Virtual Laboratory on Television Principles

By the end of this laboratory, the student should be able to:

• Explain the phenomenon of persistence of vision and its role in smooth motion portrayal in television.
• Measure and interpret the critical flicker fusion (CFF) frequency and justify the choice of field frequency in TV standards.
• Demonstrate and quantify line merging (insufficient vertical sampling) and relate it to the number of scanning lines.
• Analyse how resolution, bandwidth and noise jointly determine picture quality.
• Determine the Kell factor and explain why the effective vertical resolution is less than the number of scanning lines.
• Write a professional engineering laboratory report following the prescribed structure.

1. Image Capture and Scanning

A television camera converts an optical image into an electrical signal. The image is dissected by an electron beam that scans the photosensitive target in a sequence of parallel horizontal lines, starting at the top-left and finishing at the bottom-right — a process called raster scanning. Each complete scan of the whole picture is a frame. In interlaced scanning, each frame is split into two fields: field 1 scans the odd lines (1, 3, 5 …), field 2 scans the even lines (2, 4, 6 …). The fields are presented in rapid succession and the eye integrates them into one complete frame.

2. Persistence of Vision

The human retina retains an image for a short time after the stimulus is removed — approximately 1/16 s. If successive still frames are presented faster than the decay of this retinal impression, the brain fuses them into continuous motion. This is the basis of all motion-picture and television systems: a sequence of still frames is displayed rapidly enough that the gaps between them are invisible.

3. Flicker and the Critical Fusion Frequency

When a picture is refreshed at a low rate, the screen brightness visibly pulses — this is flicker. The Critical Flicker Fusion (CFF) frequency is the minimum refresh rate at which flicker becomes imperceptible. For the human eye at normal brightness the CFF is about 45–55 Hz, rising with brightness (Ferry–Porter law). If the whole frame were refreshed at 25 Hz (as in early mechanical TV), flicker would be intolerable. Interlacing solves this: the field rate is doubled (e.g. 50 Hz) while the frame rate stays at 25 Hz, refreshing each screen area 50 times per second with only half the video bandwidth.

4. Line Merging (Insufficient Vertical Sampling)

The picture is sampled at a finite number of horizontal scanning lines. Fine vertical detail (closely spaced horizontal stripes) in the scene must satisfy the sampling theorem: the number of scanning lines must be at least twice the number of stripe-pairs (lines) per picture height. When this condition is violated, the fine stripes cannot be resolved and merge into a uniform grey area — an effect called line merging (the spatial analogue of aliasing).

5. Picture Quality, Resolution and Bandwidth

Picture quality is governed mainly by: (i) the number of scanning lines (vertical resolution); (ii) the video bandwidth, which limits how fast the signal can change along a line, i.e. the horizontal resolution; (iii) the signal-to-noise ratio. For a picture of aspect ratio 4:3 and N scanning lines, the highest video frequency occurs when alternate picture elements along a line are black and white:

For 625 lines, 25 fps and Kell factor 0.7 this gives ≈ 5.5 MHz — the standard TV channel video bandwidth.

6. The Kell Factor

The effective vertical resolution of a TV system is less than the number of scanning lines N because the scanning lines do not in general fall exactly on the picture detail. Depending on the phase relationship between the raster and the scene, a line straddles the boundary between a dark and a bright area, producing an average instead of a clean sample. Measurements show that the maximum number of resolvable horizontal lines is roughly 70 % of N:

The Kell factor also applies horizontally (bandwidth-limited resolution) and explains why a 625-line system resolves only about 400–440 lines of vertical detail in practice.

Theory of the Experiment

The eye retains each image on the retina for about 1/16 s after it disappears. A motion-picture or TV system presents a rapid sequence of still frames; if the frame rate is high enough, successive frames overlap on the retina and the brain perceives smooth continuous motion instead of a jerky slideshow. Below the fusion rate the motion appears jerky (each frame is seen separately); above it, motion appears continuous. In this experiment you vary the display frame rate and the retinal persistence time, and observe the transition from discrete frames to smooth motion.

Procedure

1. Set the wheel speed to 1.0 and the persistence time τ to 62 ms (typical human eye).
2. Set the frame rate to its minimum (2 fps) and observe both displays. Note that the perceived display shows several separated wheels — each frame is clearly seen individually and motion is jerky.
3. Increase the frame rate in steps of 2 fps. At each setting record the frame period T, compute T/τ, and note whether the perceived motion is smooth or jerky.
4. Identify the lowest frame rate at which the motion appears smooth. Compare it with the theoretical value 1/τ.
5. Reduce the persistence time to 30 ms (as in bright light or for a peripheral/dark-adapted observer) and repeat steps 2–4. Record how the required frame rate changes.
6. Increase the persistence time to 100 ms and repeat. Explain the result using the decay of the retinal impression.
7. Sketch the perceived display at a low and a high frame rate in your report and annotate the retinal trails.
8. Answer the discussion questions and write conclusions (see Report Guidelines).

Theory of the Experiment

When the screen brightness is modulated at a low frequency, the eye perceives pulsation — flicker. As the modulation frequency is raised, the pulsation disappears at the Critical Flicker Fusion (CFF) frequency, typically 45–55 Hz at normal brightness, because the retinal response cannot follow the rapid change and the average brightness is perceived instead. CFF increases with screen brightness (Ferry–Porter law) and with the fraction of the visual field stimulated. Television standards therefore refresh each part of the screen at 50 or 60 Hz (field rate) even though the frame rate is only 25/30 Hz — this is exactly why interlacing was invented: it doubles the flicker rate without doubling the video bandwidth.

Procedure

1. Set the frequency to 5 Hz, brightness to 80 % and duty cycle to 50 %. Observe that the bar flashes on and off — flicker is obvious.
2. Raise the frequency slowly (in steps of about 5 Hz). Record, for each setting, whether flicker is visible and note the retinal ripple amplitude on the graph.
3. Determine your measured CFF — the frequency at which flicker just disappears. Repeat three times and average.
4. Increase the peak brightness to 100 % and repeat the measurement. Explain any change using the Ferry–Porter law.
5. Change the duty cycle to 20 % (narrow bright flashes) and repeat. Comment on why a short, bright flash is more visible than a dim long one at the same frequency.
6. With the ripple plot, estimate the eye's effective time constant from the decay envelope after each flash.
7. Using your results, justify the choice of 50 Hz field rate for the 625-line/25-frame PAL system and explain how interlacing achieves it without increasing bandwidth.

Theory of the Experiment

A TV picture is sampled vertically by a finite set of N horizontal scanning lines. Fine horizontal stripes in the scene (detail that varies rapidly in the vertical direction) are only reproduced if the sampling theorem is satisfied: the number of scanning lines N must be at least twice the number of stripe-pairs per picture height. When the stripes are finer than this, successive samples straddle whole stripe-pairs and the detail merges into a uniform average brightness — this is line merging, the spatial equivalent of aliasing. In this experiment a test chart of horizontal line-pairs of adjustable fineness is scanned with an adjustable number of lines; you observe merging at low N and clean resolution at high N.

Procedure

1. Set N = 150 scanning lines and the detail fineness to 30 line-pairs per picture height. Observe that the scanned image reproduces the stripes faithfully.
2. Raise the fineness in steps of 10, keeping N = 150. At each step compare the scanned image with the original and record the highest fineness that is still reproduced correctly.
3. Verify that the observed limit agrees with the theoretical Nyquist limit N/2 = 75 line-pairs per picture height.
4. Beyond the limit, describe the appearance of the pattern: merging into uniform bars, moiré beats, and incorrect (aliased) coarse patterns. Record the apparent frequency of any moiré beat and compare it with |f_detail − N/2|.
5. Repeat for N = 300 and N = 600 lines. Tabulate measured and theoretical resolution limits.
6. At a fixed fineness of 200 line-pairs, decrease N from 600 downwards. Record the value of N at which merging first appears and compare with 2 × fineness.
7. Discuss why the 625-line system cannot resolve detail finer than about 300 line-pairs vertically, and how this motivated the choice of line standards.

Theory of the Experiment

Picture quality is the subjective impression of sharpness, cleanliness and naturalness of a reproduced image. It is determined objectively by three main factors: (i) vertical resolution — the number of scanning lines N; (ii) horizontal resolution — limited by the video channel bandwidth, which sets how many picture elements (pixels) can change along each line; and (iii) noise — random amplitude fluctuations that degrade the signal-to-noise ratio. The video bandwidth needed for a system of N lines, aspect ratio 4/3, frame rate f_v and Kell factor k is:

In this experiment a test scene (a resolution wedge and portrait-like pattern) is scanned with adjustable line count, bandwidth (via horizontal low-pass filtering) and added noise. A composite Picture Quality Index is computed from the effective resolution and SNR, mimicking subjective quality rating.

Procedure

1. Set N = 625 lines, bandwidth = 5.5 MHz, noise = 40 dB (the PAL System I operating point). Observe the reproduced picture and record the Picture Quality Index.
2. Vary the bandwidth from 10 MHz down to 1 MHz in steps of 1 MHz, keeping N and noise constant. At each step record the horizontal resolution and PQI. Plot PQI versus bandwidth.
3. Determine the minimum bandwidth that gives "Excellent" quality. Compare with the theoretical 5.5 MHz and explain any difference using the Kell factor and blanking margins.
4. Restore bandwidth to 5.5 MHz. Now vary the number of lines from 1080 down to 180 in steps. Record PQI at each setting and plot PQI versus N. Explain the approximately square-law relationship between bandwidth need and N.
5. Set N = 625 and bandwidth = 5.5 MHz. Reduce the SNR from 50 dB to 0 dB in steps of 10 dB. Record PQI and describe the visible degradation (snow, loss of detail, loss of sync).
6. Find the SNR at which the quality rating drops below "Good". Compare with broadcast practice (typically ≥ 40 dB for good reception).
7. Investigate one trade-off of your own: e.g. halve N and double bandwidth, or vice-versa, and record PQI. Conclude which factor dominates quality.
8. Tabulate all results, plot the graphs, and write your conclusions (see Report Guidelines).

Theory of the Experiment

Vertical resolution would equal the number of scanning lines N if every line landed squarely on a picture element. In practice the scene detail and the raster have an arbitrary phase relationship: a scanning line may straddle the boundary between a dark stripe and a bright stripe, so the camera outputs an average instead of a clean sample. Only when the raster is perfectly aligned with the detail (the lucky "in-phase" case) is the full Nyquist limit N/2 line-pairs reached. The average over all phases — the effective vertical resolution — is found empirically to be about 0.7 N. This constant, k_Kell ≈ 0.7, is the Kell factor. In this experiment a resolution test pattern of horizontal line-pairs is scanned with N lines whose phase is stepped through all possible offsets; the number of distinguishable line-pairs (and the implied Kell factor) is measured for each phase and averaged.

Procedure

1. Set N = 400 lines and phase = 0. Observe the wedge: the finest resolved line-pair corresponds to the resolvable TV lines. Record this value and the implied Kell factor k = TV-lines / N.
2. Step the phase from 0 to 1.0 in increments of 0.05. At each step record the resolvable TV lines and k. Note that at some phases (lines aligned with stripes) resolution is high, and at others (lines straddling stripes) it is low.
3. Click Auto-sweep to scan all phases automatically; record the average Kell factor.
4. Repeat for N = 200 and N = 800. Show that the average k is approximately independent of N (≈ 0.7), confirming that Kell factor is a property of the sampling geometry, not the line count.
5. Compute the effective vertical resolution of a 625-line system using your measured k, and compare with the 405–440 TV lines quoted in practice.
6. Using k = 0.7, compute the video bandwidth of the 625-line/25 fps system and compare with the 5.5 MHz standard.
7. Discuss the assumptions of the experiment (sharp spot, no interlace interaction, sinusoidal MTF) and why real measured Kell factors range 0.65–0.75.

Each student must submit an individual report for every experiment. Reports must be type-written, in English, and submitted as a single PDF. Use A4 paper, 12-pt Times New Roman, 1.5 line spacing, justified alignment, and number all pages.

Required Structure

Quality Standards

• Report the units of every quantity and quote results to an appropriate number of significant figures.
• Where a comparison with theory is possible, compute the percentage error:  %error = |measured − theoretical| / theoretical × 100%.
• Every figure and table must be numbered and referred to in the text (e.g. "…as shown in Figure 2").
• Discussion questions at the end of each experiment are compulsory; reports without them lose marks.
• Reports must be your own work. Plagiarism and copied data attract a zero mark and disciplinary action.
• Length guideline: 6–10 pages per experiment, excluding the title page.