Color theory isn't about pretty wheels, it's about light and perception
Most people learn color from art class diagrams. Those are useful for mixing paint, but they break down fast when you move to screens, printing, or anything involving actual light. The Science Of Color is really three separate problems glued together. You have the physics of wavelengths, the biology of how cones in your eyes translate those wavelengths, and the psychology of how your brain decides what those signals mean. Getting any one of those wrong will make your work look wrong to other people. Here's the practical setup I actually use. Most of my work involves calibrated monitors for web and print production. I start with a hardware calibrator, a SpyderX or i1Display Pro, and build an ICC profile for the specific monitor I'm working on. That takes about twenty minutes. Then I set up a neutral viewing environment. Gray walls, 5000K ambient lighting, no windows with direct sunlight hitting the screen. This part matters more than people expect. A warm room light will shift your perception of blues and reds by enough to make a final product look wrong on someone else's screen. The counter-intuitive part nobody tells you early: your monitor should never be set to 6500K white point if you're doing web work. 6500K is the printing standard, D65. Web content is typically viewed on screens that run warmer, around 5500K to 6000K depending on the device. If you calibrate strictly to D65, your colors will look slightly cool and desaturated when viewed on typical consumer displays. I offset my target white point to around 5800K for web-focused work. It's a small shift, maybe 200 degrees, but it's the difference between colors that look right on most phones and colors that look flat.
Another thing that catches people off guard. Gamut mapping. When you convert from a wide gamut space like ProPhoto RGB or Adobe RGB down to sRGB for web, the conversion isn't always clean. Out-of-gamut colors get clipped or compressed, and the algorithm matters. Perceptual mapping preserves relationships between colors but shifts everything. Relative colorimetric clipping preserves in-gamut colors exactly and throws away everything outside sRGB. For photographic work, I usually go perceptual with black point compensation. For UI design and illustration, relative colorimetric keeps the core palette intact. Pick the wrong one and saturated oranges turn muddy, or blues shift toward purple in a way that's hard to trace back to the source. I ran into a specific problem last year working on a brand identity for a client. The primary color was a deep teal they specified in Pantone. I converted it to sRGB for the digital assets using the official Pantone library values. Everything looked fine on my calibrated Eizo. The client reviewed it on an iPad and called immediately saying the color looked completely wrong, more green than blue. The issue was that iPads default to a wider gamut and higher saturation than my monitor was set to display. The Pantone conversion had pushed some of those teal values outside the typical sRGB gamut boundary, and the iPad was rendering them with its own gamut expansion. My workaround was to hard-limit the specific Pantone color to pure sRGB boundaries before handing off the files, then provide the client with a note about device variation. It cost me an extra hour of file preparation but saved the project. Lighting metamerism is another trap. Two colors can look identical under your studio lights and completely different under fluorescent office lighting or direct sunlight. This happens because the spectral reflectance curves of the two pigments cross at certain wavelengths. Under a light source rich in those crossing wavelengths, they match. Under a different source, they diverge. I've seen this wreck packaging designs where the print proof looked perfect in the studio but the final product on the shelf under store lighting looked like a different color entirely. The fix is to evaluate your colors under at least two different light sources before signing off. A dedicated viewing booth with interchangeable light panels costs around fifteen hundred dollars, but you can approximate it with a simple D50 and D65 comparison lamp for a fraction of that.
Color management in software is where most people give up. Every application handles profiles differently. Photoshop respects embedded profiles. Some older design tools ignore them entirely and assume sRGB. Always check what profile your working space is actually using. In Photoshop, View > Proof Colors lets you simulate how colors will look in a different output space without permanently changing your file. It's not a replacement for actual soft-proofing with your ICC profile loaded, but it catches about eighty percent of gamut issues in ten seconds. The biggest bottleneck I see is people treating color as purely visual when it's also about consistency across media. A color that works on a backlit screen doesn't necessarily work in print. A color that prints well doesn't necessarily reproduce accurately in fabric dye. Each medium has its own gamut limitations and color formation mechanism. Additive mixing for light, subtractive mixing for pigment. They follow completely different mathematical models. Don't expect a color picker value to travel cleanly between them without a proper conversion pipeline. If you're just starting out, skip the expensive hardware for now. Calibrate with what you have. Set your monitor to a known reference, work in a controlled light environment, and learn to recognize when colors look off by comparing them side by side on multiple devices. The theory will fill in once you've seen enough failures to know what to look for.
