I'm assuming here that reciprocity failure will cause an increase in contrast because the shadows suffer greater failure than the highlights. Looking at the HP5 PDF makes it look like one gives 3x more exposure for each additional stop on the meter past 30s and I've tested this out to about 1000s of actual exposure without any dramas. A bit more contrast but I haven't tried to quantify it.
My question is, to what meter reading does one apply the reciprocity-failure correction? (a) apply it to the meter reading and the table takes into account the contrast expansion and I don't lose shadow details, or (b) the reciprocity failure curve will apply to each zone individually. The published curve could easily be computed to either approach and I don't know which it is.
Case (a) is easy and means there's not a lot of contrast expansion happening. Zone II or III stay at about where they were, midtones get a bit denser and highlights get real dense, fixable by reduced development. It also means I don't have to think much.
But if it's case (b), I figure if I apply it to the chosen exposure, what I'll be doing is making sure the midtones come out identically (for unadjusted development) but shadows will be pushed further down and lost.
- shadows meter at 1920s (III)
- midtones meter at 480s (V)
- highlights meter at 120s (VII)
If I apply the rule to the exposure I would normally go with (480s), we end up with 12960s. For (a), midtones end up at about V-VI and highlights at VIII. But if it's case (b), midtones are still at V but the shadows being 2 stops down will get 9x less effective exposure and suddenly they're below zone II and probably lost.
So perhaps for (b) I apply the correction factor to the shadows... 1920s becomes 116640s and then take two stops off to put 'em back in III gives us 29160s. At that point, the highlights are somewhere up around IX or X unless I reduce development.
Does anyone know which way the tables are formulated and therefore which meter reading the correction should be applied to? 13,000 to 30,000s is a big difference, and about 3 stops difference in total dynamic range recorded. Or should I just go buy some Acros?