Visual continuity in production

How can you maintain a consistent visual identity across multiple generated shots despite changes in framing and lighting?

1960 € per session, excl. VAT

In short

This 7-hour day, excluding breaks, structures a visual continuity process for an audiovisual production assisted by generative AI. You define the fixed attributes of characters and locations, organise references, test several shots and correct discrepancies. The result includes a continuity dossier and a documented evaluation.

Why this training?

A face that changes between two camera angles, a jacket whose details disappear or a door that has moved undermine the clarity of a scene. These flaws must be spotted before editing leads to repeated rework. Generative AI requires explicit visual direction and human checks. Separating narrative constants from permissible variations makes it possible to target corrections rather than regenerate every shot without a diagnosis. Running case study: The fictitious studio Orlisse Animation is preparing a reunion scene in a restoration workshop: two characters, a workbench and a side window appear in six shots. Variations in faces, costumes and the arrangement of props disrupt continuity between shots. Transfer to the workplace: On your next scene, apply the checklist before approving the shot breakdown, then again after generation. Record discrepancies by shot and pass the selected references, approved versions and outstanding reservations to the editing team.

Objectives of this training

  • Formalise the fixed physical, costume and spatial attributes of a scene
  • Build a reference bible that can be used shot by shot
  • Write instructions that separate fixed elements from variations in staging
  • Diagnose and correct continuity breaks
  • Measure shot compliance using a shared assessment grid

Who is it for and what are the prerequisites?

Roles and tasks concerned

  • Directors, storyboard artists and art directors using generated images
  • Editors and production managers responsible for approving visual continuity

Prior skills

  • Be able to use generative AI and rephrase a visual instruction.
  • Understand the concepts of shot breakdown, continuity matching, camera axis and shot size.

What you will work on

  1. Define critical continuity matches

    Objective: Prioritise the risks of continuity breaks. Identity, costume, props, spatial layout, lighting; narrative continuity versus superficial resemblance.

  2. Build the master references

    Objective: Stabilise the attributes to be retained. Silhouette, proportions, palette, materials, set landmarks; rights, consent for image use, prevention of deceptive deepfakes and synthetic content disclosures to be planned.

  3. Translate the bible into instructions

    Objective: Distinguish fixed attributes from permitted variations. Stable descriptive block, action, framing, lighting; use references only in line with the tool’s verified capabilities.

  4. Test changes in viewpoint

    Objective: Identify drift between shots. Side-by-side comparison, spatial landmarks, entrances into and exits from the frame; temporal checks when clips are available.

  5. Correct without shifting the problem

    Objective: Choose a targeted revision. Diagnosis by attribute, isolated modification, version tracking and checks for side effects.

  6. Approve the dossier for editing

    Objective: Document quality and reservations. Passed checks as a proportion of applicable checks, blocking defects, artistic judgement and traceability.

What you take away

  • Character and set bible with identified references
  • Continuity table for the six shots
  • Instructions dossier and version log
  • Completed quality assessment grid and list of priority corrections

Method and assessment

  • Short teaching inputs followed by immediate practice on the running case study.
  • Individual work followed by work in pairs, with peer review of the outputs.
  • Guided group review: what works, what needs correcting and what remains a human decision.
  • Preparation: Arrange access to a tool authorised by your organisation and verify its actual capabilities before the session. Prepare a fictitious or anonymised scenario and references with documented usage rights.

Observable criteria: Each character has visible and verifiable fixed attributes. ; The six shots respect the defined spatial layout or explicitly flag an exception. ; Each detected continuity break includes a location, a severity level and a decision. ; The compliance rate is calculated against applicable checks, before and after correction. ; The references have a recorded rights status and the necessary disclosures are identified..

Frequently asked questions

Do you need to generate a complete video?

No. Six key images are enough to work on continuity between shots; clips also allow you to examine temporal drift.

Can you use a real person’s face?

The exercises prioritise fictitious characters. Any use of a real person requires checks on rights, consent and the distribution context.

What can you do without a visual reference feature?

Use structured descriptions and manual comparison, documenting the limitations observed.

Can a beautiful image be rejected?

Yes, if it breaks a critical continuity match. Approval concerns the scene, not just each individual image.

Last updated: