All Things Color for Film and Digital Cinema
Subscribe | Log in

Posts Tagged ‘Histogram’

Video Waveform Visualization and Creation

As an addition to the prior post on waveforms here we examine a more visually descriptive way of showing how a video waveform is created and therefor how it can be interpreted.

Fig:1 Waveform and Live Image

Above we can see a standard Live Video Feed and its represented waveform.

Below I have created a 3Dimentional Luminance map of the same live feed. In this representation every pixel on the image is represented on the z-axis as a luminance value bw 0 and 1023 video code values. The ‘lighter’ or more intense the pixel value the higher it appears on the z-axis.

Fig: 2 Luminance Relief Map  Face On

As I rotate this image in 3D space, around the y-axis, you can see the luminance values more clearly represented as a Height map in the 3 Dimensional space.

Fig: 3 Luminance Relief Map Rotated 45 Degrees

When finally rotated a full 90 Degrees we end up with the Waveform interpretation.

Fig: 4 Luminance Relief Map Rotated 90 Degrees

Adrian Hauser


Waveform Vs Histogram interpretation in Digital Cinema Cameras

Understanding how to read Histograms present on many new digital cinema cameras can be tricky and are easily misenterpreted.

For analysis, I will use the below still from the film Daybreakers as reference.

Ethan Hawke in "Daybreakers" (2010)

The following snapshots have been taken using DpxRead available on the Panavision website.

Immediately we can see that there is a massive difference in the way these two images are represented in each of the graphical/statistical graphs.

Histograms represent the volume/percentage of light levels exposed within a particular image. The resulting graph shows the distribution/intensity plot of those levels.

Histogram Exposure

The above image reference image  is quite ‘moody’. The histogram shows us this quite literally but surprisingly shows nothing of where the midtones sit. This is because Histograms work with percentages and Ratios of light. If for the most part an image is dark, say 60% of the overall area , then the rest of the histogram has to be interpreted with the remaining 40 percent of image area. For that reason the intensity represented by the histogram for the remaining light values is visually a lot lower than the Dark spike shown in our reference.

To better show the way a histogram graphs an images lightness values I have put the below gradient into the scopes.

Linear Grayscale Ramp

We can see that because there is an even amount of each light level within the gradient image the Histogram shows an even intensity/distribution of each ‘level’.

Video Waveforms on the other hand give us a lot more visual information with which to evaluate your exposure and contrast ratio. As well as showing us the distribution of light values the graph is also plotted across the horizontal plane of the image. With this additional ‘axis’ one can easily determine where within the frame a particular item sits in its digital exposure value. This makes it easy to find for example the exposure of someones skin tone in relation to the background subject matter.

Histograms are Cheap and Easy to display from a programmers POV but in my opinion are quite useless in representing photographic content and should not be used for indepth exposure analysis.

Adrian