A history museum is preparing a clearly marked illustrative panel based on an archival photograph of a railway waiting room. A modern plastic display stand intrudes into a new photo taken for the exhibition. Staff want a neutral wooden-looking prop in its place, but they must not present an AI reconstruction as something documented in the historical room.
Skylum describes Luminar Neo GenSwap as a generative tool that replaces or adds objects inside a brushed selection using a text prompt. This guide treats it as an illustration-making aid with strong provenance rules. Every pictured before-and-after state is conceptual artwork, not a real museum image or tested GenSwap output.
Classify the source image before touching pixels
An archival photograph and a contemporary staged photo demand different editing standards. If the file is a primary historical document, do not replace objects and then present the result as unaltered evidence. The hypothetical project begins with a contemporary exhibition set photo inspired by an archival scene; its goal is a transparently labeled visual concept.
Record the photographer, date, rights, original filename and exhibition caption. Keep an untouched copy. Identify what must remain unchanged: the genuine explanatory sign, the wall geometry, the floor edges and any original photographic marks. This list becomes the baseline for all edits.
Choose a narrow region and an honest replacement
From Luminar Neo’s Catalog, Skylum describes opening GenSwap, brushing over the object or area, supplying a simple English prompt and choosing Swap. For a staged exhibit, select only the temporary plastic stand, leaving the surrounding caption and room edges outside the mask.
Prompt for a plain wooden pedestal suitable for a museum staging concept; do not specify a historical maker or claim authenticity. The documentation recommends a mask large enough for the desired replacement and short, concrete prompts. This does not guarantee a one-to-one physical replacement or faithful historical reconstruction.
Inspect the material boundary and contact shadow
Compare the source and candidate at the mask edge, especially where the stand meets the floor. A generated pedestal may float, gain an extra leg, or distort floorboards that should remain straight. The visual comparison depicts a questionable edit with a broken floor line and a review direction that preserves the line.
That drawing is not a record of model performance. Check perspective against nearby architecture and verify that light direction remains plausible. If a correction requires repainting protected signage, reduce the selection or stop the edit. A visually seamless image can still misrepresent the original exhibit.
Do not invent provenance through decoration
An AI-generated wooden pedestal with ornate Victorian carvings could imply the museum possesses an artifact it does not own. Keep the prop visually generic and avoid generating dates, maker stamps, plaques, insignia or accession numbers. If the exhibit design needs such information, typeset it separately from the generated picture using an approved source. The illustration’s warning callout highlights a decorative plate with a fabricated number; the acceptable direction uses an unlabeled display support. Museums should distinguish “interpretive visual” from “documentary photograph” in both the image caption and the collection management record.
Use version comparison rather than repeated blind swaps
Skylum documents repeating Swap, selecting among variants with Undo and Redo, and saving the chosen output to Generative Creations. Keep at least one intermediate image for comparison rather than relying on memory of what changed. Inspect a full-resolution composite and a small-size exhibition mockup.
If a later variant changes the wall molding or archival sign, revert even if the replacement prop itself looks improved. Use a checklist that marks source-preserved, changed region, reviewer and reason for rejecting a variant. The decision should be based on what the exhibit communicates, not only on which candidate is more photogenic.
Protect rights and contextual interpretation
Confirm that the photographer or institution permits uploading and altering the source. Keep internal permissions for collection photos, images of visitors and any licensed artworks within the frame. The availability of a generative tool does not grant permission to reuse copyrighted signage or to portray a named historic location inaccurately.
For sensitive history, discuss whether a composite is appropriate at all. A straightforward diagram, a contemporary labeled photograph or an unedited archival original may explain the same point with less risk. Review the institution’s own authenticity and AI disclosure policy before public display.
Prepare a visible disclosure and accessible caption
For a public panel, use a caption such as “Illustrative reconstruction using a contemporary staged photograph and an AI-generated display prop.” Name the genuine historical source separately if it influenced the scene, without implying the generated prop appears in it. Alt text should describe what is visible and acknowledge the reconstructed nature of the prop where that matters to interpretation.
Avoid small technical disclaimers baked into the image; place the explanation in readable exhibit text. A visitor should be able to tell whether a visual is a genuine historical record, a staged photograph or an artist’s hypothetical reconstruction.
Approve the composite with an archival reviewer
Ask a collection specialist to inspect the original beside the edited image, focusing on signage, floor pattern, room layout and historical claims. Ask a design reviewer to assess contrast, crop and whether the caption will remain near the image on mobile and in print.
Archive the source, masked area, prompt, chosen variant and final caption. If any viewer is likely to mistake the result for historical evidence, change the presentation or withdraw the composite. GenSwap can speed up a staged design concept, but the authoritative record is still the original image and the museum’s provenance documentation.
How We Research
We prepared this guide on November 4, 2026 from named official product documentation: Skylum: Generative tools in Luminar Neo; Skylum: Adding an object using generative AI; Skylum: Start editing in Luminar Neo. We checked the documented entry point, editing and export workflow, access requirements, and boundaries between tool features and human approval. We used original fictional scenarios and explanatory illustrations to demonstrate plausible editorial errors; we did not run these services, upload real client files, collect performance measurements, or make screenshots of their interfaces. Pricing, availability, integrations, permissions and model behavior can change, so recheck the developer sources and your account before any production use. A named editor should compare the outcome against original evidence, verify relevant rights and obtain approval before public distribution.
Frequently Asked Questions
Yes. Skylum describes selecting a region, entering a prompt and generating one or more replacement variants.
No. Preserve the source and visibly disclose reconstruction or generated alterations.
Reject the edit or remove the fabricated detail; use only documented labels outside the generated imagery.
Skylum says users can generate additional versions and move among them using Undo and Redo.
No. They are original instructional diagrams showing what an editor should check.