If we were to start with the delayed choice quantum eraser setup in the attached image, what would the interference pattern look like for the photons at D0 if D3 and D4 were eliminated (and the two beam splitters that feed them) and the beam splitter configuration between D1 and D2 was altered so that 75% of the photons from each slit to go to D1 and 25% of the photons from each slit to go to D2? Even though 75% of the TOTAL photons arrive at D1, D1 is still receiving an equal amount of photons from EACH slit so the which-way information could not be determined. Likewise for D2. Would the interference pattern at D0 still be identical to the 2nd attached image, with equally dark out-of-phase bands for the photons whose quantumly entangled pairs correspond to D1 and D2? Or would the D1 corresponding bands be three times as dark as the D2 corresponding bands?
if the beam splitter configuration between D1 and D2 was altered so that 75% of the photons from each slit to go to D1 and 25% of the photons from each slit to go to D2?
This behaviour is unphysical from any beam splitter.
Take a beam splitter with two input ports, $\rm in_1$ and $\rm in_2$, and two output ports, $\rm out_1$ and $\rm out_2$. Because of conservation, if you impose the requirement that
- each photon entering $\rm in_1$ goes out of $\rm out_1$ 75% of the time and out of $\rm out_2$ 25% of the time,
then it must also be the case that
- each photon entering $\rm in_2$ goes out of $\rm out_1$ 25% of the time and out of $\rm out_2$ 75% of the time.
The reasons for this are technical; they're explained in depth in this previous answer.
If you were to use such a beam splitter, then you'd get a result halfway between the which-way path (D3/D4) and the quantum-eraser path (D1/D2): you would still see some interference, but with a reduced contrast, and with the distributions moving away from the center: