dynamic range vs 'dynamic range'

OK, someone school me. Dynamic Range is the range of intensities which can be recorded by a sensor, such as film. So, when we say slide film has ~5 stops of dynamic range, it can record an intensity of 1 up to 32 on a linear scale. Modern digital sensors and color print film have more dynamic range.

Here's my question, which has the greatest bit-depth (I'm sure there is another, correct term for this concept)? Digital has only about 10-12 bit depth (I think I remember hearing that somewhere). A 16-bit scan, should have 16-bit depth. So here's the question: does film support that bit depth when being scanned? When there are 32-bit scanners, will color slides support 32-bit depth?

Thanks, Jon

My measurements for slide film (E100) show a dynamic range of about 7 stops (128:1), and color neg (Portra 160VC) of about 11 stops (2048:1). This from shooting gray scales and plotting the density vs. exposure. A bit depth of 10 is a dynamic range of 10 stops or 2**10 = 1024:1. Depth of 12 is 4 times that.
Scanners that produce 16 bit files are often using only 12 of the 16, so 4096:1. Scanning a limited range slide on a scanner capable of a wider range will not add info that isn't there on the film. But the density range on the slide is larger than 7 bits, the contrast is increased. So you need the 12 bit scanner, or more, to read all the info that is on the slide.
The largest dynamic range sensor is B/W film. A very large dynamic range can be had by shooting several pics at different exposures, and combining the results into an HDR file, where the pixel data is not an 8 or 16 bit integer, but a floating point number. Photoshop CS2 can do this. I have seen a photo of a dark room, black cat under the desk, and full sunlight out the window, with texture in all. 18 stops?

Jim,

Correct me if I'm wrong, but dynamic range only refers to a sensor's range of recording, not the detail which is recorded in that range. For example, if a sheet of film only had 1 stop of dynamic range, recording only between 1 and 2 intensity levels. It could still theoretically be able to differentiate between 1 and 1.5, right? So, even if slide film has only 5-7 stops of DR, it coult have even more bit depth than that, right? So, 1, 1.25, 1.75, and 2 represent 1 stop of DR and 2-bit depth. This could theoretically go one indefinitely. My question is how far?

I think you can consider film to have infinite "bit depth" (as you term it) meaning that for all practical purposes it can represent every color and shade within its palette or gamut, not just some number such a 256 or 16,000 or millions of colors.

And in your response to Jim -- one stop of range is not 2 bits, it is infinite in practical terms. Think of it as everything between f5.65 and f8. That can be lots and lots even bazillions of colors and tones!

Now that I've crawled out on this limb I know that some renowned Physicist lurking on this forum will explain it properly. All I ask is that I not be made fun of.

; >)))

Henry,

That's my understanding of it also. Theoretically, it should be infinite. Of course, there must be some limiting factor - but this may be well beyond the ability of any recorder to detect. I just want to be sure that's actually the case.

The dynamic range is based upon the electron capacity (full well capacity) of the CCD or CMOS detector. Bit depth only tells you the number of discernable steps in the dynamic range.....not the dynamic range itself. This range is dependant upon the sensor and the quality of the analog & digital conversion process. Any errors in the conversion & noise detract from the full potential dynamic range. You could increase bit depth on an Epson 4990 to 24 or 32 bits, but the dynamic range would remain unchanged as the present electronics would not be capable of any further resolution.

If you want to reduce the noise in your scanner, keep your scanner in a colder room and don't worry about extra bits!

Dynamic range is what the film or sensor capable of recording.

Bit depth is an entirely different thing.

In digital, the signal captured go through an Analog/Digital (A/D) sensor, and the signal is put INTO a certain DEPTH of a digital file. The digital file may be an 8 bit, 12 bit, 14 bit, or 16bit in depth.

Bigger depth gives more tonality steppings. For example, if you have a 5 stop dynamic range signal coming, and is put into an 8 bit file, then that 5 stop will be divided among 255 levels (not linearly though). The same 5 stop signal can also be put into an 16 bit file, where the 5 stop will be divided among ~65,000 levels - thus better tonality/gradation compared to the 8 bit file.

35mm DSLR capture in 12 bit, and it will either be saved an as 16 bit TIFF, 8 bit TIFF, or 8 bit JPEG.
Older medium format backs capture in 14 bit, and should be saved in a 16 bit file.
Latest MF backs, I believe captures 16 bit.
When you scan film, it's the same. You have that 5 stop, 7 stop, or 9 stop, negative. You put it under a scanner and you have an option of putting those information into an 8 bit, or 16 bit files.

I don't think bit depth has an equivalence in the film realm.

I believe Henry is right, one stop of dynamic range can have an infinite number of steps theoratically (of course in practice one stop may not really be infinite. but this is a different discussion altogether.)

In theory any part of the film's density range could be scanned to 12 bits. But the scanner starts at black, not at the lower end of the selected range. So setting the white and black points close together does not expand that range to 12 bits, the histogram shows many missing levels. The expansion does not generate new in-between levels. But if we imagine a scanner that can select a range and digitize that range to 12 bits, then the limit is noise, both from the film grain (the film's noise) and the scanner's sensor and electronics.

The ladder analogy is often used:

- dynamic range is how tall the ladder is,
- bit depth is how many steps there are in the ladder.

-> it is pointless to have too many steps on a short ladder, which is why smaller sensor DSLR only sample on 12 bits. The digital backs appear to have more real DR, which explains why the sampling is done on 16 bits.

As far as scanning is concerned, you'll find that even on an Imacon Precision 3, it is real hard to get more than 3.5 usable stops of DR. This means that very little detail is to be found in the shadows of a slide whose highlights are just short of being blown when there is 3.5 stops between the 2 areas of the image.

Therefore, if you shoot slides knowing that you'll end up scanning them on an imacon or lower end scanner, it is a good idea to be careful with scenes having more than 3.5 stops of DR.

Regards,
Bernard