CP-lecture06-Exposure and High Dynamic Range Imaging

Exposure and high dynamic range imaging

Overview

• Exposure control.
• Light metering.
• Our devices do not match the world.
• High dynamic range imaging.
• Radiometric calibration.
• Other aspects of HDR imaging.
• Tonemapping.

Exposure control

$$Exposure=Gain\times Irradiance\times Time$$

• Shutter speed -> Exposure time
  • Motion deblurring
• Aperture -> Irradiance
  • Aperture also controls size of circle of confusion for out of focus objects
  • Sharp depth of field (“bokeh”) is often desirable, and form of bokeh is determined by shape of aperture
  • Lens “Speed”
    • A “fast” lens is one that has a very large max aperture.
• ISO -> Gain
  • Side effect: Image becomes very grainy because noise is amplified
  • ISO is not an acronym, It is pronounced (roughly) eye-zo

Light metering

SLR cameras use a separate low-resolution sensor that is placed at the focusing screen

the world has a high dynamic range, but the sensors have a low dynamic range, and images have an even lower dynamic range

Two challenges:

1. HDR imaging – which parts of the world do we include in the 8-14 bits available to our device?
2. Tonemapping – which parts of the world do we display in the 4-10 bits available to our device?

High Dynamic Range Imaging

key idea:

1. Exposure Bracketing: capture multiple LDR images at different exposures
2. Merging: Combine them into a single HDR image

Ways to vary Exposure

• Shutter speed
  • shutter times usually obey a power series, say 1/4, 1/8, 1/16, 1/32, 1/64, 1/128, 1/256, 1/512, 1/1024 sec
• F-stop
• ISO
• Neutral density(ND) filters

RAW image formation model
$$I_{\text{linear}}(x,y)=\text{clip}[t_i\cdot L(x,y)+\text{noise}]$$
where $t_i$ is the exposure time.

For each pixel:

1. Find “valid” images
2. Weight valid pixel values appropriately
3. Form a new pixel value as the weighted average of valid pixel values

Radiometric calibration

Measuring the camera’s response curve
Merging non-linear exposure stacks(series of image with different exposure time): use inverse transform to estimate linear image
If you can’t measure the response curve, take a look at the image’s EXIF data, which often containing relevant information like the Tone Reproduction Curve. If nothing else, take the square of your image to approximately remove effect of tone reproduction curve.

Other aspects of HDR imaging

The final fused HDR image gives relative radiance. If we know exact radiance at one point, we can convert relative HDR image to bsolute radiance map
Problem with basic HDR approach: very sensitive to movement. So the modern automatic HDR solutions oinclude an alignment step before merging exposures
Store HDR images:

• most standard image formats - integer 8 bits
• some image formats - integer 12/16 bits
• HDR images - floating point 32/64 bits
  Formats: portable float map(.pfm), radiance format(.hdr), OpenEXR format(.exr)

Tonemapping