Why isnt this possible....

[Roadkill_6mm]

A bit of fun type thread...

I've been wondering, from what I understand a digital camera works by light going through the lens and hitting the sensor. The sensor is made up of tiny 'areas' which measure and record the light hitting it this creating a picture.

As all this is controlled electronically regarding sensitivity etc I have a couple of questions....

Why do I need a ND filter to reduce the amount of light hitting the sensor? Why cant I just compose my image and tell the camera how long I want to expose for and it will set the sensitivity of the sensor automatically?
Would this even be possible??

If that IS possible (in theory) think of the applications.....want to expose the foreground but under expose the sky? Simple the camera can scan the image it sees, recognise which bits are sky and automatically underexposed them on the sensor resulting in an image looking like it was taken with a gradual ND filter.

Surely this sort of thing is possible?

Or am I being a numpty??

Neil

 

[djinn]

don`t get giving em ideas, auto metering is considered blasphmy

 

[Strangways]

The problem is that it is such a wide contrast range. However the Fuji F500 does it by taking two images at the same time and blending them together. It is already happening.

 

[Musicman]

Why cant I just compose my image and tell the camera how long I want to expose for and it will set the sensitivity of the sensor automatically?

Because light is delivered in quantum energy packets - photons.

The sensor is fixed in its sensitivity. Essentially, it's a bucket that collects photons dropping into it. Once it's full up and starts brimming over, it's blown the highlights. Where there's little light, very few photons end up in the bucket.

Increasing the ISO amplifies the data gathered by multiplying it by a scale factor. Reducing the ISO tells the camera's software to throw away data, reducing the signal to noise ratio and introducing noise into the final image. AFAIK, Canon cameras use a base ISO of 100 and Nikon, ISO 200. Knock my 5D down to ISO 50, and the result will be slightly noisier than ISO 100.

What you can't do is change the size of the bucket on the fly (it's fixed in silicon) which is essentially what you're suggesting.

 

[Roadkill_6mm]

Ah, ok. I wasnt thinking of the sensor as a bucket.

I was thinking of it as a grid - with each square measuring the amount of light (or photons if you wish) and each of these tiny squares makes up the picture.
My thought was controlling the exposure of each of these squares individually to give different exposure to different parts of the cell.

It would appear sensors don't, and probably can't work like that.

Shame

 

[Musicman]

Ah, ok. I wasnt thinking of the sensor as a bucket.

I was thinking of it as a grid - with each square measuring the amount of light (or photons if you wish) and each of these tiny squares makes up the picture.
My thought was controlling the exposure of each of these squares individually to give different exposure to different parts of the cell.

You're right about it being a grid: a grid of separate light buckets (sensor sites, AKA photodiodes).

The way to control the amount of light falling onto them is by use of a shutter,* but that acts across all of them at the same time.

You could expose the image for the highlights so that none are blown (no individual bucket fills up) and then increase the gain on the remaining ones, which would be to selectively bring up the shadows in the camera or PP, but, as above, you'll introduce noise into those areas.

* or ND/grad filters, etc.

 

[Faldrax]

Ah, ok. I wasnt thinking of the sensor as a bucket.

I was thinking of it as a grid - with each square measuring the amount of light (or photons if you wish) and each of these tiny squares makes up the picture.
My thought was controlling the exposure of each of these squares individually to give different exposure to different parts of the cell.

It would appear sensors don't, and probably can't work like that.

Shame

This is something I have thought about as well - and I suspect part of the answer might be 'currently'.

Sensors are getting better and better at handling a range of sensitivities - look at ISO 1600 images now compared to 3 years ago, for example. That's 4 stops (from a base 200), if they can get another couple of stops, and processors continue to increase in power, then having different regions at different sensitivities might be an option.

Still not what you need to replace a 10-stop, but good enough for a lot of ND Grads.