Nano Banana Pro Swap Person & Face Swap Prompts

A “swap person” prompt asks Nano Banana Pro to take one person’s identity from a reference image and place it into another image or scene — replacing someone in a photo, putting yourself into a poster, or keeping the same face across a set of edits. The examples here are public prompts labelled Nano Banana Pro that state which identity to keep and what to replace.

Write a swap as two jobs in a fixed order. First say which image supplies the identity (“the person in image 1”) and list the features that must not change: face shape, eyes, skin tone, hairline, distinguishing marks. Then say what the other image supplies — pose, outfit, lighting, background — and what gets replaced. Ask the model to match the light direction and colour of the target scene so the swapped face does not look pasted on. Only swap people who have agreed to it, never to deceive, and label edited images where a real likeness could be mistaken for a real photo. The gallery below uses the same model-filtered query as the dated evidence count, so the cards are published examples rather than generated filler.

Example starting brief: “Use the person in image 1 as the identity reference: keep the same face shape, eyes, skin tone, freckles and hairline. Replace the person standing on the left in image 2 with them, keeping image 2’s pose, navy coat, street background and evening light. Match the warm light from the right on the face, natural skin texture, no extra people, no text.” Treat it as a controlled starting point: change one variable group, inspect the output and keep the model-specific limitation visible.

Browse all prompts →

@IamEmily2050

@ChillaiKalan__

@saniaspeaks_

@aimikoda

@zayleeai

@Samann_ai

You’ve reached the end ✦

Evidence and editorial status

7 published prompts matched the read-only query in the snapshot used for this page. The collection is below the 20-prompt index gate and remains in the backlog.

Edited 2026-10-01. Counts are evidence snapshots, not search volume or ranking data.

Prompt variables

Change one group at a time so the result can be compared and reused.

which image supplies the identity, named by order
identity anchors that must not change
what the target image keeps: pose, outfit, background, light
light direction and colour match on the swapped face
consent, disclosure and no-extra-people constraints

Model recommendations

Nano Banana Pro. The label on every card here; the examples use it for multi-image identity references and plain-language keep/replace instructions. Limit: Identity can drift when the two images differ strongly in angle or light; results are not a guaranteed likeness.
Nano Banana 2. Worth trying the same keep/replace structure when your account offers it. Limit: PicGens has not run a controlled swap comparison between versions, so treat any difference as something to test.

Failure modes

The swapped face looks pasted on. Describe the target scene’s light direction and colour and ask for the face to match it.
The result mixes features from both people. Name the identity image explicitly, list its anchors first and avoid describing the replaced person’s face.
Extra people or duplicate hands appear. Say how many people should be in the final image and keep the target pose simple.

Method: this model×scene page uses one model-filtered full-text query over published prompt rows. The count shown comes from the static catalog snapshot the page was built from and is rechecked for metadata, links and sitemap output; it is not search volume, ranking or model-performance data. No editorial relevance sample has been recorded for this page yet. Public examples remain subject to their source terms and product/licence review. Last edited 2026-10-01.

Briefing Nano Banana for face-preserving photo edit prompts

which image supplies the identity, named by order
identity anchors that must not change
what the target image keeps: pose, outfit, background, light
light direction and colour match on the swapped face
consent, disclosure and no-extra-people constraints

Variables to test in order

Use one variable group per comparison. Example starting brief: “Use the person in image 1 as the identity reference: keep the same face shape, eyes, skin tone, freckles and hairline. Replace the person standing on the left in image 2 with them, keeping image 2’s pose, navy coat, street background and evening light. Match the warm light from the right on the face, natural skin texture, no extra people, no text.”

which image supplies the identity, named by order
identity anchors that must not change
what the target image keeps: pose, outfit, background, light
light direction and colour match on the swapped face
consent, disclosure and no-extra-people constraints

Nano Banana fit and limits

Nano Banana Pro: The label on every card here; the examples use it for multi-image identity references and plain-language keep/replace instructions. Limitation: Identity can drift when the two images differ strongly in angle or light; results are not a guaranteed likeness. Nano Banana 2: Worth trying the same keep/replace structure when your account offers it. Limitation: PicGens has not run a controlled swap comparison between versions, so treat any difference as something to test.

Failure checks for this model×scene pair

The swapped face looks pasted on. Fix: Describe the target scene’s light direction and colour and ask for the face to match it. The result mixes features from both people. Fix: Name the identity image explicitly, list its anchors first and avoid describing the replaced person’s face. Extra people or duplicate hands appear. Fix: Say how many people should be in the final image and keep the target pose simple.

A safe production handoff

Keep the prompt, reference permissions, output review and final design pass separate. A public example demonstrates an approach, not a licence or an outcome guarantee.

Frequently asked questions

What is a good Nano Banana Pro swap person prompt?: Name the identity image, list the features to keep, then say which person in the target image is replaced and what of that image stays — pose, outfit, background and light. End with constraints such as no extra people.
How do I keep the face the same in a Nano Banana face swap?: Use a sharp, front-facing identity photo, list its anchors before anything else and make only one change per request. Large differences in angle between the two images make identity harder to keep.
Is it okay to swap someone’s face into a photo?: Only with that person’s permission and never to mislead. Use photos you have the right to edit, and disclose the edit wherever the image could be taken as a real photograph.
Does the same swap prompt work in Nano Banana 2?: The keep/replace structure is a reasonable starting point, but PicGens has not tested it across versions. Run it on your own permitted images and compare.