Comment by tambourine_man
2 days ago
It could be an exposure issue. Film has a response curve with a big “shoulder” in the highest values. It makes it really hard to blow out highlights.
Digital sensors have a linear response to light, so if the highlights are a bit over a threshold, they are gone.
If you’re willing to tolerate more noise and shoot RAW, you could underexpose, perhaps by as much as 4 stops, and apply a strong curve in post. It would pretty much guarantee no blown out highlights.
Most people find luminance noise aesthetically pleasing up to a point and digital is already much cleaner than film ever was, so it’s a worthy trade off, if you ask me. But “Expose To The Left/Right” is a heated topic among photographers.
Spitballing, but a HDR digital camera could be designed with a beamsplitter similar to that of the 3CCD ( https://en.wikipedia.org/wiki/Three-CCD_camera ) designs that projects to an assembly with only a sensor behind it, another to an assembly that has a 4 stop neutral density and sensor, and another to an assembly that has an 8 stop neutral density and and sensor.
This way it wouldn't suffer from any parallax issues and sensor images should then also line up to allow it to be reconstructed from the multiple sources.
That said... HDR images can be "bland" with it being washed out. It would probably take a bit more post processing work to get the image both high dynamic range and providing the dynamism of what those old Saturn V launches showed.
Back in the day, Fuji had a sensor in which half the pixels had a 2-stop neutral density filter, IIRC. It was a 12MP sensor, with an effective resolution a bit higher than 6MP. It was amazing the amount of highlight you could recover in Adobe Camera Raw and the TIFs/JPGs were beautiful, as it’s usually the case with Fuji.
Alas, it didn’t work out in the market, people weren’t willing to trade half their resolution for more DR, turns out. Also, regular sensors got much wider latitude.
I think what you need is a plain old half-silvered beam splitter, not the 3CCD prism. The dichroic prism uses combinations of coatings to separate RGB into different optical paths. You don't need that.
Half silvered would work too. I was using the 3CCD prism as an example of splitting an image into multiple parts and avoiding issues with parallax. If you stacked cameras on top of each other with their own lenses, you'd inevitably get slightly different focal lengths, focal distances, and camera positions.
Aside with the dichotic filter... there was a neat trick that was lost (and refound) with a sodium 589nm dichromic (notch) prism. https://youtu.be/UQuIVsNzqDk