How a deep tech explainer video actually gets made
The sequence from first call to delivered master, what happens at each stage, and where a deep tech film goes wrong in ways a SaaS film never does.
- The order is script, storyboard, technical review, build, grade. Animation before technical review is the single most expensive mistake available.
- Deep tech adds one stage a SaaS film does not have: a factual review by your engineers, before anything is rendered.
- The technique is chosen after the script, not before. Choosing 3D at the brief stage is choosing the answer before the question.
- Most of the risk sits in the first week, where changes are cheap. Almost none of it sits in production, where they are not.
Most explainer video processes are published as a five-step graphic with a rocket at the end. This is the actual sequence, including the parts that are awkward to advertise.
What happens before anyone writes a script?
You send a spec sheet and we ask what the film has to survive. Not what it should say, what it has to survive: a conference floor with no sound, a procurement lead who watches thirty seconds, an investor who has already seen four companies claim the same thing this month.
That question decides length, technique and structure before a word is written. A film for a booth loop and a film for a sales call are different films, and producing one and cutting it down produces two bad films.
Why does the script come before the technique?
Because the technique is an answer, and you cannot pick an answer before you have the question. A brief that opens with we need a 3D explainer has already skipped the only decision that matters.
The script determines what has to be shown. Sometimes what has to be shown is a mechanism inside a sealed unit, and that is 3D. Sometimes it is an argument about why the industry is optimising the wrong component, and that is 2D and motion graphics. Sometimes it is a founder who is genuinely compelling on camera, and the honest answer is a shoot and a good editor.
We have made all three for deep tech companies. The films that failed were the ones where the technique was chosen in the brief.
What does technical review actually involve?
Your engineers read the script and mark anything that is untrue, imprecise, or true but misleading. It usually takes two sessions. It is the stage that most separates a deep tech film from a SaaS one, and the stage most often skipped.
Skipping it is expensive in a specific way. A factual error found in the script costs an afternoon. The same error found after animation costs a re-render. Found after delivery, at a conference, in front of specialists, it costs the credibility of everything else in the film.
If a heart rate was 120, it had to be 120, not 100. At a conference full of experts, get one detail wrong and you lose your credibility.
That is Cloudphysician, describing a medical AI film. The same sentence applies to a torque figure, a process order, a spectral response or a tolerance.
When does the storyboard get signed off?
After technical review and before any production begins. The storyboard is the last cheap moment. Everything after it costs real hours.
This is also where length gets fixed rather than estimated. A storyboard that runs long is trimmed on paper, which is free. A film that runs long is trimmed in the edit, which means paying to produce material you then discard.
What actually happens during production?
For 3D, modelling from your CAD or spec, then lookdev, then animation, then lighting and render. For live action, a shoot day and an offline edit. For 2D, design of the visual system, then animation.
The useful thing to know is that this stage is deliberately the least interactive. If the earlier stages were done properly there is little to decide here, and a client who is being asked for many decisions during production is usually being asked to fix something that should have been settled in week one.
What comes back at the end?
A graded master, plus a version set: homepage, sales, social and booth cuts, with and without captions. Source files, project files, audio stems and captions are yours.
The one thing we cannot hand over is licensed stock or library music, which stays under its own licence. We say up front which shots those are. Where a film needs to outlive a licence, we shoot or build the frame instead.
Where do deep tech films go wrong?
Two places, consistently.
The first is technique chosen before script, which produces a technically accomplished film about the wrong thing. It looks like a success until it fails to move anyone.
The second is a film made before the positioning is settled. Animation is the most expensive possible way to discover that your story is not right. If the positioning is genuinely open, that is a short piece of work to do first, and it is cheaper than reanimating.
How do you know whether a film is worth making at all?
Ask whether the product can be understood from a screenshot. If it can, a well-built page section will usually outperform a film for a fraction of the cost, and we will say so.
Film earns its budget when the thing is invisible, when it is inside a reactor or a cell or a network stack, or when it has no physical form at all. That is most of deep tech, which is why this page exists. But it is not all of it.
Written by Mejo Kuriachan. More in the blog, the glossary and the FAQ.