10 minutes ago, matt77hias said:

So radiance?

No. To be deliberately simplify -- let's say that radiance is what the sensor is measuring. Exposure would determine what the maximum measurement could possibly be.

The sensor isn't measuring radiance really though - Radiance is energy (emitted or received) per angle, per area, per time. Shutter speed increases time. Aperture increases angle. ISO increases efficiency (the percentage of energy actually captured and not lost as heat, reflected, etc). Exposure is the combination of these three -- so increasing exposure actually allows the same radiance input to act over a longer period of time / larger angle, which allows more energy to be delivered to the sensor.

You can have a very high exposure value (large shutter time, large aperture, large ISO value), but if you don't shine any light (radiance) through the lens, the sensor won't report any values

Likewise if you shine a constant radiance value through the lens, but vary the exposure value, the sensor will report higher or lower energy values.

tl;dr - it's basically an arbitrary multiplier that you can use to make everything brighter or darker.

41 minutes ago, Hodgman said:

No. To be deliberately simplify -- let's say that radiance is what the sensor is measuring. Exposure would determine what the maximum measurement could possibly be.

Radiance is actual energy (emitted or received) per angle per area per time. Shutter speed increases time. Aperture increases angle. ISO increases the percentage of energy actually (not lost as heat, reflected, etc). Exposure is the combination of these three -- so increasing exposure actually allows the same radiance input to act over a longer period of time / larger angle, which allows more energy to be delivered to the sensor.

You can have a very high exposure value (large shutter time, large aperture, large ISO value), but if you don't shine any light (radiance) through the lens, the sensor won't report any values

Ah so the pixel intensity:

$$I_{j} = \int_{\Lambda} \int_{T} \int_{\mathcal{M}} h_{j}(\lambda, t, \mathrm{path}) L(\lambda, t, \mathrm{path}) \mathrm{d}\lambda \mathrm{d}t \mathrm{d}\mathrm{path}$$

Edit: Damn, I was hoping for MathJax support on GameDev.net.

$$I_{j}$$ -> intensity of pixel j

$$\Lambda$$ -> wavelength domain

$$T$$ -> time domain (so basically the shutter interval)

$$\mathcal{M}$$ -> the scene manifold (which includes the camera sensor)

$$h_{j}$$ -> response function of pixel j (which includes the sensitivity)

$$\mathrm{path}$$ -> some path connecting the camera point through the camera aperture via some possible zero-sequence of other points on the light

$$L$$ -> radiance

 and suddenly there was MathJax support (though I don't know the inline keyword for Gamdev.net). Probably a delay due to the server.