BeginnerLesson 4 of 206 min read

Filming with What You Have: Shape, Light, Sound and File Size

Use simple numbers for aspect ratios, file sizes and background noise to film well with a phone.

In this lesson you will

  • Read an aspect ratio
  • Estimate a file size and plan storage
  • Understand a decibel ratio for noise
  • Apply cheap fixes for light and sound

About 28 minutes

Good enough, for now

Kofi borrows his sister’s phone, props it against a pile of books on the balcony table, and films the basil. The picture is fine. The sound is awful: wind, traffic and a neighbour’s radio. He assumes he needs a better camera. He does not. Almost every beginner problem is about three cheap things: where the light comes from, how the sound is recorded, and what shape the picture is. This lesson covers the numbers behind each, so that you can make good choices with the gear you already own.

The shape of the picture

Video comes in shapes called aspect ratios, which are the width divided by the height, written like 16:9. The program finds the ratio from a few common sizes by dividing both numbers by their greatest common divisor, and it prints the size in megapixels, which is the number of pixels in millions.

import math, random
from math import gcd
sizes = [(1920, 1080), (1280, 720), (3840, 2160), (1080, 1920), (1080, 1080)]
for w, h in sizes:
    g = gcd(w, h)
    print(str(w) + " x " + str(h) + ": " + str(w // g) + ":" + str(h // g) + ", " + str(round(w * h / 1e6, 1)) + " MP")
minutes, mbps = 8, 12
size_mb = mbps * minutes * 60 / 8
print("an", minutes, "minute video at", mbps, "Mbps:", round(size_mb), "MB")
raw_hours = minutes * 3 / 60
month = 8 * size_mb * 3 / 1000
print("8 videos with 3x raw footage:", round(month, 1), "GB")
def tone_and_hiss(noise, seed):
    r = random.Random(seed)
    signal = [0.3 * math.sin(2 * math.pi * 200 * i / 8000) for i in range(8000)]
    hiss = [r.uniform(-noise, noise) for _ in range(8000)]
    return signal, hiss
def rms(xs):
    return math.sqrt(sum(x * x for x in xs) / len(xs))
for label, noise, seed in (("bedroom", 0.01, 1), ("kitchen", 0.05, 2)):
    sig, hiss = tone_and_hiss(noise, seed)
    snr = 20 * math.log10(rms(sig) / rms(hiss))
    print(label + ": voice-to-noise " + str(round(snr)) + " dB")
print("doubling the level adds:", round(20 * math.log10(2), 1), "dB")
1920 x 1080: 16:9, 2.1 MP
1280 x 720: 16:9, 0.9 MP
3840 x 2160: 16:9, 8.3 MP
1080 x 1920: 9:16, 2.1 MP
1080 x 1080: 1:1, 1.2 MP
an 8 minute video at 12 Mbps: 720 MB
8 videos with 3x raw footage: 17.3 GB
bedroom: voice-to-noise 31 dB
kitchen: voice-to-noise 17 dB
doubling the level adds: 6.0 dB

Three of the sizes reduce to the same shape: 16:9, the wide shape of a normal video, whether it is 1280 pixels across or 3840. A size of 1080 by 1920 is its sideways twin, 9:16, the tall shape of a phone held upright. And 1080 by 1080 is 1:1, a square. The choice of shape depends on where people will watch. Wide suits normal videos on a computer, a tablet or a television. Tall suits short videos made for phones, as lesson 7 shows. Do not mix: a wide video shown in a tall frame gets black bars, and a tall video in a wide frame does too.

How big do the files get?

A video file is a stream of data, and its size is simple arithmetic: the rate in megabits a second, times the length in seconds, divided by eight to turn bits into bytes. At a made-up rate of 12 megabits a second, an 8 minute video comes to about 720 megabytes. Real rates depend on your camera or app. If you shoot three times as much footage as you keep, a common habit, eight finished videos need about 17.3 gigabytes. That is more than many phones can spare, so plan to copy footage to a computer or drive and clear the phone. Before you start a project, check your free storage, not after the phone says it is full.

Light

  • Face the window, not the back to it. Light from behind turns you into a shadow.
  • Soft is kinder than hard. A window on a cloudy day gives soft light. Direct sun makes strong shadows.
  • Keep the light steady. If clouds come and go, the picture will flicker in brightness. Close a curtain or choose a time of day when the light is even.

Sound is more than half

Viewers forgive a plain picture. They rarely forgive bad sound. The enemy is noise: any sound that is not your voice. You can measure how much of it there is. The decibel is a unit that compares two sound levels on a scale that matches how we hear. A useful measure is the voice-to-noise ratio, the level of your voice compared with the level of the background, in decibels. A higher number is cleaner.

The program builds a pretend voice, a pure tone, and adds different amounts of random hiss, standing in for a quiet bedroom and a noisy kitchen. The bedroom comes out at 31 decibels and the kitchen at 17. The numbers are for a synthetic signal, so they are not what you would measure on a real microphone, but the lesson carries over. A gap of fourteen decibels is large: the kitchen hiss is about five times the amplitude of the bedroom’s. The rule of thumb in the last line is useful: every doubling of amplitude adds about six decibels.

Cheap fixes that work

  1. Move closer to the microphone. Halving the distance raises your voice relative to the noise more than any gadget.
  2. Choose a quiet, soft room. Curtains, carpets and bookshelves soak up echo. A wardrobe full of clothes is a classic studio.
  3. Turn off what hums. Fans, fridges and radios are heard on a recording even when you stop noticing them.
  4. Record ten seconds of silence in the room, and listen to it with headphones before you film.
  5. Use a simple external microphone only after you have tried the free fixes.
Common mistake: wind and distance

Filming outside on a windy day with the phone’s built-in microphone. Wind noise cannot be removed afterwards. Film where it is still, or add a basic cover and film close.

Made-up sound

The sizes, bit rate and synthetic sound are made up. Real bit rates, file sizes and levels depend on your device and settings. The decibel numbers show the idea and are not a measurement of a real room.

Record ten seconds of silence in two rooms of your home. Listen with headphones and decide which one you would film in.

Practice for this lesson

Check what you learned

Five quick questions. Pick an answer and you will see straight away whether it is right and why.

What does an aspect ratio of 16:9 describe?

  1. The tall shape of a phone video
  2. A square
  3. The wide shape of a normal video
  4. The file size

A tall video reduces to 9:16.

How is a video file size estimated?

  1. Width plus height
  2. Frames minus seconds
  3. Pixels times colours only
  4. Rate times seconds, divided by eight

Dividing by eight turns bits into bytes.

Why plan storage before filming?

  1. Phones cannot film in wide shape
  2. Footage takes far more space than finished videos
  3. Storage changes the aspect ratio
  4. Raw footage is always small

Shooting three times what you keep is common.

What does a higher voice-to-noise ratio mean?

  1. The room is noisier
  2. The file is bigger
  3. Your voice is louder compared with the background
  4. The picture is wider

A higher number is cleaner.

Which is the cheapest big improvement to sound?

  1. Move closer to the microphone in a quiet, soft room
  2. Buy the most expensive camera
  3. Film with the window behind you
  4. Add music over the noise

Distance and room matter more than gadgets.

Summary

  • Aspect ratio is width divided by height: wide, tall or square.
  • File size is rate times seconds divided by eight.
  • Sound matters more than picture; reduce noise at the source.
  • Move close, pick a soft quiet room, and face the light.

Next lesson

Lesson 5 covers editing: cuts, silences and the pace of a video.

Questions

Do I need a microphone?

Not at first. Test the phone in a quiet room and close to your mouth. If the recording is clear, you may not need one. Buy one only when the free fixes have not solved the problem.

Should I film in the highest quality?

A good quality is worth it, but check your storage and editing speed first. A sharp, steady, well-lit video at a moderate quality beats a blurry one at the highest setting.

Can bad sound be fixed in editing?

A little, such as lowering a steady hum. Wind, echo and loud noise are hard or impossible to remove. It is always cheaper to record clean sound.

Progress is saved in this browser.