How I Turned a Real IMO 2026 Exam Screenshot Into a Fully Narrated Solution Video (Step by Step)

Why this workflow matters

Most AI video tools either force you to type out a problem by hand or generate a fake, re-drawn version of your source material. For competition math — where authenticity matters — that's a dealbreaker. A redrawn diagram isn't the real IMO paper. A paraphrased problem isn't citable. Here's exactly how I used TimaLens to take a real screenshot of IMO 2026 Problem 5 and turn it into a fully animated, narrated walkthrough — keeping the original exam image intact inside the video itself.

Step 1 — Write the script prompt

I started with a simple instruction, not a full script:

"Help me solve this math problem in a simple, easy-to-understand way. Explain each step clearly, as if you were teaching a beginner."

TimaLens uses this as the narration brief — you're not writing scene-by-scene copy, you're telling it how to teach, and it plans the structure for you.

Step 2 — Insert the real image (not a visual reference)

This is the step people get wrong. TimaLens has two different image features, and they do opposite things:

Real Image vs. Visual Reference — what's the difference? A Visual Reference is for consistency — a logo or character you want redrawn the same way across every scene. A Real Image is for authenticity — TimaLens keeps the image exactly as you gave it, unedited, and treats it as something to narrate around, not redraw.

Real Image vs. Visual Reference — what's the difference? A Visual Reference is for consistency — a logo or character you want redrawn the same way across every scene. A Real Image is for authenticity — TimaLens keeps the image exactly as you gave it, unedited, and treats it as something to narrate around, not redraw.

Step 3 — Choose language and voice

TimaLens offers cloned and pre-built voices per language. For a teaching-style walkthrough, I went with Neutral Female — balanced and clear, which suits a step-by-step explanation better than a casual or cheerful tone would for competition math.

Step 4 — Set the target duration

Durations range from a 30-second short to a full 15-minute course-length video. For a single olympiad problem, 2 minutes (Standard) hit the right balance — enough time to actually teach the key insight without padding.

Step 5 — Pick orientation and drawing style

- Orientation: Landscape (16:9) — better for a resource page and for embedding in a full write-up. - Drawing hand style: I chose a direct ink (No hand) for a clean, precise feel that matches a math-heavy explanation — cartoon or realistic hands work better for more casual, consumer-facing content.

Step 6 — Review and launch

The summary screen confirms everything before it spends credits: target duration, format, voice, and — importantly — that the real image was picked up correctly ("1 narration-aligned" image). Hit generate, and TimaLens builds the full scene sequence, narration, and whiteboard animation automatically.

The result

A fully narrated, step-by-step solution video — with the actual IMO 2026 Problem 5 exam image appearing on screen exactly as published, not redrawn or reinterpreted.

FAQ

What's the difference between Real Image and Visual Reference in TimaLens? Visual Reference keeps a logo or character consistent across scenes by redrawing it each time. Real Image keeps an uploaded image exactly as-is — untouched — and narrates around it. Use Real Image for anything that needs to stay authentic, like an exam screenshot, a chart, or a document page. Do I need to write a full script for TimaLens? No. A single instruction describing the tone and goal (e.g., "explain this simply, like teaching a beginner") is enough — TimaLens plans the scene-by-scene narration itself. Can I use a real photo or screenshot inside a TimaLens video? Yes — upload it through Insert Real Image and add a short description of what it is and how it should be used. TimaLens keeps it unedited in the final video. How long does it take to generate a video like this? Generation time depends on length and scene count; a 2-minute video with one inserted image typically completes in a few minutes.