you are right! but i only give you 66.666% mark for that 2 points. the exact answer should be:

Perspective is ~~only~~ a function of the subject distance **and focal length**, so the sensor doesn't matter. = 100% mark answer

I disagree. Perspective is the relation between distance and relative size (that objects farther away are smaller, and by how much), which purely depends on the subject distance. The angle of view (which depends on focal length and sensor size) determines the crop, but you can get exactly the same perspective by cropping later, this doesn't change the perspective. You can correct the smaller angle of view of a smaller sensor by choosing a shorter focal length, which will yield the same perspective and crop, only the DOF will be different.

80% mark... 100% = Bokeh is mainly a function of the lens **aperture**.

The shape is determined by the iris, but the other attributes like how well defined the aperture shape is and if the edges are obvious depends on the under- or overcorrection of spherical aberrations, which is determined by the lens design (and setting of the defocus control ring on the few lenses that have one).

let me tell you a secret, that nobody on earth do care to put it on the internet. COC is a subjective matter, it varies with each different observer, trust me, it cannot be counted in the engineering equation. just like colors or your favourite food.

This statement is not relevant to the discussion. Whatever criterion you choose, it will scale proportionally with sensor size. If you view on a 24" screen, the image from the 1/3rd inch sensor has been enlarged much more to make it that large (in mm, not in pixels). So the CoC is also enlarged, which means the CoC on the sensor must be smaller to compensate.

yup yup! PRINTING! it has to do with resolution! thats why sensor size (pixel count actually) comes into equation/discussion. if you talk about resolution, then yes... 1/3" sensor lose!

Sure, the resolution provides a lower bound, no point in choosing a CoC that's lower than the resolution of either the sensor or the output device (times scaling). But I'm not at all talking about pixel count, for my example, you could just assume both sensors are 1080p, everything except the sensor size (and derived values like focal length, for the same angle of view) can be identical. Actually, for the simple DOF equation, you assume pixel size << CoC, and when viewing (on a screen or on paper), your eyes should be unable to see individual pixels, otherwise the resolution was way too low, or you're too close (which is why CoC assumes a certain viewing distance). So as long as the CoC is much larger than the pixel pitch, the pixel pitch doesn't matter, and you could just as well assume a constant pixel pitch.

man! i dont understand what you are talking about! are you saying that when i put a 50mm f1.2L @f8 on the 5D2, the image will be similar effect to the same lens @f1.2 on the EOS 350D? (the smallest DSLR sensor)?

I'm saying that you can compensate for the deeper DOF on smaller sensors by using a proportionally larger aperture and proportionally shorter focal length (to compensate for angle of view). The 350D sensor is 1.6x smaller, so the lens should be 1.6x shorter for the same angle of view (~30mm), and the aperture should be a factor 1.6 larger (f/2 on the 5D would be equivalent to f/1.2 on the 350D). This can be derived from the DOF equation, for the simplified case for a symmetrical lens (probably not a bad approximation for a 50mm lens), the hyperfocal distance h=f

^{2}/NC, with N the f-number and C the CoC. The near and far limits are a function of h, the subject distance v and v - f. Assuming v >> f (usually true except for macro), we can approximate v - f with v, and the sensor size/aperture/focal length only changes h. If you change the sensor a factor x, both f and C will change (see above), but h must stay the same (same DOF). So f

_{0}^{2}/N

_{0}C

_{0} = f

_{1}^{2}/N

_{1}C

_{1} = (x*f

_{0})

^{2} / N

_{1}(x*C

_{0}) => N

_{1} = x * N

_{0}. If you do it the hard way without my assumptions of a symmetrical lens and v >> f, the change in N will be slightly different than x. You can argue about that exact value, I'm not going to claim that it's exactly 1.600 or something like that, but it's clearly not 1, so the sensor size does have an effect, since you've effectively gotten extra lower stops (f/1.2 on 350D is now f/2.0, so I can go two stops lower than before, and lose two stops on the other end). This is why large format photographers use these extreme aperture's like f/64 to get enough DOF.