This is deprecated, use the app in https://github.com/lindemeier/painty

Digital PaintMixer

Mix paint by from base pigments using the Kubelka-Munk equations. Based on the paper of Lindemeier et al. [1].

This is work in progress

I'm porting the unpublished original implementation to this public repo step by step.

Example

This is a collection of base pigments whose absorption and scattering coefficients I've measured in my article [1]:

• Primary Magenta
• Carmine Red
• Cadmium Red Medium
• Raw Umber
• Cadmium Orange Hue
• Cadmium Yellow Hue
• Primary Yellow
• Leaf Green
• Phthalo Green
• Cobalt Blue Tone Deep
• Ultramarine Blue
• Lilac
• Lamp Black
• Titanium White

Figure 1: This is a visualization of the pigments and their resulting reflectance on black and white background.

Create a palette from an input picture and mixing them from base pigments

Command to extract a pallete of size 6 from a input image:

 ./PaintMixer -b ../palettes/BasePigments/AcrylicsBasePigments.json -i ../example/pexels-photo-977737.jpeg -n 6 -o ../example/pexels-photo-977737-extractedPalette.json

Figure 2: The input image [2] we will extract a palette from using the pigments from Figure 1 .

Figure 3: The extracted palette rendered on black and white background.

These are the weights computed with the PaintMixer solver:

Primary Magenta	Carmine Red	Cadmium Red Medium	Raw Umber	Cadmium Orange Hue	Cadmium Yellow Hue	Primary Yellow	Leaf Green	Phthalo Green	Cobalt Blue Tone Deep	Ultramarine Blue	Lilac	Lamp Black	Titanium White
0	0	0	0	0	0	0	0.0947	0.262	0.0903	0.215	0	0.337	0
0	0	0	0	0	0	0	0	0	0	0	0.019	0	0.981
0	0	0	0	0	0.00757	0	0.192	0.59	0.139	0.0424	0	0	0.0292
0	0	0.0854	0	0.171	0.0692	0.674	0	0	0	0	0	0	0
0.146	0.211	0.437	0	0.206	0	0	0	0	0	0	0	0	0
0	0	0	0.0063	0	0	0	0.0369	0.212	0.216	0.259	0.0625	0.206	0

Table 1: These are the weights for mixing the paints as shown in Figure 3 from the base pigments shown in Figure 1.

These are the absorption and scattering coefficients of the extracted palette computed using the PaintMixer solver and output as json:

{
    "palette": [
        {
            "K": [
                1.4671067147397455,
                0.9698479705161202,
                0.7784896610234372
            ],
            "S": [
                0.043076041668283649,
                0.08198151765139729,
                0.06474356110909095
            ]
        },
        {
            "K": [
                0.06539400295725369,
                0.06421413897814343,
                0.06737760085328282
            ],
            "S": [
                0.7612055495574275,
                0.9165666516471837,
                1.0169171824109196
            ]
        },
        {
            "K": [
                1.558403138979944,
                0.706115436691196,
                0.7202637267172865
            ],
            "S": [
                0.06732445406550918,
                0.11887217439153461,
                0.11310752780489106
            ]
        },
        {
            "K": [
                0.013819327449075598,
                0.3409922592740801,
                2.53557022496248
            ],
            "S": [
                1.0679933129986213,
                0.6562281343632745,
                0.01595292961985854
            ]
        },
        {
            "K": [
                0.13424884989254433,
                1.6172395172573063,
                1.8094305792798447
            ],
            "S": [
                0.7523791290476137,
                0.1791898328836181,
                0.1282115084482811
            ]
        },
        {
            "K": [
                1.5054829224326548,
                0.9997198355146348,
                0.584845819500503
            ],
            "S": [
                0.041619867600404109,
                0.10086032379657013,
                0.11154146006304615
            ]
        }
    ]
}

References

[1]: T. Lindemeier, J. M. Gülzow, and O. Deussen. 2018. Painterly rendering using limited paint color palettes. In Proceedings of the Conference on Vision, Modeling, and Visualization (EG VMV '18). Eurographics Association, Goslar Germany, Germany, 135-145. DOI: https://doi.org/10.2312/vmv.20181263

[2]: Image from (https://www.pexels.com/photo/scenic-view-of-beach-during-dawn-977737/), accessed May 22, 2019.