Level 5 exam — Midjourney & Visual AI

Exam — fresh questions

12 new questions to validate Production & generative video: visual guidelines, generative video, animation and voice, assisted marketing, watermark and C2PA traceability.
Answer everything, then click Check: your score and explanations will appear.

5.1 · Recall

1. What do AI visual guidelines contain?

Only a list of software Aesthetic pillars, template prompts, locked parameters, references and prohibitions Only production pricing No rules, just examples

The framework that lets ten visuals from three people look like they came from the same hand.

5.1 · Application

2. Sonia wants to unify her client's visuals in Algiers. What is her first methodical step?

Let each contributor invent their own prompts Write a style theory without generating Generate broadly then select the visuals that “are” the brand Impose the most complex style possible

Guidelines are built by induction: generate, select, analyse the constants, formalise.

5.2 · Recall

3. What is the central difficulty of generative video?

Temporal consistency: stability from one frame to the next The size of the final file The name of the editing software The choice of typeface

Objects, characters and lighting must stay stable — far harder than a beautiful still image.

5.2 · Application

4. Sonia must animate her client's validated banner. Which approach does she choose?

Image-to-video: animating the validated image describing only the movement Redescribing the whole scene on every take Promising a long video with no editing Generating a complete film in one block in text-to-video

Framing and style already fixed: ask only for the movement, simple and slow, then assemble in the edit.

5.3 · Application

5. Sonia creates a virtual presenter for her client. What does she do first, ethically?

She releases with no disclosure at all She has it give financial advice without validation She uses a client's face without warning her She picks a generated character and plans an “AI-generated content” disclosure

Generated character or written consent, validated text, short sequence, mandatory transparency.

5.3 · Arbitration

6. The client asks Sonia to have the avatar speak a real person's testimonial without their consent. Decision?

Accept, the client takes everything Refuse without explanation Require explicit, documented consent from the person, or switch to a generated character Publish anyway with no disclosure

Without written consent it is a deepfake with serious legal and human consequences — non-negotiable.

5.4 · Application

7. The client hesitates between two moods for the campaign. What does Sonia suggest?

Decide alone by taste Produce two on-guidelines variations and launch a real A/B test Publish both with no filter and no measurement Debate in meetings until consensus

Lower trial cost lets you test instead of guess: data decides.

5.4 · Application

8. Sonia produces many visuals for a campaign. Which safeguard protects the brand?

Change style with every visual Remove all validation to go fast Publish everything generated Visual guidelines + a quality/compliance filter before every release

Without guidelines, volume becomes patchwork; without a filter, you release defects and protected elements.

5.4 · Arbitration

9. The client's management wants “as many visuals as possible” with no indicator. Sonia's decision?

Publish unfiltered to go faster Produce in bulk with no measurement Refuse any AI production Reframe on one indicator: test, measure performance, keep what works

“Producing more” is not the goal: volume serves selection, performance decides.

5.5 · Recall

10. What is the logic of the C2PA / Content Credentials standard?

Infallibly detecting every fake Banning generated images Certifying the authentic with signed provenance rather than detecting fakes Imposing a visible watermark everywhere

Signed history of creation and edits — necessary but insufficient against disinformation.

5.5 · Application

11. Sonia delivers her visuals to the client in Algiers. What does her traceable delivery contain?

Images alone, with no documentation Each visual with its prompt, version, parameters and an AI disclosure Only the invoices Visuals with no disclosure at all

Track (prompts, versions, sources), disclose when expected or mandatory, preserve metadata.

5.5 · Arbitration

12. Sonia is asked to republish a dubious-origin realistic image involving people. Decision?

Verify the origin before any republication and have it validated by a lawyer if sensitive Assume “uncertified” means “fake” and accuse Remove all transparency disclosure Republish at once with no checks

Verify the source, seek independent confirmation, validate legally — without hasty conclusions.

What's next?

Score ≥ 80%: move to Level 6 “In practice” — real cases, troubleshooting and the Midjourney & visual AI toolbox (NO exam).
Score 50-79%: reread the explanations above — each one re-explains the concept — then redo the 48h challenges and template sheets of the failed sections.
Score < 50%: revisit sheets 5.1 (guidelines by induction) and 5.5 (C2PA and traceability), your two vital moves, before continuing.