What the two stops are actually paying for
The thread traced the figure to the manufacturers' own data sheets. The example given was Portra 800, whose sheet specifies a Wratten 80A filter under 3200K tungsten and tells you to meter at ISO 200 — two stops down. The point the correction turns on is that the adjustment describes the filter, which is a deep blue piece of glass that does not let much light through. It is not a property of tungsten light.
Why it gets misread
- The data sheet prints the filter and the exposure adjustment together, so the adjustment reads as though it belongs to the lighting condition rather than to the glass in front of the lens.
- A commenter in the thread cited the data sheets as the reason for a flat two-stop rule; the correction was that the same sheets name the filter the rule depends on.
- One reply called the original advice inaccurate on this point, and another agreed, so the thread did not treat it as a matter of taste.
If you are correcting colour afterwards instead
Then the compensation does not apply, which was the original commenter's point and the thread did not dispute it. One reply described sending film to a lab whose scanner corrects the white balance and removes the blue cast, then finishing the correction digitally, and said the result was good enough that no filter was used at all. The trade named in that reply was a slight blue that survives in greens — visible in grass, between the small shadows.
The short version
- Filtering on set: fit the filter the data sheet names and lose the stops it costs.
- Correcting later: no filter, no exposure compensation, and a colour cast to deal with in the scan or the grade.
- Either way the two stops belong to the filter. Nothing about tungsten light itself asks for them.