Getting Better Color From AI Art

Using Style Reference commands, specific Pantone colors can be incorporated into projects with AI tools.

At a recent presentation by Marshall Atkinson, coach at Atkinson Consulting; and Jim Raffel, CEO and color management consultant at ColorCasters, the topic was getting better color from AI. Whether that is images customers are generating with AI and providing to the shop, or using AI tools to get better color across the board, the pair shared a few tips printers can learn from.

1. Pick your tools. There are many, many AI tools on the market right now — and it seems like more are popping up every day. It would be easy to get overwhelmed and bogged down trying to learn them all. One of the first suggestions Atkinson and Raffel shared is that print service providers should experiment, but then pick just a few tools to really learn and work with, rather than trying to become an expert at all of them. Every AI tool set has its own rules, its own quirks, and its own methods for getting things done. Find the ones that align with your business needs, and then really learn to use those specific tools.

That said, don’t get so locked into just one or two tools that you ignore the rest of the AI world. While becoming an expert in just a handful is critical to being successful at using those tools, the rapid pace of technological evolution means both the tools you’re using and the ones you aren’t are both evolving rapidly. Make it a regular habit to at least check out what else is out there, in case something has launched that would actually serve your business needs better.

2. Use the Style Reference commands. If you are using AI to help manage color, one great way to do it, the pair suggested, is to use the Style Reference command — sref on many tools, but check the one you’re using for the exact command — to create a color match. For example, you can go to the Pantone website, they noted, pull the URL for a specific Pantone color swatch you are trying to match, and then use the sref tool to tell the AI tools that is the color you are trying to hit. It will analyze the colors and give you information such as the RGB, CMYK, hex, and L*A*B information so you can not only make sure the right color is being used, but also that the right color information is being fed back into your color management system.

3. Don’t be afraid to upscale. There are some amazing AI tools out there for upscaling images, and every printer should consider checking them out. How often does a client send in a web image and expect it to print beautifully in a high-resolution application? Like other AI tools, these are evolving and getting better at a rapid pace, so it is worth getting to know what’s out there and how they work. Being able to reliably upscale images without needing to go back to the client will save a huge amount of both time and frustration on all sides. Atkinson and Raffel suggested checking out Trace Journey, GigaPixel, and Vector.ai as tools for this they have used and are playing with themselves.

4. Pick the right color space. With more customers providing AI-generated elements as part of their projects, it’s important to note that these generators use the RGB color space. Choosing the right conversion tools is key to ensuring it looks the way they want when it is switched to CMYK. Raffel and Atkinson noted that it really depends on which tools were used to generate the art as to which color profile to use when converting. Don’t be afraid to go in and try using a few of them to see what the results look like before finalizing the PDF for print and sending it to the RIP. It can and will make a massive difference in the color of the final application, so getting it right at this stage means fewer reprints and customer frustrations on the back end.