AI-generated supplementary shots

How can you produce cutaway shots that are consistent with a video, then check their quality before editing?

1250 € per session, excl. VAT

In short

In a half-day session of 3 hours you use generative AI to create supplementary shots for a short video. You prepare prompts, compare variants and check continuity, artefacts and rights. You leave with a documented selection, ready to be tested in the edit.

Why this training?

A cutaway shot conceals a transition, supports an explanation or provides a visual pause. However, an appealing image that is inconsistent with the setting, lighting or message can make the edit less credible. Generation allows you to explore missing illustrative footage, but requires careful selection. Participants learn to distinguish a usable shot from a misleading or unstable result, then document their choices rather than multiply attempts without criteria. Continuing case study: The fictitious organisation « Avelune Céramique » is preparing a 45-second video presenting an imaginary pottery workshop. A fictitious interview requires three illustrative shots: tools on a table, a shelf of pieces and a material detail. No real faces or identifiable brands are used. Transfer to the workplace: For a future video, reuse the prompt template for a specific cutaway requirement. Test two variants in the edit and record the faults, corrections and final decision before any release.

Objectives of this training

  • Identify three cutaway requirements in a short sequence
  • Write a prompt specifying subject, framing, lighting and movement
  • Produce variants using a tool authorised by your organisation
  • Compare results using a visual quality assessment grid
  • Select and document shots that can be used in the edit

Who is it for and what are the prerequisites?

Roles and tasks concerned

  • Communications officers producing short editorial videos
  • Content creators and video editors looking to supplement existing footage

Prior skills

  • Have already formulated simple requests to a generative AI tool.
  • Be familiar with the concepts of framing, continuity and shot duration.

What you will work on

  1. Choose what the cutaway should show

    Objective: Turn gaps in the edit into precise visual requirements. Functions of a cutaway shot; distinction between illustration and evidence; usable duration, format and continuity with the main shot.

  2. Describe a shot that can be generated

    Objective: Formulate verifiable prompts and limit ambiguity. Subject, environment, shot size, lighting, texture and movement; authorised references; parameters based on the capabilities actually available.

  3. Correct and select variants

    Objective: Assess usability rather than aesthetics alone. Object distortions, flickering, unstable textures, impossible movements; colour consistency, light direction and temporal continuity.

  4. Prepare for responsible use

    Objective: Make the handover of shots conditional on their final check. Rights to references and conditions for using outputs; consent from the people depicted; deepfake risks; synthetic content disclosures according to the context and applicable rules.

What you take away

  • List of three shots with their narrative function and target duration
  • Set of prompts and variant log
  • Folder of selected shots with technical specifications
  • Visual checklist and rights traceability record

Method and assessment

  • Short instructional inputs followed by immediate practice on the continuing case study.
  • Individual work followed by work in pairs, with peer review of the outputs.
  • Facilitated group review: what works, what needs correcting and what remains a human decision.
  • Preparation: Check access to a tool authorised by your organisation and its actual video generation capabilities, as well as the available usage limits. Prepare a fictitious or anonymised clip, its text and its technical specifications; check the rights to any visual reference provided.

Observable criteria: Each shot corresponds to an identified passage and does not contradict the commentary. ; The prompts specify framing, lighting, movement and elements to exclude. ; Each variant receives a score from 0 to 2 for relevance, continuity, stability and absence of artefacts. ; Faults are identified at normal speed and frame by frame; any critical fault results in rejection. ; The provenance of references, permissions and required disclosures are documented..

Frequently asked questions

Do we need to film new footage?

No, the exercise uses supplied fictitious or anonymised material and references that are authorised for use.

Can a recognisable employee be reproduced?

This is not planned as part of the exercise. Real-world use requires, among other things, appropriate consent and a rights check.

What if generation is too slow?

Limit the tests and continue the analysis using prepared variants; clearly identify the shots actually produced during the session.

Will all the shots be suitable for release?

No. Selection depends on the faults identified, continuity, rights and the context of release.

Last updated: