Blue is not a single colour, it is a problem you solve differently depending on your medium
Most people approach blue backwards. They see a reference and try to mix the perfect blue from scratch, which means juggling phthalo and ultramarine, adjusting tints, fighting with opacity, and wondering why their swatch looks nothing like the Pantone book. That is the wrong way to start. You begin with the undertone. Blue is almost never a flat neutral, and treating it as one is what makes every colour mixing tutorial disappoint. Here is the practical breakdown. I am going to explain how blue actually behaves across pigments and light, where the common traps are, and what I actually do when a client sends me a reference with a blue I cannot replicate by eyeball.
What Colours Can Make Blue A Comprehensive Guide To Mixing And Understanding Blue Tones
Strictly speaking, you cannot make blue from other colours in subtractive mixing. Blue is a primary in both RYB and modern pigment systems. If you mix cyan, magenta, and yellow together you get brown or black, never blue. The real question is always more useful: which existing blues get you closest to the target, and what tiny adjustments do you need to make? I work mostly in acrylics and digital illustration, so I think about blue in two layers. First, I identify the hue family. Second, I adjust warmth, saturation, and value independently. Mixing a custom blue from two primary pigments usually starts with either phthalo cyan or ultramarine blue as the base. From there you add a warmer or cooler modifier, not random extra colours. The mistake most beginners make is adding yellow to blue and calling it teal without considering that the yellow will drag the saturation down. Phthalo blue mixed with a touch of yellow gives you a bright cyan, but a yellow ocher or cadmium yellow light will muddiness things fast. If you are mixing digitally, the logic is similar but faster. Set your base to a standard cyan around #00AAAA or an ultramarine around #312BE2, then shift the hue slider toward green for cool blues or toward red for warm blues. The HSB model is cleaner for this than RGB because saturation and brightness are separated. You do not want to fight chroma while also trying to fix value.
I ran into a specific problem last year on a branding project where the client needed a very particular cool, slightly greyed-out blue for a website they were building. The reference was a desaturated blue with almost no green bias. I started with phthalo cyan, added white, then a tiny amount of burnt sienna to desaturate it. The result was too green. The workaround was to switch to a blue violet base instead, pull it toward neutral grey with a complementary orange, and then reposition the hue slightly. That took about twelve minutes total. Trying to force desaturation through green would have cost me an hour and still looked wrong under different lighting.
Get the Full Details
Understanding blue undertones before you touch another pigment
This is the part nobody stresses enough. Warm blues lean toward red or violet. Cool blues lean toward green or cyan. Ultramarine blue is inherently warm. Phthalo blue is inherently cool. Cerulean sits somewhere in the middle but leans cool. Cobalt blue is relatively neutral with a slight cool bias. Knowing where your starting pigment lives on that spectrum determines everything that follows. If you need a sky blue, ultramarine is a poor base even though traditional painting classes tell you to use it. Ultramarine with white creates a lavender tint because of its red bias. A phthalo or cerulean base gives you a much cleaner sky blue with less subsequent correction. If you need a deep navy, ultramarine is a better choice because its red undertone keeps the resulting dark from leaning too cyan. The other counter-intuitive thing is how transparent versus opaque pigments behave when layered. Phthalo blue is extremely transparent. Ultramarine is moderately transparent. Mixing them together does not produce a predictable intermediate colour because transparency affects how light interacts with the layer underneath. I learned this the hard way on a print commission where I mixed phthalo over a white ground and expected it to look the same as the digital mockup. It did not. The opacity difference shifted the perceived saturation enough that the proof came back looking flat and washed out. The fix was using a more opaque blue like cerulean for the final layers, or adjusting the digital file to account for the lower chroma in print.
How to mix specific blue families reliably
I keep a mental shorthand for common blue families so I do not have to recalculate each time. The shorthand works because it maps to real pigment behaviour rather than abstract colour theory rules. Cool cyan blues start with phthalo cyan. Add a trace of yellow if you want a brighter green-leaning cyan, but be careful with the yellow. One dot is usually enough for a standard tube mix. Add white for tint. Do not add black, because black will desaturate too aggressively and shift the hue. Use a complementary mix or a gray neutral if you need to darken without losing chroma. Warm sky blues are harder because most warm blues tend to go lavender instead of sky blue. Start with a neutral to warm blue like cerulean, add white, and if it leans too green you correct with a microscopic amount of red or violet. If it leans too purple you correct with a microscopic amount of yellow. The corrections are so small that pre-mixing tiny separate pots of red and yellow modifiers near your workspace saves time. I keep little ceramic dishes with drops of permanent rose and hansa yellow light within reach for exactly this reason.
Deep navy and midnight blues do not come from black plus blue. That produces a dull, lifeless colour. Mix ultramarine with a touch of phthalo cyan for a deeper cool navy, or ultramarine with a hint of burnt sienna for a warmer, richer navy. The key is keeping some chroma in the mix. Pure black kills chroma immediately and you cannot get it back. Teal and turquoise are really blue-green hybrids. Start with phthalo cyan and shift toward green. If you use a green primary like viridian, you lose saturation compared to shifting a cyan with yellow. Beginners often use viridian for teal and then wonder why the colour looks chalky next to the reference. The viridian path is not wrong, it just produces a different quality of teal that is lower in intensity. Use it when you want muted teals, avoid it when you want vivid turquoise.
Pigment limitations you should accept rather than fight
Not every blue you see in digital references is achievable with every pigment system. Phthalo blue is potent and stains almost everything. It is also chemically aggressive in some mediums and can cause drying issues in oil paint if used in large proportions. Ultramarine blue is safer across mediums but fades under strong UV over decades. That is why museum conservators prefer it less for outdoor frescoes, even though artists love it for indoor work. If longevity matters for a specific project, you need to choose your blue with that in mind rather than picking the one that matches the reference best and hoping for the best. In digital work, the equivalent limitation is gamut. Your display cannot show every saturated blue that exists in nature or in print. The sRGB gamut covers most screen work but misses the deepest cyans and some electric blues. If you export for print, CMYK has an even smaller gamut for blues, particularly bright cyans. I have lost count of how many times a designer sent me a file that looked fine on screen and printed as a muddy desaturated blue. The solution is to proof in CMYK mode while you work, not after the fact.
A practical workflow I actually use
When I get a blue reference, I do three things in order. First, I identify whether it is cool or warm. Second, I estimate the value and saturation relative to a full scale. Third, I pick the closest base pigment or digital colour. Only after those three steps do I start mixing. For physical mixing, I prepare my base, test on a scrap surface that matches the final material, compare it under the same lighting as the reference, and then adjust in tiny increments. I write down the ratios once I hit a close match because memory is unreliable and the same tube of ultramarine from different batches can vary slightly. For digital work, I sample the reference, set my base, then refine using HSB sliders with saturation and brightness adjusted separately. I keep a copy of the original sampled colour on a secondary monitor so I can compare continuously rather than relying on memory after a few tweaks.
There is no shortcut around this process other than volume of practice. You will mix wrong colours until you stop making the same mistakes twice. The undertone awareness and the pigment limitation knowledge cut the failure rate significantly, but they do not eliminate it. That is just how colour mixing works.
