The Exposure Triangle Is Misleading And Unnecessary.

Increasing the gain (aka amplification, defined for this purpose as voltage at the input to the analog to digital converter divided by number of electrons counted) has the potential to reduce input-referred read noise (particularly if the conversion gain is increased by reducing the size of the pixel capacitor). The improvements may be small once the gain has been increased past a certain point. The downside is clipping the ADC before the sensor saturates.

That has some jargon in it. If you don't understand a term, ask me and I'll explain.

I, on the other hand, really like the live histograms on MILCs. But it's taken me a long time to be able to come to terms with what they mean in terms of the raw files.

Not to go into much detail, analogue amplification is a part of signal conditioning, just a ubiquitous tool of electronic design, helps to reduce noise.

I use in-camera histograms for orientation purposes. They kind of suck, and I don’t even bother to remember how my different bodies calculate them, but they give me an enough accurate approximation for my workflow requirements.

It’s usually pretty easy to find out how the histogram is implemented in a specific body, but you still have to remember to do the “correction” in your head.

Not everything has to be 100% accurate to be useful.

Ok, thanks Jim. To start my parroting attempt:

• the benefit of analogue amplification is potentially a reduction in read noise. Read noise is noise added by the camera's circuitry? Read noise was high in early cameras and had a major impact on overall noise, but in modern cameras it is much lower and can often be ignored? So analogue amplification is no longer anything like as important as it was 25 years ago?

• the downside of analogue amplification is the risk of clipping before it would otherwise happen?

• So there was always a trade off in its usefulness, and with modern sensors that trade off is more towards not turning up the amplification?

• In the early years, the ISO knob turned up the amplification but these days it tends to do something else?

When a histogram tells me that my red channel is severely clipped, while it is not, it's not about not being 100% accurate, it's about being grossly misleading.

In the last 10 years or so I haven't seen a camera that doesn't have an analogue amplifier before analogue-to-digital conversion happens.

Not always. Judging by early Kodaks, could be multiplication, could be hybrid.

What does that mean and how is it different from amplification?

Multiplication happens after analogue-to-digital conversion.

OK, so bear with me... analogue amplification physically changes the input before ADC meaning the ADC produces a different value to what it would have done without amplification, while multiplication just changes the value after ADC? What are the reasons/advantages to doing it either way?

Thanks.
D

EDIT: I have this vague memory of Phil Askey saying something in a review years ago. A Nikon review, I think. Something about Nikon using analogue amplification to balance the 3 channels prior to ADC. Something like that. Didn't you once write a raw convertor specifically to convert Nikon raws better than anyone else? You probably know about Nikon do stuff better than most.

Yes

Design considerations are SNR, price, complexity, parts availability, current consumption, heat, space, reliability.

Ah, those sound like engineering/manufacturing things. Any advantages/disadvantages to the photographer? Or can those choices be safely ignored as it ends up with the same result anyway?

i set my camera up using the histogram for every photoshoot ,just to put the levels in the right position for the subjects im shooting, i also have the zebras set to 110% as a backup in case i accidently move a dial by mistake.

I don't know what Nikon did with the D1, but all the Nikon digital cameras I've used in the last 20 years did the white balance prescalinig digitally, after the ADC. I think it ought to be done in post.

Do you do quality control by looking at the raw histograms?

Digital gain introduces histogram combing. It's not harmful if there's enough noise for good dithering.

That would start with D2X, D1 series and D2H(s) didn't use what Nikon called "white balance pre-conditioning, optimized distribution of analog and digital white balance gain" (which, in fact, is calibrating of white balance as one of the methods used to cope with sample variations).
My second D1 is still in good shape ;)
PS. D1 applies full white balance to raw data.

Design considerations are SNR, price, complexity, parts availability, current consumption, heat, space, reliability.

Ah, those sound like engineering/manufacturing things. Any advantages/disadvantages to the photographer? Or can those choices be safely ignored as it ends up with the same result anyway?

Those decisions don't automatically mean either way, advantages or disadvantages. I examine a camera and decide whether I like it or not. Generally, I avoid using ISO settings where multiplication is in use, as it adds nothing useful to raw data - but clips highlights.