Elzatian AI animator · AI content creator · AI trainer

Getting started

Learning AI video in Singapore: what to learn first

Learn the decisions before the tools. The order that works is: write the shot list, lock the character and the look, approve a still, then animate, then finish the sound — because each stage makes the next one cheaper. Beginners who start with tools spend their first month re-rendering; beginners who start with the order finish something in the first week.

3 min read 8 sections Written by Elza

Summary

The order matters more than the tools. Write the shot list first, lock the character and the look, approve a still before animating, then finish the sound — because each stage makes the next one cheaper. A shot of the wrong length costs nothing to fix in a shot list and a full render to fix in motion. The first month should be one scene finished end to end, sound included, rather than a film attempted in pieces.

The tools are the last thing to learn

Every few months the tools change, and every time they do, the people who learned only the tools start again. The people who learned the order — what has to be decided before what — move to the new tool in an afternoon.

So the first thing to learn is not a generator. It is the sequence: script and shot list, character bible, style lock, keyframe, motion, master.

First, write the film down

A shot list is the film as shots before anything is generated: what each shot is for, the emotion it has to land, how long it lasts, and how the camera behaves.

Duration is written first because a shot of the wrong length is a shot you re-render. This is the cheapest stage of the whole film and the one that decides whether the expensive stages are needed once or five times.

Second, lock the character and the look

The character bible answers who the character is — face, build, wardrobe, colour. The style lock answers how the film looks — light, palette, detail, camera behaviour.

Both are approved once and then pointed at on every shot. Without them, the face drifts and the light changes between shots, which is the failure that makes otherwise good work look broken.

Third, approve a still before you animate

A keyframe costs a fraction of a clip and tells you almost everything: whether the composition works, whether the character is right, whether the light matches the shot before it.

Approving the still first is the single habit that saves the most credits. The expensive thing in this work is not generating — it is generating again.

Fourth, motion and the master

Animate from the approved frame with one clear action per shot and one motivated camera move. A single action in two seconds holds; three actions in six do not.

Sound is the last stage, and it is not decoration added at the end. It is what turns a sequence of good shots into something an audience reads as a film.

What to practise in your first month

One scene, not a film: a handful of shots, finished all the way through, including the sound. Then a second one, faster.

The habit to build alongside it is diagnosis. When a shot fails, work out which instruction stopped being obeyed — the reference was not attached, there were two actions, the camera had no reason to move — instead of re-rolling and hoping. That is the difference between learning this and getting lucky.

Our take

Beginners who struggle are almost never short of tools; they are short of an order. Re-rendering a shot whose purpose was never decided is the most common way credits disappear, and it is avoidable at the shot-list stage, before anything is generated.

The habit worth building alongside the order is diagnosis — asking which instruction stopped being obeyed, rather than generating again and hoping. That is what turns one good shot into a repeatable method, and it is the part no tool release can hand you.

Where this is taught

AI Film Lab: Cinematic Storytelling — 2 Days, taught live in a small cohort. Ask on WhatsApp (+65 9188 6948) for the next dates.

See the course Hand me a brief

Keep reading

The other articles

Choosing a course

How to choose an AI course in Singapore

Choose an AI course in Singapore by what you leave with, not by the syllabus. The courses worth paying for end with something you made on your own material and a method you can repeat; the rest end with a certificate and a folder of clips you cannot explain. Provider name, course length and subsidy are all secondary to that one question.

Read

Funding

Before you spend SkillsFuture Credit on an AI course

SkillsFuture Credit can be used for courses listed on the MySkillsFuture portal, and eligibility is decided per course rather than per subject — so a course being about AI tells you nothing about whether it is claimable. Before you commit, confirm three things: that this specific course is listed as eligible, that the subsidy in the quote applies to you, and that you know what you leave with. The last one matters most, because a subsidised course that teaches you nothing is still the expensive option.

Read

Commercial work

AI video for a small business in Singapore: what to make first

Three videos earn their keep before anything else: a short product spot that shows the thing working, a founder or team story that explains why anyone should care, and vertical cut-downs of both for social. Brand films, event recaps and animations are later purchases — worth making once those three are already doing their job.

Read

Short-form

Making AI videos for TikTok without them looking fake

AI short-form reads as fake for three reasons, and picture quality is not one of them: the first second is a logo instead of a sentence, the face changes between cuts, and the sound does not belong to the room. Fix those three and an AI-made video can sit in a feed without being spotted — which is the only thing that makes it worth making.

Read

AI news · productivity

Singapore uses AI more than most places — the results are not following

Singapore has an unusual combination: a workforce that uses AI about as much as anywhere in the world, and a set of employers that has barely started. In the same year, 56 per cent of Singapore workers reported using AI at least several times a week, while 71.5 per cent of firms had not adopted it at all and only 15 per cent of workers strongly believed it would improve their job. When use is high and confidence is low, the missing piece is usually not the tool. It is the method.

Read

AI news · creativity

AI video's consistency problem, and what Google's new research changes

On 24 September 2026 Google Research published a multi-agent framework for long-form AI video that sets out to fix two named failures: identity drift — a character's face, attire or scenery quietly changing between shots — and cascading failure, where one bad asset corrupts everything generated after it. It is the clearest sign yet that the industry has stopped treating AI video as a picture-quality problem and started treating it as a continuity problem. That is the problem a shot list and an approved keyframe already solve.

Read

The method behind these pieces is written out in the guides — the six stages, the four rules behind every prompt, and how a character survives a whole film.