Just in case anyone was wondering (I was), it turns out that my smartphone camera produces a better SMI color score off a ColorChecker Passport target than a full frame Nikon D610 DSLR .

My latest phone, a late 2017 incarnation of the LG V34, produces raw DNG files, so I went poking around.  From what I could gather the sensor is most likely Sony’s IMX 234^[1], 1/2.6″, Back Side Illuminated, stacked and based on the latest and cleanest Exmor RS technology.   The sensor’s 1.12um pixels produce 16MP raw files with 10-bit depth, which I understand to be typical for current phone cameras.  Other features include phase detect AF, an electronic shutter with variable integration time, HDR, hot pixel suppression and raw noise reduction (ugh!) – plus a slew of video features.

I was curious about color ‘accuracy’, since I had heard reports that phone cameras did not produce quality tones in the past.  The data sheet included the following plot of the sensor’s Color Filter Array sensitivity to wavelengths, before lenses and/or other (UV/IR) filters are introduced in the imaging path:

This looks fairly typical of good quality pigments found in regular digital cameras^[2] so it bodes well for the potential color ‘accuracy’ of the smartphone.  Accuracy is between quotes because I only tested the sensors over a very limited set of tones, the 24 contained in an X-Rite ColorChecker Passport Photo.

An often used metric for color ‘accuracy’ is Sensitivity Metamerism Index, which is computed for a specific illuminant by taking 100 minus 5.5 times the average deltaE76 of (in this case) the 18 color patches of the CC24 compared to a known reference.  An SMI score of 100 represents unattainable color perfection (the sensor produces 100% correct values for all color patches), 90 state-of-the-art excellence, 85 pretty good, 80 decent – and 75 unacceptable, in 2018.

I captured raw files of the CC24 Passport in the mid-April afternoon city sun with a Nikon D610 DSLR and my LG V34 smartphone.  Correlated color temperature turned out to be about 5300K according to RawTherapee 5.4 and ACR CC 2018.  RT produced virtually identical CCT values from the NEF (D610) and DNG (V34), while ACR got 5300K from the NEF and 4800 from the DNG. 5300K it is.

I followed the procedure outlined in this earlier article to obtain an individual compromise color matrix from raw data to the XYZ color space, once for each camera.   The matrices were optimized so that the CC24 tones captured in the raw files would show the smallest average CIEDE2000^[3]  perceptual difference to the reference values published in BabelColor.com’s database.

End result: in these conditions the Nikon D610 came away with an SMI score of 84, while the LG V34 with 86.  Go Smartphone.  On the other hand the D610 has a more robust-looking matrix, see for instance the off diagonal blue coefficient.

For my own reference, here is the relative data.  LG smartphone first:

And the Nikon D610’s:

Notes and References

<sub>1. The Sony IMX234’s datasheet can be found here.
2. Typical digital camera pigment sensitivities can be seen here.
3. CIEDE2000 is a modern measure of color difference, you can find its definition here.</sub>