Canonical: https://instain.ru/en/articles/remove-an-unwanted-person-from-a-photo-with-nano-banana/ Language: en A practical guide # Remove an unwanted person from a photo with Nano Banana A person in the background can spoil an otherwise good travel photo. This guide covers targeted removal with a text prompt: what to ask the model to fill in and where to look for subtle changes elsewhere in the image. ![A female traveler with a turquoise backpack on the left and a highlighted passerby in a red jacket on the right.](https://instain.ru/assets/images/article-nano-banana-ubrat-cheloveka-s-foto-cover-6c12fa2f0b49-1200.webp) Removing one passerby requires a precise description of the target. An educational AI-generated illustration, not the result of processing an image in Nano Banana. ## Choose a photo that is worth editing Imagine a photo taken on a walk: your companion stands on the left with a turquoise backpack, while a person in a red jacket walks through the shot on the right. You want to remove the passerby and keep the photo close to the original. Nano Banana accepts an image and a text description of the changes for this task. The example below is educational; it is not a report of personally testing a particular file. Look at what the passerby obscures. A small section of a stone wall or path is easier to assess using neighboring details. If the person blocks your companion’s face, a complex inscription or half a sculpture, the model will invent the missing part. It does not know what was there when the photo was taken. For an important detail, it is better to find another shot of the same scene. Save the original separately. Use your own photo at the best available quality rather than a reduced copy from a messaging app. Before uploading, make sure you can use the image and share it with the service you selected. ## Identify the person to remove and define the edit A request such as “remove the person” is not specific enough when several people are in the shot. Give a position and a distinctive detail: the passerby in the red jacket on the right, next to the stone wall. Then identify who should stay: the person with the turquoise backpack on the left. If their clothing colors are similar, add their relative positions rather than assuming the model will guess your choice. Decide what should fill the cleared area. In our example, that means continuing the existing wall and paved path under the same lighting. Asking for a beautiful background could lead to more extensive changes to the scene. Here, the request needs to describe the content of a specific area. ## Upload the photo with one complete prompt Open a Nano Banana mode that supports editing a source image and add the photo you selected. Google’s documentation describes this workflow as editing with an image and text. In this example, you specify the target area by describing the person and their position in the shot. Example prompt for practice: “Remove only the passerby in the red jacket on the right, next to the stone wall. Fill the cleared space with a natural continuation of the existing wall and paved path. Keep the position and appearance of the person with the turquoise backpack on the left, including their clothes, face and hands. Keep the framing, viewpoint, lighting and all other objects. Also remove the shadow cast by the passerby being removed onto the path, while preserving the shadows of all other objects. Do not add new people, writing or decorative details.” The prompt mentions a shadow because there is one in this scene. For another photo, first check whether the dark patch actually belongs to the person you want to remove. If there is no visible reflection, there is no reason to ask for one to be removed. Otherwise, the instruction could affect another area that fits the description. ## Check the removal area and any remaining traces Once you have a result, zoom in on the area where the passerby stood. Check the line of the wall and the joints in the paving: they should continue in consistent directions. Identical repeating stones, a suddenly blurred patch or an abrupt color transition can reveal generated content even when no silhouette is visible. Inspect the edge of the edited area. A piece of a shoe, a narrow strip of the jacket or a shadow can remain as a trace of the person. On wet surfaces and in glass, check for a reflection. Removing the figure alone does not confirm that every associated detail has disappeared. Assess the lighting across the whole photo. If the shadows cast by the wall and trees remain, but the regenerated area has an unnaturally uniform, light-colored rectangle, the edit is still noticeable. Compare it with nearby surfaces that are lit in a similar way. ![Three courtyard details: a passerby, a shadow left after removal and a clear background with continuous paving joints.](https://instain.ru/assets/images/article-nano-banana-ubrat-cheloveka-s-foto-illustration-1-9b9ba65e8251-1200.webp) The original scene is on the left, the center shows a remaining shadow and a broken paving joint, and the corrected version is on the right. This is an educational AI-generated illustration, not a Nano Banana test. ## Compare the main subject and the geometry of the whole shot After checking the background, look again at your companion on the left. Compare their face, fingers, shoulder outline and the shape of the turquoise backpack with the original. Asking the model to preserve these details guides the generation but does not technically lock them. An attractive result can still contain unwanted changes. Check the edges of the shot, the horizon, windows and fence lines. Switching between the original and edited photos at a similar scale can help. If one version has become noticeably larger or has a different aspect ratio, resolve that difference first so you do not mistake cropping for an accurate local edit. Also view the photo at its normal size. A defect that is barely visible when enlarged can sometimes look like a patch on a small screen. For printing, also check the detail at a size close to the intended print. ## Fix a specific defect and keep the version that works If the person has been removed but a horizontal joint in the wall stops abruptly, ask specifically for that joint to continue. Do not also change the weather, clothing colors and sharpness in the same request: that makes it harder to tell what helped and which parts of the photo changed again. Save the current version before the next attempt. Nano Banana supports successive edits in a conversation. Use this when you can identify a small, specific defect to fix. If the main subject’s face or position has changed noticeably along with the background, return to the original and define the removal area again. Continuing to edit an already distorted photo can take it further away from the original details. ## Save a separate version and open the finished file Download the result you chose and open it outside the chat. Check that the entire correct version has been saved and that its resolution suits your planned publication or print. Keep the original and the prompt together so you can repeat the task later or discard an unsuccessful edit. The generated wall or path may look plausible, but it is still new content. If you need to show the exact appearance of a monument, inscription or another obscured object, compare it with a separate photo or keep the original shot. Removing a passerby can improve a photo’s presentation, but it does not recover information that was never visible in it. In the Nano Banana section on the YesAI website, you can upload your own photo and describe the edit in a message. Our example needs just one source image: identify the person on the right to remove and the subject on the left to preserve. Explain which background details should continue into the cleared area. Before submitting, check the selected model and the payment terms in your account. [Edit your own photo with Nano Banana on YesAI ](https://yesai.su/en/nanobanana) ## Sources and documentation - [Google: image generation and targeted editing with Nano Banana](https://aistudio.google.com/docs/image-generation) - [Google: source photos and successive edits](https://blog.google/products-and-platforms/products/gemini/updated-image-editing-model/) - [YesAI: Nano Banana, source images and text instructions for edits](https://yesai.su/en/nanobanana) [All articles ](https://instain.ru/en/articles/) ## Tools mentioned in this article Stars are directory scores, not user reviews or traffic statistics. [How to interpret the scores](https://instain.ru/en/methodology/) Images ### [Nano Banana](https://instain.ru/en/ai/nano-banana/) Creating images and making precise edits with words 4.68 / 5 model Assistants ### [YesAI](https://instain.ru/en/ai/yesai/) Chat, images and videos in one browser 4.70 / 5 service