actually that's a nice exercise so let's have a look at this... if I was a math teacher, I'd be tempted to make that my next exam
b1,b2,b3 = black area in panel 1,2,3
c1,c2,c3 = constant amount of black in panel 1,2,3 (fonts, frame,scale/lines)
w = width of bars in bargraph (20 pixel)
h_max = height of bar graph that would represent 100% ( = 200 pixel)
scalefactor = factor by which graph in panel 3 is scaled (=0.254)
totalblack = total black area
totalarea = area of whole comic (= 740 pixel *180 pixel = 133200 square pixel )
r = radius of pie graph (= 73 pixel)
(1)
totalblack = b1 + b2 + b3these are the 4 variables... (everything else is known)
(2)
b1 = c1 + \pi r^2 totalblack/totalarea(3)
b2 = c2 + w h1 + w h2 + w h3 = c2 + w (h1+h2+h3)we'll have to assume that the bargraph shows a percentage - h_max would then be 100% and h1,h2,h3 accordingly:
h1 = h_{max} b1/totalblack;h2 = h_{max} b2/totalblack ;h3 = h_{max} b3/totalblackhence (and, because of eq. 1 ),
(3a)
b2 = c2 + w h_{max} /totalblack (b1+b2+b3) = c2 + w h_{max} /totalblack (totalblack) =c2 + w h_{max}(the amount of ink used in graph 2 does not depend on the other graphs, because we show a bar for each panel and those bars always add up to 100% or h_max - that is of course not true, if we adjust h_max so that our bars don't become too small/big)
Panel 3 shows the whole thing scaled by scalefactor, which of course contains an even smaller version scaled by scalefactor^2 compared to the original which contains... etc. etc. (of course with pixels, you're not able to show that, but you'd approximate that by the correct grey values)
(4)
b3= c3 + \Sigma_{n=1}^{\infty}(scalefactor^n totalblack) = c3+ totalblack \Sigma_{n=1}^{\infty}(scalefactor^n) = c3 + totalblack \frac{scalefactor}{1-scalefactor}we have 4 variables and we have 4 equations - seems we got exactly one solution.
in the last step, the geometric progression was replaced by a simple fraction (showing how you get that with a bit of hand-waving):
x=\Sigma_{n=1}^{\infty}(scalefactor^n)multiply both sides with scalefactor
scalefactor x=\Sigma_{n=2}^{\infty}(scalefactor^n)substract both equations from each other
(1 - scalefactor) x =\Sigma_{n=1}^{\infty}(scalefactor^n)-\Sigma_{n=2}^{\infty}(scalefactor^n)= scalefactorsolve for x
x = \frac{scalefactor}{1 - scalefactor) }so, putting all that back in eq. (1), we end up with:
totalblack - totalblack \frac{\pi r^2}{ totalarea} - totalblack \frac{scalefactor}{1-scalefactor} = totalblack ( 1 - \frac{\pi r^2}{ totalarea} - \frac{scalefactor}{1-scalefactor}) = c1+c2+c3+wh_{max}i.e.
totalblack = \frac {c1 + c 2 + c3 + w h_{max}}{\frac{1-2 scalefactor}{1- scalefactor}-\frac{\pi r^2}{totalarea}}
from which we trivially get the fraction totalblack/totalarea for graph 1, b1 and b3 and hence h1,h2,h3 for graph 2 (b2 is already fixed with c2+w h_max) .
Measured from the comic (c1,c2,c3 are just the frames, graph-scales, text, etc, not the sector in the pie graph, the bars or the scaled graph):
c1=4705; c2=4997; c3=3259; w=20; hmax=200; scalefactor=0.254 ; totalarea=133200;r=73
(the corrections for possibly non-square pixels is left as an exercise to the reader...)
I.
Edit: correction: the scalefactor in a single dimension is 0.254 - for the used area that means A=xy ---> (x*0.254)*(y*0.254) = A*(0.254^2), right? So scalefactor=0.254^2=0.064516 (for the area // amount of ink used)