dos & don’ts appeared first on The US Spreadshirt Blog.
Looking for cool gifts? Check Rott515 store!