Controlled image retouching

How can you correct an image with AI without distorting its subject or compromising its distribution?

1250 € per session, excl. VAT

In short

In a half-day session of 3 hours participants learn to make local corrections to an existing image using a tool authorised by your organisation. They define the elements to preserve, carry out targeted retouching, compare versions and prepare a usable file accompanied by quality and rights checks.

Why this training?

Removing an object or correcting a background can introduce artificial edges, inconsistent shadows or invented texture. An image that looks attractive at first glance is therefore not necessarily faithful to the original. Assisted retouching requires a distinction between defects to correct and information to retain. Localised requests, systematic comparison and distribution checks help determine when to accept a version, revert to an earlier one or opt for a manual correction. Ongoing case study: The fictitious organisation Avelune Céramique is preparing a presentation sheet for a vase. The photograph has a cable in the background, a cool colour cast and overly wide framing. These defects must be corrected without changing the vase’s outline, actual colour or texture. Transfer to the workplace: For your next authorised image, reuse the retouching brief and checklist. Compare the original with the exported file, record any remaining defects and obtain approval for any changes that could alter the perception of the product.

Objectives of this training

  • Identify correctable defects and visual elements to preserve.
  • Formulate a localised retouching instruction with explicit constraints.
  • Make successive corrections while retaining the original image.
  • Compare versions using a fidelity and quality assessment grid.
  • Export a file suited to the medium and document its conditions of distribution.

Who is it for and what are the prerequisites?

Roles and tasks concerned

  • Communications staff who adapt photographs for digital media.
  • Marketing or administrative staff who occasionally correct presentation images.

Prior skills

  • Have previous experience of making simple requests to generative AI.
  • Be able to open, duplicate and save an image; distinguish between pixel dimensions and file size.

What you will work on

  1. Diagnose and define the scope of corrections

    Objective: Turn observed defects into verifiable instructions. Assessing framing, light and textures; distinguishing between correction and misleading transformation; copyright and permission to import.

  2. Carry out targeted retouching

    Objective: Correct an area without unnecessarily regenerating the rest. Instructions specifying the area, change and elements that must remain unchanged; local selection if available; retaining versions and understanding the tool’s limitations.

  3. Adjust the rendering without distorting the subject

    Objective: Improve presentation while preserving fidelity. Colour cast, shadow consistency, cropping and textures; risks of invented details; using manual correction where necessary.

  4. Check and prepare for distribution

    Objective: Deliver a checked and traceable image. Inspection at 100 %, checking in the intended format, export; consent from people depicted, modification disclosures and prevention of misleading deepfakes.

What you take away

  • Retouching brief with editable areas and protected elements.
  • Log of instructions and intermediate versions.
  • Exported final image with the original retained.
  • Comparison grid and rights and disclosure sheet.

Method and assessment

  • Short teaching inputs followed by immediate practice on the ongoing case study.
  • Individual work followed by work in pairs, with reciprocal review of the outputs.
  • Guided group discussion: what works, what needs correcting and what remains a human decision.
  • Preparation: Arrange access to a tool authorised by your organisation and check its actual retouching and export capabilities. Prepare a fictional or anonymised image authorised for use, along with a brief specifying the expected dimensions and elements that must remain unchanged.

Observable criteria: The corrections meet the brief without altering the vase’s characteristics. ; No visible joins, doubled edges or inconsistent shadows remain when examined at 100 %. ; The comparison assigns each criterion a status of compliant, to be corrected or not applicable, with justification. ; The file meets the requested dimensions and format and opens correctly. ; Rights, any required consent and necessary disclosures are recorded before distribution..

Frequently asked questions

Can we work on an internal photograph?

Only if it is anonymised, authorised for import and compatible with confidentiality rules. Otherwise, use the fictional case study.

Does the tool need to offer local selection?

This is preferable for isolating a correction. Exercises are adapted to the verified features; any unintended changes must be checked.

Can we alter a person’s face?

That is not the intended exercise. Any use in which a person is identifiable requires consent and rights to be checked; misleading deepfakes are excluded.

Must the use of AI be disclosed?

Check the applicable obligations, channel and internal rules. Document the changes and add the required disclosures without removing credits that are due.

Last updated: