Sunday, 4 October 2026

How Do You Film Something That Happens Too Quickly to See?

 


How Do You Film Something That Happens Too Quickly to See?

Sometimes the camera does not merely record an event — it reveals one.

Most of the time, we think of a camera as a device for preserving something we have already seen.

A wedding ceremony happens, and we film it.

Someone gives a presentation, and we record it.

A musician performs, and the cameras capture the performance.

But there is another type of filmmaking that I find particularly fascinating.

What happens when the event takes place too quickly for us to see properly in the first place?

A balloon bursts.

A ball hits the ground.

A droplet strikes the surface of water.

A ball falling.



A golf club, tennis racket or cricket bat strikes a ball.

A machine operates at high speed.

A piece of scientific apparatus oscillates.

A sailing dinghy hits a wave and throws spray into the air.

To our eyes, these events may appear almost instantaneous.

Point the right camera at them, however, and an event lasting a fraction of a second can be stretched into several seconds of detailed movement.

Suddenly, we are not simply recording what happened.

We are discovering how it happened.


Our Eyes Are Remarkable — But They Have Limits

Human vision is extraordinarily good at detecting movement.

It is less good at examining the individual stages of something that happens in a tiny fraction of a second.

Imagine dropping a rubber ball onto a hard surface.

We see:

fall → impact → bounce

But that simple description hides an enormous amount of information.

During the impact the ball may:

  • deform;

  • flatten;

  • store elastic energy;

  • change direction;

  • recover its shape;

  • begin accelerating upwards.

Much of that may happen in only a few milliseconds.

Film the same event at a sufficiently high frame rate and suddenly the impact becomes visible.

The ball does not simply "bounce".

We can actually watch the physics of the bounce taking place.

That is where high-frame-rate video becomes such a useful filmmaking tool.


What Does Frame Rate Actually Mean?

Video is really a sequence of individual images displayed rapidly enough to create the impression of continuous movement.

A conventional video might be recorded at:

25 frames per second (fps)

That means the camera records 25 individual images every second.

Other common recording rates include:

  • 24 fps;

  • 25 fps;

  • 30 fps;

  • 50 fps;

  • 60 fps.

But cameras capable of slow-motion recording may offer:

  • 100 fps;

  • 120 fps;

  • 200 fps;

  • 240 fps;

and specialist high-speed cameras can go vastly beyond this.

The important point is simple:

the more frames we capture during the event, the more individual moments we have available to examine afterwards.

Suppose I record something at 100 fps and play it back at 25 fps.

Each second of real action contains 100 recorded frames.

At 25 frames per second, those frames take four seconds to display.

So:

Slow-motion factor = recording frame rate / playback frame rate

In this example:

100 / 25 = 4

The action therefore appears four times slower.

Record at 200 fps and play back at 25 fps and we obtain:

200 / 25 = 8

The action can now appear eight times slower.

This is the basic principle behind slow-motion filmmaking.


The Bursting Balloon Experiment

A bursting balloon is an excellent demonstration because our everyday experience of it is almost entirely dominated by the sound.

BANG!

The balloon appears to vanish.

But that is not what actually happens.

With high-speed video, it may be possible to observe the rubber beginning to tear and then rapidly peeling away as the stored elastic energy is released.

The contents of the balloon behave differently again.

Fill the balloon with water and the result can be particularly spectacular.

For a very short time, the water may retain approximately the shape of the balloon even though the rubber that contained it is disappearing.

Gravity and surface tension then take over and the mass of water collapses.

At normal speed:

pop — splash.

In slow motion:

a sequence of physical processes becomes visible.

This is a wonderful example of the difference between simply making an attractive video and using video as an observational tool.


A Bouncing Ball Can Become a Physics Experiment

Another beautifully simple subject is a bouncing ball.

I could place a camera low down, close to the floor, and arrange the shot so that the point of impact fills a significant part of the frame.

At normal speed, the impact is difficult to study.

Slow it down and we can begin asking questions.

How much does the ball deform?

How long is it in contact with the floor?

Does a tennis ball behave differently from a squash ball?

What about a table-tennis ball?

Or a solid rubber ball?

Now the filming has become an experiment.

Add measurements and it becomes even more interesting.

If the ball is dropped from height h1 and rebounds to height h2, we can compare the rebound.

A simple measure is:

Rebound percentage = (h2 / h1) x 100%

We could repeat the experiment with different balls, different surfaces or different temperatures.

The camera has become part of the measuring equipment.


Water Looks Completely Different in Slow Motion

Water is one of my favourite subjects for this sort of filming because its behaviour is simultaneously familiar and extraordinarily complicated.

Drop something into a bowl of water and we see a splash.

Film it closely and slowly and there is much more happening.

We might see:

  • the initial depression of the surface;

  • a crown of droplets forming;

  • individual droplets separating;

  • a cavity developing;

  • water collapsing back towards the centre;

  • a vertical jet forming afterwards.

Even a single falling droplet can produce remarkable images.

A droplet hitting a thin layer of liquid may create a crown-like structure that exists for only a tiny fraction of a second.

The challenge is therefore not simply recording at a high frame rate.

We also need to think about magnification, focus, exposure and lighting.


Slow Motion Needs Light — Lots of It

This is one of the practical problems that is easily overlooked.

If I increase the recording frame rate, every individual frame exists for a shorter period.

That generally means there is less time for light to reach the camera sensor.

If the shutter speed also needs to be fast enough to freeze motion, the problem becomes even greater.

The result?

A high-speed shot that looked easy in your imagination can suddenly become very dark.

The temptation is to increase ISO dramatically.

That may work, but there is a price: increasing amplification can make noise more visible and reduce image quality.

So one of the secrets of good high-frame-rate filming is often surprisingly simple:

provide more light.

That might mean:

  • brighter continuous lighting;

  • moving lights closer to the subject;

  • using a wider aperture;

  • choosing a faster lens;

  • increasing ISO carefully;

  • controlling ambient light.

The more extreme the slow motion, the more important lighting becomes.


And Then There Is Motion Blur

There is another interesting decision.

Do we want every frame to be razor sharp?

Or do we want some motion blur?

If a rapidly moving object travels a significant distance while each frame is being exposed, it will appear blurred.

A faster shutter speed reduces that blur.

For analytical filming, where I might want to identify the exact position of an object, a short exposure can be extremely useful.

For cinematic footage, however, removing all motion blur can make movement look rather harsh or unnatural.

So once again, there is no single "correct" camera setting.

It depends upon the purpose of the film.

Are we making a measurement, teaching a concept or creating a beautiful sequence?

Those can require quite different choices.


Getting Close Changes Everything

Slow motion becomes even more impressive when it is combined with close-up or macro filming.

Imagine filming a small mechanical switch.

At ordinary viewing distance, we see it operate.

Move much closer and slow the movement down and we may see:

  • components flexing;

  • springs compressing;

  • contacts moving;

  • vibration after impact.

The same principle can be applied to:

  • machinery;

  • tools;

  • manufacturing processes;

  • laboratory equipment;

  • sports equipment;

  • musical instruments;

  • moving mechanisms.

Sometimes the interesting part of an event is only a few millimetres across.

The solution is therefore not simply:

"Film it slower."

It may be:

"Film it slower, closer and with better lighting."


Focus Becomes Critical

Close-up filming introduces another problem.

Depth of field becomes very small.

A moving object may be sharp at one point and blurred a few millimetres later because it has moved outside the plane of focus.

Autofocus can sometimes help, but for predictable experiments I often prefer the certainty of planning the shot carefully.

One useful technique is to place an object temporarily at the exact position where the important event will occur.

Focus on that point.

Lock the focus.

Remove the temporary target.

Then perform the experiment.

The event happens exactly where the camera is expecting it.

This is particularly useful when the interesting moment happens too quickly for a camera's autofocus system to react meaningfully.


The Camera Needs to Know Where the Action Will Be

This highlights one of the major differences between ordinary filming and high-speed filming.

With an interview, I can react to the person.

With an event lasting 1/100th of a second, I cannot.

Everything has to be anticipated.

Where will the balloon burst?

Where will the ball strike the surface?

Where will the droplet land?

Where will the machine component move?

Where will the athlete's foot make contact?

Where will the spray appear?

Good slow-motion filming is therefore often as much about preparation as camera technology.


Why Multiple Cameras Can Be So Useful

A fast event may look completely different from different directions.

Suppose I were filming a ball striking a surface.

One camera could provide the main wide shot.

Another could be positioned close to the impact point.

A third might look along the surface.

A fourth could concentrate on a particular component.

Now the editor can move between:

context → action → detail → explanation.

This is particularly valuable for educational and demonstration videos.

The viewer first understands where something happened and then gets to see exactly what happened.

That is one reason I find multi-camera production so useful. Different cameras do not merely provide alternative pictures.

They can provide different kinds of information.


Slow Motion Is Not Just a Special Effect

Slow motion is frequently associated with dramatic filmmaking.

A runner crosses the finishing line.

Champagne sprays into the air.

A wave crashes over a boat.

A musician strikes a cymbal.

Used well, these shots can certainly look spectacular.

But slow motion can do something much more important.

It can explain.

Consider a sports coach examining someone's movement.

At normal speed, the action may look correct.

Slow it down and perhaps the foot lands differently from expected.

Perhaps the racket angle changes just before contact.

Perhaps the sailor moves their weight too late during a manoeuvre.

Perhaps the golf club face is not where the player imagined it was.

Video provides something memory cannot.

Evidence that can be replayed.


Science Education Is an Obvious Application

This is particularly useful in science teaching.

There are many experiments where students understand the theory but struggle to observe the important event.

High-speed video can help with subjects including:

  • collisions;

  • momentum;

  • projectile motion;

  • oscillations;

  • waves;

  • elasticity;

  • vibration;

  • fluid motion;

  • resonance.

Imagine demonstrating a collision between two dynamics carts.

Students see the collision.

But slow-motion footage allows us to examine the moment of contact.

Add data from force or motion sensors and something even more interesting becomes possible.

We can compare:

what the camera shows

with

what the sensors measure.

That turns a video into part of a much richer investigation.


Industrial and Engineering Filming

The same principle extends beyond education.

Imagine a company has a production process that involves something moving quickly.

At normal speed, everything may appear to be functioning correctly.

But perhaps there is an occasional problem.

A component bounces.

A package shifts.

A belt vibrates.

A mechanism fails to engage properly.

A liquid splashes at a particular stage.

A product is damaged during transfer.

High-frame-rate video may help reveal the sequence.

Of course, genuinely high-speed industrial analysis can require specialist cameras, specialist lighting and sometimes synchronised measurement equipment.

But the principle remains the same:

slow the event down until the process becomes understandable.

That footage can then have several uses.

It might support:

  • training;

  • technical explanation;

  • troubleshooting;

  • presentations;

  • product development;

  • marketing.


A Technical Film Can Also Be a Promotional Film

This is where things become particularly interesting for a filmmaking business.

The same footage that explains a process can also make excellent promotional material.

Imagine a manufacturer whose product contains a beautifully engineered mechanism.

Most customers will never see how it works.

A carefully produced sequence could begin with the complete product, move to a close-up and then show the mechanism operating in slow motion.

Graphics could identify important components.

A voice-over could explain the engineering.

The final sequence could return to the finished product.

Instead of simply telling customers:

"This is precision engineered."

the film demonstrates it.

That is much more powerful.


Ordinary Objects Can Become Extraordinary

One of the pleasures of photography and filmmaking is discovering that we do not always need exotic subjects.

A glass of water can be interesting.

A bouncing ball can be interesting.

A vibrating ruler can be interesting.

A bursting balloon can be fascinating.

The important thing is changing the way we look at them.

Macro photography changes our sense of scale.

Time-lapse photography compresses time.

High-speed photography stretches time.

Each technique gives us access to something that normal human observation struggles to provide.


A Practical Experiment I Would Try

A simple demonstration could use three cameras or three recording settings.

Film the same bouncing ball at:

25 fps

50 fps

100 or 120 fps

Keep the camera position and subject as similar as possible.

Then place the three clips next to one another during editing.

At 25 fps we see the event.

At 50 fps we begin to examine it.

At 100 or 120 fps, the deformation and recovery become much easier to study.

Then repeat the experiment with a much faster shutter speed and sufficient lighting.

The comparison would demonstrate that high-speed filming is not simply about pressing a "slow motion" button.

It is a combination of:

frame rate + shutter speed + lighting + lens + focus + composition + timing.


The Real Skill Is Knowing What We Want to Discover

Modern cameras make extraordinary technology accessible.

But owning a camera capable of recording 100 or 120 frames per second does not automatically produce an interesting film.

The first question should be:

What are we trying to see?

If I want to study a bouncing ball, the camera needs to concentrate on the impact.

If I want to analyse a machine, I need to know which part of the mechanism matters.

If I am filming sport, I need to anticipate the movement.

If I am making a promotional film, I need to decide which detail tells the strongest story about the product.

The equipment matters.

But the idea comes first.


From Filming an Event to Revealing It

There is something rather wonderful about pressing play on a piece of high-speed footage for the first time.

You know what happened.

You were standing there when it happened.

Yet suddenly you can see something you did not see while it was happening.

A ball deforms.

A droplet becomes a crown.

A mechanism flexes.

A sail shakes.

A spring oscillates.

A splash becomes a complex moving structure.

A fraction of a second becomes several seconds of information.

That is why high-frame-rate and close-up filming can be valuable for far more than spectacular social-media clips.

It can help businesses explain products.

It can help engineers understand processes.

It can help coaches analyse movement.

It can help teachers demonstrate science.

And it can help filmmakers create images that make people stop and look.

Because sometimes the most interesting thing to film is not something we have never seen before.

It is something we have seen hundreds of times — but have never really been able to see.

Sometimes the camera does not merely record an event. It reveals one.

#Filmmaking #SlowMotion #HighSpeedVideo #VideoProduction #ScienceEducation #IndustrialVideo #ProductVideo #MacroVideo #Photography #Videography #STEM #ScienceCommunication #PhilipMRussellLtd

Saturday, 3 October 2026

Can AI Plan My Working Week?

 


Can AI Plan My Working Week?

A diary tells you when things happen. Could AI tell you when they should happen?

For many small businesses, the scarcest resource is not equipment, premises or even money.

It is time.

That becomes particularly obvious when one business contains several very different activities.

In my own case, a working week might include:

  • GCSE and A-level tuition;

  • preparing lessons and practical experiments;

  • filming;

  • photography;

  • video editing;

  • producing graphics and social-media material;

  • making personalised products;

  • answering enquiries;

  • preparing invoices and other administration;

  • maintaining equipment;

  • developing websites and other online material;

  • sailing and sailing-club work;

  • writing articles such as this one.

Individually, none of these necessarily presents a scheduling problem.

Put them all into the same week and something interesting happens.

The diary may say that there are enough hours.

Reality may disagree.

So I decided to consider a rather different use for artificial intelligence.

Could AI actually plan my working week?

Not merely put appointments into a calendar, but decide when different types of work ought to happen.

That turns out to be a much more interesting problem.


A Calendar Is Not Really a Plan

Suppose my diary contains a tuition lesson at 4.00 pm and another at 6.00 pm.

The calendar knows about two appointments.

But what does the working day really contain?

The first lesson may require preparation.

Perhaps I want to set up a practical experiment beforehand.

Afterwards, I may need to make notes about the student's progress.

The second lesson might require completely different equipment or teaching material.

Meanwhile, there may be an email enquiry waiting for a reply, a video that needs editing and an article that needs publishing.

The calendar records the fixed events.

It doesn't necessarily understand all the work surrounding them.

That is the first important distinction:

An appointment is not the same thing as a workload.


Give AI a Deliberately Difficult Week

A useful experiment would be to give an AI system a fictional but realistic collection of jobs.

For example:

Monday

9.00–11.00 — edit a promotional video
11.30 — answer customer enquiries
1.00 — prepare an A-level Physics practical
4.00–5.00 — GCSE Maths tuition
5.30–6.30 — A-level Physics tuition
7.00–8.00 — GCSE Science tuition

Tuesday

Morning — filming
Afternoon — editing and administration
Evening — tuition

Wednesday

Photography job during the morning
Product-production work during the afternoon
Tuition from late afternoon onwards

Thursday

Write and publish blog material
Prepare practical work
Film demonstrations
Evening tuition

Friday

Complete outstanding editing
Accounts and invoices
Equipment preparation
Tuition

Then add the activities that don't fit conveniently into neat boxes:

"Allow time for sailing."

"Write several social-media posts."

"Back up this week's video footage."

"Prepare next week's lessons."

"Deal with new customer enquiries quickly."

"Allow some time for jobs that overrun."

Suddenly this is no longer a simple diary exercise.

It is a scheduling problem.


The First AI Timetable Might Be Terrible

This is where the experiment becomes interesting.

If I simply tell AI:

"Fit all of these jobs into my week."

I might receive a beautifully organised timetable.

It might also be completely impossible.

For example:

8.00–9.00 Write blog
9.00–10.30 Film experiment
10.30–11.30 Edit video
11.30–12.00 Administration
12.00–1.00 Photography
1.00–2.00 Prepare lessons
2.00–3.00 Product production
3.00–4.00 Answer emails
4.00–9.00 Tuition

It looks wonderfully efficient.

There is only one problem.

Nobody actually works like that.

Where is lunch?

Where is the time required to change the studio setup?

What happens when filming takes twenty minutes longer than expected?

When is the camera footage transferred?

When are batteries charged?

What happens if a parent telephones?

What if an experiment takes longer to prepare than expected?

And after five hours of tuition in the evening, am I really going to start another complicated piece of work?

A timetable can be mathematically possible while being humanly ridiculous.


AI Needs Constraints

The solution is not necessarily to abandon AI.

It is to give it better information.

Instead of saying:

"Plan my week."

I could say:

"Plan my week subject to the following constraints."

For example:

  • tuition appointments are fixed;

  • allow at least 15 minutes between online lessons;

  • allow longer gaps where equipment must be changed;

  • do not schedule demanding creative work after a long evening of teaching;

  • include lunch;

  • include preparation time before practical lessons;

  • allow time after lessons for recording notes;

  • keep some unscheduled time each day;

  • group similar activities where possible;

  • allow travel time for work away from home;

  • treat customer enquiries as relatively high priority;

  • don't assume every job will finish exactly on time.

Now the AI has something much closer to the real problem.

And this is an important lesson about using AI generally:

The quality of an AI plan depends enormously upon the constraints we give it.


Preparation Time Is Real Work

This is particularly important in education.

A one-hour lesson does not necessarily represent one hour of work.

Suppose I want to demonstrate Young's modulus experimentally during an A-level Physics lesson.

Before the student arrives, I may need to:

  • select the wire;

  • set up the apparatus;

  • check the measurement equipment;

  • make sure the experiment works;

  • prepare questions;

  • locate the appropriate notes;

  • perhaps set up cameras or a visualiser;

  • prepare extension material.

The diary says:

Physics lesson — 1 hour.

The real workload might be considerably longer.

The same applies to filming.

A 20-minute finished video is certainly not a 20-minute job.

There may be:

planning -> setup -> filming -> retakes -> transferring files -> editing -> graphics -> sound -> rendering -> uploading -> promotion.

One advantage of using AI for planning is that it can be instructed to recognise these hidden tasks.


Travel Time Is More Than Driving Time

Suppose I have a photography or filming job away from my normal workplace.

AI needs to know that a 10.00 am appointment ten miles away doesn't mean I can work normally until 9.59.

There may be equipment to load.

Then there is travel.

Parking.

Unloading.

Setting up.

And afterwards everything happens in reverse.

A two-hour filming appointment could occupy half a working day.

That is the difference between event duration and event footprint.

A good planning system needs to understand both.


What About Sailing?

This produces another interesting scheduling problem.

Some activities are flexible but dependent upon external conditions.

Sailing is a good example.

A conventional diary might say:

Wednesday afternoon — sailing.

But Wednesday might be wet with virtually no wind, while Thursday afternoon could be ideal.

So perhaps AI shouldn't always allocate the activity immediately.

It might instead reserve a flexible block and reconsider it as better information becomes available.

The same principle could apply to outdoor photography, drone work where permitted, exterior filming or jobs requiring good natural light.

This moves us from a static timetable towards a dynamic plan.


Priorities Matter More Than Filling Empty Spaces

One of the biggest mistakes in time management is assuming that every empty space needs filling.

Imagine three jobs:

A: Reply to a potential new tuition customer.
B: Reorganise some archived photographs.
C: Finish tomorrow's lesson preparation.

All three are legitimate work.

They are not equally urgent.

An AI planner therefore needs more than duration.

It needs some concept of:

urgency + importance + deadline + consequences of delay.

The customer enquiry may deserve an immediate response.

Tomorrow's lesson preparation has a definite deadline.

The photograph archive might wait until there is genuinely spare time.

This is where AI potentially becomes much more useful than an ordinary calendar.


Batch Similar Jobs Together

There is another optimisation that humans sometimes overlook.

Changing tasks has a cost.

If the video studio is already set up, it might make sense to record three short videos rather than record one, dismantle everything and rebuild the studio tomorrow.

Similarly, if I am producing personalised products, several jobs using the same equipment might sensibly be grouped together.

The same applies to:

  • photography;

  • invoicing;

  • emails;

  • lesson preparation;

  • blog writing;

  • video editing;

  • social-media scheduling.

AI could identify these clusters.

Instead of asking:

"Where is there an empty hour?"

it could ask:

"What other work fits naturally with what is already happening?"

That is potentially much more valuable.


The Problem of Overruns

Real work doesn't obey the calendar.

A lesson might lead to an important question that deserves another few minutes.

A filming sequence might require another take.

A computer might decide to install something at precisely the wrong moment.

An experiment might refuse to behave.

An urgent enquiry might arrive.

If every minute has already been allocated, one delay causes a cascade through the rest of the day.

A better AI-generated timetable would deliberately contain buffer time.

Perhaps a 60-minute job receives a 75-minute block.

Perhaps there is a spare 30-minute recovery period during the afternoon.

Perhaps Friday afternoon contains a larger block for unfinished work.

At first sight, this looks inefficient.

In reality, it may make the entire week more efficient because the timetable becomes resilient.


AI Should Understand Energy as Well as Time

There is another factor that calendars rarely record.

Not every hour is equivalent.

For many people, there are times when concentration is particularly good.

Those periods are valuable.

They should perhaps be used for work such as:

  • complex writing;

  • planning;

  • difficult editing;

  • problem solving;

  • financial decisions;

  • technical development.

Other tasks require less uninterrupted concentration.

Those might include:

  • routine administration;

  • file organisation;

  • preparing equipment;

  • uploading completed material;

  • straightforward production work.

So instead of treating the week as forty identical one-hour boxes, an AI planner could consider the type of attention each job requires.

That begins to look much more like real management.


Could AI Replan the Week?

This is perhaps the most exciting possibility.

Monday doesn't go according to plan.

A filming job overruns by 45 minutes.

One editing job isn't completed.

A new customer enquiry arrives.

An appointment is cancelled on Wednesday.

The traditional diary simply records these events.

AI could potentially ask:

What should change now?

The unfinished editing could move into Wednesday's cancelled appointment.

A low-priority administration job could move to Friday.

The new enquiry could be dealt with immediately.

Nothing important needs to disappear.

The plan simply changes.

That is potentially one of AI's greatest advantages.

It can keep replanning.


But Should AI Be Allowed to Control the Diary?

There is an important distinction between:

AI suggesting

and

AI deciding.

I would be very happy for AI to say:

"You have scheduled too much work on Thursday."

Or:

"If you move the editing to Tuesday morning, you can film three items consecutively on Thursday."

Or:

"You haven't allowed preparation time for Friday's practical lesson."

That is useful assistance.

I would be considerably less comfortable with an autonomous system deciding to cancel a lesson, rearrange a customer appointment or promise a delivery date without approval.

AI can analyse the workload.

The human should retain control of important commitments.


A Better Experiment: Give AI the Real Calendar

The obvious next stage would be to connect the planning system to the information that already exists.

Imagine AI being able, with appropriate permission, to examine:

  • the calendar;

  • incoming enquiries;

  • existing customer commitments;

  • project deadlines;

  • production jobs;

  • lesson schedules;

  • recurring administration;

  • task lists.

Then ask:

"Plan next week, but don't alter any appointments without asking me."

That is far more interesting than asking AI to produce a generic timetable.

It is beginning to behave like a virtual office manager.


A Possible AI-Generated Day

A realistic day might look more like this:

8.30–9.00 — Check urgent enquiries and day's commitments
9.00–10.30 — Concentrated video editing
10.30–10.45 — Buffer
10.45–11.45 — Film two short demonstrations while studio is configured
11.45–12.15 — Transfer footage and reset equipment
12.15–1.00 — Lunch
1.00–2.00 — Prepare afternoon lessons
2.00–2.45 — Administration and customer replies
2.45–3.15 — Buffer and tuition setup
3.15 onwards — Teaching schedule
After final lesson — Brief notes only; no major creative work

That looks less impressively packed than the earlier timetable.

And that is exactly why it might work.


The Objective Shouldn't Be 100% Utilisation

This may be the most important conclusion from the whole experiment.

If AI tries to maximise every available minute, it has probably misunderstood the task.

A business is not automatically more productive because every box in the diary contains something.

There needs to be capacity for:

  • thinking;

  • unexpected opportunities;

  • overruns;

  • customer enquiries;

  • equipment failures;

  • preparation;

  • rest;

  • simply doing something properly.

Perhaps the instruction to AI shouldn't be:

"Fit as much work as possible into my week."

Perhaps it should be:

"Create a week in which the important work gets completed reliably without making the timetable unrealistic."

Those are very different objectives.


AI as the Assistant, Not the Boss

I think this is where AI could become particularly valuable to a small business.

Not because it can somehow create more hours.

It cannot.

Instead, it can make the demands upon those hours more visible.

It can spot forgotten preparation.

It can question unrealistic assumptions.

It can group related jobs.

It can protect time for important work.

It can suggest priorities.

And when the inevitable disruption occurs, it can help rebuild the plan.

The final decision, however, remains human.

I know whether an extra half-hour with a student is worthwhile.

I know whether the weather makes sailing worth rearranging an afternoon for.

I know when a filming project needs another take rather than being declared "good enough".

And I know that sometimes the sensible decision is simply to stop.

That context matters.


So, Can AI Plan My Working Week?

Yes — but only if I tell it what a real working week actually looks like.

Simply feeding appointments into AI is unlikely to achieve much.

Give it preparation times, travel, priorities, deadlines, buffers, dependencies, preferred working patterns and realistic limits, however, and something much more useful begins to emerge.

The interesting future may not be an AI that runs my diary without me.

It may be an AI that looks at the diary and says:

"You have planned this as though nothing will go wrong."

And that might be exactly the warning a busy small-business owner needs.

A diary tells you when things happen. Perhaps the next generation of AI tools will help us decide when they should happen — and, just as importantly, when they shouldn't.

Friday, 2 October 2026

VCV Rack 2 — Building a Synthesiser That Doesn't Physically Exist

 


VCV Rack 2 — Building a Synthesiser That Doesn't Physically Exist

Imagine being given hundreds of electronic music modules and being told: build whatever instrument you want.

Not choose an instrument.

Not select a preset.

Build the instrument itself.

You decide what produces the sound. You decide what changes its pitch. You decide how long each note lasts, how bright it sounds, whether it moves around the stereo field and whether the next note is deliberately chosen or partly determined by chance.

You can even decide whether it needs a keyboard.

That is the wonderfully strange world of modular synthesis, and software such as VCV Rack 2 makes it possible to explore that world without filling an entire room with electronic equipment.

For me, this is one of the fascinating sides of creating music with computers. The computer is no longer simply pretending to be a piano, organ or conventional synthesiser.

It can become an electronic laboratory in which we construct instruments that may never have physically existed at all.


What Is a Modular Synthesiser?

Most electronic keyboards present themselves as finished instruments.

You switch them on, select a sound and play.

There may be enormous sophistication underneath, but the basic signal path has already been designed for you.

A modular synthesiser turns that idea around.

Instead of receiving a finished instrument, you receive the building blocks from which an instrument can be constructed.

Those building blocks might include:

  • oscillators;

  • filters;

  • amplifiers;

  • envelope generators;

  • low-frequency oscillators;

  • sequencers;

  • mixers;

  • effects;

  • clock generators;

  • random voltage generators;

  • logic modules;

  • control processors.

Initially, many of them do absolutely nothing.

That is an important point.

You can place an oscillator into an empty modular rack and it may be happily producing a waveform, but unless you connect it to something useful, you will hear nothing.

The instrument only begins to exist when we start making connections.

And that leads to perhaps the most interesting question in modular synthesis:

Why would I connect these two things together?


Starting With an Empty Rack

One of the best ways to understand VCV Rack is to start with almost nothing.

An empty rack can initially look rather intimidating.

Where are the sounds?

Where are the presets?

Where is the instrument?

The answer is:

We haven't built it yet.

So let us construct one.


Step 1 — We Need Something That Makes a Sound

Our first component can be an oscillator.

An oscillator repeatedly generates an electrical or mathematical waveform.

Common waveforms include:

  • sine;

  • triangle;

  • sawtooth;

  • square.

Each has a different harmonic structure and therefore a different sound.

A sine wave is comparatively pure.

A sawtooth contains many harmonics and can sound bright and buzzy.

A square wave has another distinctive harmonic structure.

Already we have choices.

But there is an immediate problem.

Connect the oscillator directly to the audio output and we may simply hear:

BEEEEEEEEEEP.

It does not stop.

That is because the oscillator does not necessarily know anything about musical notes. It is simply oscillating continuously.

We need to turn that raw electronic signal into an instrument.


Step 2 — Controlling the Loudness

We could next introduce a VCA — Voltage Controlled Amplifier.

Despite the name, in a synthesiser a VCA is commonly used to control the level of a signal.

Now we can potentially turn the oscillator up and down electronically.

But something has to tell the VCA when to become louder and quieter.

So we need another module.


Step 3 — Giving the Sound a Shape

Enter the envelope generator.

A common envelope uses four stages:

Attack — Decay — Sustain — Release

Often abbreviated to ADSR.

Imagine pressing a key.

Attack

How quickly does the sound reach its initial maximum level?

A percussive sound may have an extremely fast attack.

A slowly swelling string-like sound may have a much longer one.

Decay

After reaching its initial peak, how quickly does the sound fall towards its sustained level?

Sustain

What level does the sound remain at while the note continues?

Release

What happens when the note ends?

Does the sound disappear instantly?

Or does it fade away gradually?

This is where modular synthesis starts becoming particularly interesting.

The oscillator creates the sound.

The VCA controls its level.

The envelope controls the VCA.

One module is controlling another.

We are no longer merely generating sound.

We are constructing behaviour.


Step 4 — Now Give It Some Pitch

Next we need a way of deciding which notes the oscillator should play.

A conventional keyboard is one possibility.

But modular synthesis does not insist that music must start with somebody pressing a key.

We could use a sequencer instead.

Imagine programming eight notes:

C - E - G - A - G - E - D - C

The sequencer can repeatedly send control information telling the oscillator which pitch to produce.

We now need some timing information as well.

So we introduce a clock.

The clock advances the sequencer.

The sequencer controls the oscillator.

The oscillator produces the waveform.

An envelope shapes each note.

The envelope controls the VCA.

The VCA feeds the audio output.

Our empty rack has become a musical instrument.

But we have barely started.


Step 5 — Making the Sound More Interesting With a Filter

Suppose our sawtooth oscillator sounds rather harsh.

We can pass it through a filter.

One common example is a low-pass filter.

Very broadly, this allows lower frequencies through while progressively reducing frequencies above its cutoff region.

Turn the cutoff down and our bright sawtooth becomes darker and softer.

Turn it up and more of the harmonics return.

That is useful.

But turning a knob by hand is not the only option.

What if something else turned the knob for us?

This is where the idea of control voltage becomes central to modular synthesis.


Control Voltage — The Language Between Modules

In a physical modular synthesiser, voltages can be used to control parameters.

One voltage might determine pitch.

Another might control volume.

Another might change filter cutoff.

Another could alter the speed of an effect.

VCV Rack recreates this concept in software.

This means that instead of thinking:

"I will adjust the filter."

we can start asking:

"What else could adjust the filter for me?"

That question opens an enormous creative world.


Enter the LFO

An LFO — Low Frequency Oscillator — is another oscillator, but it normally operates much more slowly than an audio oscillator.

Rather than hearing its oscillation directly, we can use it to control something else.

Connect an LFO to filter cutoff and the filter can automatically open and close.

Connect it to pitch and we can create vibrato.

Connect it to amplitude and we can create tremolo.

Connect it to stereo position and the sound can move from side to side.

And there is nothing stopping us from using several LFOs.

One might slowly alter the filter.

Another could control pulse width.

Another could alter the speed of the first LFO.

Now things are becoming considerably more complicated.

And much more interesting.


What Happens If One Control Controls Another Control?

This is where modular synthesis begins moving beyond the architecture of many conventional instruments.

Suppose LFO 1 controls the filter.

Instead of running LFO 1 at a constant speed, LFO 2 could alter the frequency of LFO 1.

The filter movement itself now speeds up and slows down.

We could then use an envelope to control the depth of that modulation.

Or a sequencer could select different modulation amounts for different steps.

Or a random voltage could occasionally change something.

The possibilities multiply extremely quickly.

A modular synthesiser therefore encourages a different style of thinking.

Rather than asking:

"Which sound shall I select?"

we ask:

"What process shall I construct?"


The Wonderful World of Randomness

Random generators are particularly fascinating.

At first, randomness may seem to be the opposite of composing music.

Surely we want control?

But randomness does not have to mean complete chaos.

Imagine a sequence in which seven notes are deliberately programmed, but every eighth note is selected randomly from a restricted range.

Or imagine a filter whose cutoff changes very slightly and unpredictably with every note.

Or a percussion pattern in which there is a 20% chance that an additional sound will occur.

Now the music can become slightly different every time it plays.

We can decide how much uncertainty we permit.

That creates an intriguing relationship between composition and probability.

The composer designs the rules.

The synthesiser explores the possibilities permitted by those rules.


A Practical Experiment — Build a Generative Instrument

This would make an excellent VCV Rack experiment.

Start with:

  1. one oscillator;

  2. one sequencer;

  3. one envelope;

  4. one VCA;

  5. one filter;

  6. one clock;

  7. one audio output.

Get the basic sequence working.

Then add an LFO controlling the filter.

Next, introduce a random voltage generator.

Use the random voltage very gently.

Perhaps it changes the filter cutoff.

Perhaps it occasionally changes the duration of a note.

Perhaps it changes which step the sequencer moves to.

Then add delay or reverb.

Listen for a few minutes.

The fascinating question is:

Have we written a piece of music, or have we built a machine that writes variations of a piece of music?

That is a much more interesting question than simply asking what a particular synthesiser module does.


Effects Become Part of the Instrument

Effects such as delay and reverb are often considered the final stage of music production.

Record the instrument first.

Add effects afterwards.

A modular environment does not force us to think that way.

A delay can become part of the instrument itself.

Its delay time could be modulated.

Its feedback could change automatically.

A signal could be split so that the original sound travels down one path while another version travels through several effects.

Those signals could later be recombined.

We can even feed signals back into earlier parts of the system.

Feedback needs careful control, but creatively it can produce fascinating results.

The distinction between instrument, effect and composition begins to blur.


Why VCV Rack Can Become Enormous

A physical modular synthesiser has an obvious limitation.

Every additional module costs money.

It occupies physical space.

It needs power.

It needs another patch cable.

Sooner or later, the rack is full.

A virtual rack changes those limitations dramatically.

VCV Rack has access to a large ecosystem of modules, and the range of possibilities can become almost absurd.

There are modules for traditional synthesis functions, but also modules involving:

  • probability;

  • logic;

  • switching;

  • sequencing;

  • clock division;

  • mathematical functions;

  • sample manipulation;

  • spectral processing;

  • MIDI;

  • recording;

  • mixing;

  • visualisation;

  • unusual modulation;

  • experimental sound generation.

A virtual modular rack could become far larger than anything I could realistically accommodate as physical hardware.

That does not necessarily mean bigger is better.

In fact, one of the most useful exercises may be to impose restrictions.


Try Building a Synthesiser With Only Ten Modules

This is a challenge I particularly like.

Give yourself a maximum of ten modules.

Now build something musically interesting.

Suddenly every module has to earn its place.

Do I really need another oscillator?

Could one LFO control several destinations?

Could the sequencer perform two jobs?

Could one envelope control both amplitude and filter movement?

Restrictions often encourage creativity.

It is rather like photography.

Owning twenty lenses does not automatically produce better photographs.

Sometimes going out with one camera and one lens forces you to think much harder about the picture.

Modular synthesis can be similar.


Patch Cables Are Ideas

The visual appearance of VCV Rack is part of its appeal.

As the patch develops, virtual cables begin crossing the screen.

Initially there may be three or four.

Later there might be twenty.

Eventually the screen can look like a plate of multicoloured electronic spaghetti.

But every cable represents an idea.

This controls that.

This triggers that.

This signal is being sent there.

This output is affecting this parameter.

That makes modular synthesis unusually visual.

You can often understand something about how an instrument works simply by tracing its connections.


A Brilliant Way to Learn About Sound

This is also why I think software such as VCV Rack has considerable educational value.

Concepts that can initially appear abstract become practical.

Want to understand frequency?

Listen to an oscillator while changing it.

Want to understand harmonics?

Compare sine, triangle, square and sawtooth waves.

Want to understand filters?

Look at and listen to what happens when high-frequency components are removed.

Want to understand amplitude envelopes?

Change attack and release times dramatically.

Want to understand modulation?

Connect an LFO and hear the result.

Want to understand frequency modulation?

Use one oscillator to alter another.

Suddenly ideas from physics, mathematics, electronics, computing and music begin meeting in the same place.

That is precisely the kind of crossover I find particularly interesting.


Music, Physics, Mathematics and Computing Meet

A synthesiser is an excellent example of subjects refusing to stay in their traditional boxes.

There is physics in oscillation and sound waves.

There is mathematics in periodic functions, harmonics and signal processing.

There is electronics in the concepts inherited from physical synthesisers.

There is computing in the software implementation.

There is psychology in our perception of pitch, rhythm, timbre and loudness.

And ultimately there is music.

The technical knowledge is not the final objective.

The final question is whether the result sounds interesting.


From Experiment to Music Production

VCV Rack does not have to remain an isolated experiment.

Sounds and sequences created through modular synthesis can become part of a much larger music-production workflow.

A modular patch might provide:

  • an evolving background texture;

  • an unusual bass line;

  • rhythmic percussion;

  • sound effects;

  • an ambient atmosphere;

  • a strange transition;

  • a repeating sequence;

  • a cinematic drone;

  • an entirely new electronic instrument.

That material can then become part of a larger project inside a DAW.

This is where the possibilities become especially relevant to my own interests in music, video and film production.

Instead of searching through a library for precisely the sound I want, I can potentially ask a different question:

Could I build it?


Designing Sounds for Film and Video

Imagine I need a sound for a science-fiction video.

I could search through presets until I find something suitable.

But modular synthesis offers another route.

Perhaps I start with two oscillators slightly detuned from one another.

I slowly modulate their pitch.

I pass them through a filter whose cutoff changes almost imperceptibly.

I add some random movement.

I send the result through a large reverb and a slowly changing delay.

Now I have not merely selected "Sci-Fi Atmosphere 27".

I have designed a sound specifically for the project.

For another film I might need mechanical tension.

A clock, several sequencers, noise sources and carefully controlled random triggers might create something completely different.

That is where synthesis becomes sound design.


You Do Not Have to Understand Everything Before Starting

The enormous choice of modules can make modular synthesis appear intimidating.

My approach is the same one I use with many technical subjects:

start with something that works, then change one thing.

Build one oscillator.

Make it audible.

Add an envelope.

Add a filter.

Add a sequencer.

Then experiment.

What happens if this cable goes there?

What happens if the LFO is much slower?

What happens if the envelope controls the filter as well as the amplifier?

What happens if the sequence has five steps rather than eight?

What happens if a random generator controls the timing?

Some experiments will sound dreadful.

That is perfectly useful.

You have discovered something.

Others will produce results you would never have deliberately programmed.

Those unexpected discoveries are one of the pleasures of modular synthesis.


Save the Patch — Because You May Never Recreate It

There is another lesson worth learning.

Save interesting patches.

When a modular system becomes sufficiently complicated, recreating exactly the same configuration may be surprisingly difficult.

A tiny adjustment can transform the result.

That is particularly true when randomness, feedback and multiple interacting modulation sources are involved.

The patch itself becomes part of the composition.

In effect, you have designed a new instrument.


The Bigger Idea — Stop Thinking About Presets

Perhaps the most important thing VCV Rack teaches is not how to operate a synthesiser.

It teaches a different way of thinking.

A conventional instrument encourages the question:

"What can this instrument do?"

A modular system asks:

"What instrument would I like to exist?"

That is a much bigger question.

And because the instrument is software, we can experiment with ideas that would be difficult, expensive or simply impractical to construct physically.

Hundreds of modules.

Dozens of connections.

Sequencers controlling sequencers.

Randomness controlling modulation.

Envelopes altering effects.

Oscillators controlling other oscillators.

A patch can be tiny and elegant or enormous and wonderfully ridiculous.

There is no requirement that anybody has built the instrument before.


Conclusion — Build the Instrument First, Then Play It

Modern music technology gives us access to extraordinary libraries of instruments and sounds.

Sometimes, however, the most interesting sound is the one that is not already in the library.

VCV Rack 2 gives us something different.

It gives us the components.

An oscillator does one job.

A filter does another.

An envelope does another.

A sequencer does another.

Individually, they may seem quite simple.

The magic appears in the connections.

And that is why the most interesting question in modular synthesis is rarely:

"What does this module do?"

It is:

"What would happen if I connected it to that?"

Then you make the connection.

Sometimes you get silence.

Sometimes you get noise.

Sometimes you get something awful.

And occasionally you hear something and think:

I have never heard an instrument do that before.

Perhaps that is because, until a few seconds ago, that instrument didn't exist.


Practical Challenge

Open VCV Rack 2 with an empty rack and set yourself one rule:

Do not load a finished synthesiser patch.

Start with a single oscillator and build an instrument one module at a time.

For every new module, ask:

What job do I want this module to perform?

Then ask the more interesting question:

What else could it control?

That second question is where modular synthesis really begins.

#VCVRack #VCVRack2 #ModularSynthesis #Synthesizer #Synthesiser #ElectronicMusic #MusicProduction #SoundDesign #MusicTechnology #DigitalMusic #HomeStudio #DAW #CreativeTechnology #MusicEducation #PhilipMRussell