# AI photo editor URL: https://dreambench.ai/tools/ai-image-editor Say what should change and it changes: the time of day, the weather, a color, an outfit, something added or taken away. The rest of the photo is left alone, and the edit is matched to the light and grain of the original. ## What is an AI photo editor? An AI photo editor changes a picture from a sentence instead of layers and brushes: you write the edit as a prompt, and an image model makes it, keeping the rest. It runs in the browser, for anyone who knows how a photo should look but can't retouch one. It is part of every DreamBench plan, from $9.99 a month. ## How to edit a photo with AI 1. Upload a photo: Any picture you have the right to edit: a portrait, a product, a room, a street. 2. Describe the change: One sentence, the way you would tell a retoucher: what changes, and where. Leave out what should stay. Then press Make the edit. The first time, create a free account and choose a plan. 3. Download the edit: The change is made and blended in; the rest is untouched. Compare before and after, then download it full size. **Choices.** - What to change: a short description, needed (up to 240 characters) ## Edit a photo with a prompt Write the change as you would say it: turn the car blue, put a red scarf on her, make the sign say OPEN. Words in quotation marks are asked for exactly as written; check the spelling in the result. One change per run works best; run the result again for the next. ## Example prompts for photo edits One change per prompt, named and placed. A color: turn the car metallic blue. The light: make it golden hour, low sun from the left. Clothing: put a red wool scarf on her. A removal: remove the trash can on the left of the sidewalk. Text: make the sign say “OPEN”. The weather: turn it into a snowy evening. ## Remove an object or a person from a photo Yes, AI can remove a person from a photo. Type remove the man in the background and the space is filled to match the scene. Because the whole picture is drawn again here, a removal that keeps every other pixel as it was is the object remover's job, or the remove people page's. Tool: [AI object remover](https://dreambench.ai/tools/ai-object-remover). ## Edit a product photo for a shop listing Change the backdrop, the surface or the colorway of a product in a sentence: put the mug on a wooden table, make the sneakers white. For a product placed in a studio or lifestyle setting with its label kept, the product photo page is made for that job. Tool: [AI product photo generator](https://dreambench.ai/tools/ai-product-photo). ## Change the time of day, the weather or a color Golden hour instead of noon, snow instead of rain, a green door instead of a gray one. Light-changing edits are applied across the whole scene, so the shadows and reflections move with them rather than the change sitting on top. ## Replace the sky in a photo A flat white sky swapped for clear blue, a sunset or a starry night, with the light on buildings and water matched to it. The sky replacement page offers six skies and keeps everything below the skyline. Tool: [AI sky replacement](https://dreambench.ai/tools/ai-sky-replacement). ## Turn a photo into a cartoon, anime or a sketch Redraw a photo in a style instead of retouching it: cartoon, anime, comic book, 3D animation, watercolor, pencil sketch or a coloring page, with the people still recognizable. Tool: [Photo to cartoon](https://dreambench.ai/tools/photo-to-cartoon). ## Works with - What it takes: JPEG, PNG, WebP or HEIC, up to 40 MB, straight from a phone or a camera. An iPhone's HEIC photo becomes a JPEG on your device as you choose it, in any browser. A picture larger than 2048 pixels on its long edge is scaled to 2048 before the run. - What you get: A JPEG in your photo's proportions. - Instagram, LinkedIn and Facebook: The edit keeps your photo's proportions, so it posts at the shape you shot it; the picture is drawn again at about 1.4 megapixels, which is more than a feed shows, and the AI image upscaler adds pixels for a print. - Canva, PowerPoint and Google Slides: Place the JPEG in a design or a slide like any photo. - Photoshop and Lightroom: Open the JPEG for any finishing by hand: the edit arrives as one flattened picture, with no layers or mask. - The DreamBench composer: Animate the edit opens the composer with the edited photo as the reference for a clip. ## Which model runs it, and where it fails The AI image editor runs on [GPT Image 2.5](https://dreambench.ai/models/gpt-image-2.5). GPT Image 2.5 Flare is the smaller of OpenAI's two GPT Image 2.5 models, built for high-quality everyday images. It takes your photo and an instruction and draws the picture again. It declines some photos under OpenAI's content rules, photos of real people among them. A run the model declines under its content rules is refunded. - Words instead of layers: No masks, brushes or layers. The instruction is read by a model that edits from references and keeps what you did not mention. - Only what you asked: The prompt locks the framing, people, objects, colors and details outside the change. - Blended, not pasted: Light, perspective and grain are matched so the edit reads as photographed that way. - One change at a time: Chain edits by running the result again. Each run does one thing well; a list of five does none of them well. **Where it fails.** - A vague request, make it better or more professional, gives a vague result: name the change and where it goes. - Several changes in one instruction are often done half each; split them into runs. - The picture is drawn again, so small text, small faces in a crowd and fine patterns can drift; compare before and after. - It edits the photo you upload and takes no second picture, so it can't copy a face, a product or a style from another image. ## AI image editor FAQ - **How do I edit a photo with AI?** To edit a photo with AI, upload it, type the change in one sentence, and press Make the edit. The model changes only what you named, such as the car's color or the time of day, and keeps the rest of the photo, matching the light and grain so the edit blends in. Make it with any DreamBench plan and download it full size. - **What kinds of edits work best, and can AI change the clothes in a photo?** Changes you can name and place: a color, the time of day, the weather, an object added or removed, a backdrop. Clothes too: put him in a navy suit, make the dress red, add a red wool scarf. The face, pose and background stay; patterns and logos on the new garment are the model's, so check them. Vague requests such as make it better give vague results. - **Can AI change the background of a photo?** Yes: say what should be behind the subject, such as put her in front of a brick wall, and the backdrop is redrawn with the light matched. For a chosen backdrop with the subject's own pixels kept, use the AI background changer; for a cutout, the AI background remover. - **Can AI add a person to a photo, or copy one from another picture?** It can add a person you describe, such as a man in a gray coat on the left, drawn to the scene's light. It cannot add a particular person from another picture: the editor takes one photo and no second image, so it can't copy a face, a product or a style from elsewhere. Describe the change in words instead. - **How do I write a good prompt for a photo edit?** Say what changes, where, and how it should look, in one sentence: turn the front door dark green. Name what should stay only when it touches the change, like the brass handle; the editor keeps the rest anyway. Put any words that should appear in the picture in quotation marks, and ask for one change per run. - **Will the rest of my photo change, and why does an AI edit look fake?** The instruction tells the model to change only what you asked for and to keep the framing, people, objects, colors and details as they are. An edit looks fake or changes a face because the picture is drawn again: small faces, fine text and patterns can drift, and a vague prompt gives a vague edit. Name one change and where it goes, and compare before and after. - **Can I make several edits?** Yes, one at a time: run the result through the editor again with the next change. Each run keeps what came before. Several changes in one instruction tend to be done half each, so a run per change gives the cleaner result, and you can stop at whichever version you like best. - **Can it edit a video?** No. It edits one still photo at a time; a video's frames would each need the same change. To get a moving version, press Animate the edit: the composer opens with the edited photo as the reference for a short clip, on any DreamBench video model. - **Can I edit a photo like this in Canva or on my iPhone instead?** Canva's Magic Edit changes a photo from a prompt inside Canva's editor and plans, on an area you brush or on the whole picture. An iPhone's Photos app adjusts light and color and removes objects with Clean Up on models that support it. This tool takes the whole photo and a sentence in any browser, and changes the light across the scene when you ask. - **Is this like Photoshop's Generative Fill?** Generative Fill repaints an area you select inside Photoshop and puts the result on a generative layer of its own, with an Adobe membership. This page takes a sentence about the whole photo in any browser and returns one flattened JPEG; for a brushed area or layers, Photoshop is the tool. - **Can I edit photos with AI on my iPhone or Android phone?** Yes, in either phone's browser, with no app to install. Choose a photo from your camera roll, an iPhone's HEIC photo included, which is converted on the phone before the run. Type the change, press Make the edit, and download the result full size. One photo per run: there is no batch upload, so edit several one by one. - **Can ChatGPT edit photos?** Yes, ChatGPT can edit photos: upload one and describe the change in a chat. This page does the same job with a fixed instruction that keeps everything you didn't mention, runs it on the model named under Which model runs it, and gives you the edited photo to download at full size. - **Can I get a PNG with a transparent background, and is there a watermark?** No to both. The edit downloads as a JPEG, which has no transparency; for a cutout of the subject on a transparent background, use the AI background remover. DreamBench adds no watermark to the download, though the image model may embed an invisible mark of its own. - **Why was my photo declined, and what happens then?** A photo is declined when the image model's content rules refuse it, photos of real people among them. The same photo and instruction are then sent once to a second image model, with no second charge; if that one declines too, the run fails and is refunded. The second model draws the picture again as the first would, so compare before and after in the same way. - **Does the edit keep my photo's resolution?** The edit keeps your photo's proportions, not its pixel count. The photo is read at up to 1536 pixels on its long edge and drawn again at about 1.4 megapixels in those proportions, so a 12-megapixel phone photo comes back smaller, as a JPEG. For a print, enlarge the edit with the AI image upscaler, which adds pixels without redrawing anything. - **Can I use the AI photo editor without paying?** No. Making an edit needs a DreamBench plan, and you create an account on the way to it. The button shows each run's price in credits before you press it, and a run that fails is refunded. - **How much does the AI image editor cost?** A run starts at 13 credits from your DreamBench plan, depending on the photo, and once you have a plan the button shows each run's price before you press it. Starter, $9.99 a month, includes 1,100 credits a month, at least 57 runs of the AI image editor, plus AI video. A run that fails is refunded. - **Do I need an account to use the AI image editor?** Yes: a free account and a DreamBench plan. The tools are part of every plan, from Starter at $9.99 a month. - **Which image files does the AI image editor take?** A photo up to 40 megabytes: JPEG, PNG, WebP, HEIC or another format your browser opens. A photo straight from a phone is fine, an iPhone's HEIC included: the page scales a large picture down to 2048 pixels on its long edge before the run, so the result is never larger than the page says. - **Can I use what the AI image editor makes commercially?** Yes. Under the DreamBench Terms you may use what a tool makes for personal and commercial purposes, including advertising, monetized content and work for clients, with no separate commercial-use upgrade. The Terms don't guarantee copyright in a result or clear another person's rights in your upload, so only use photos you have the rights to. - **Is my photo used to train AI?** No. DreamBench doesn't use your prompts, uploads or videos to train or fine-tune AI models, and any future training program would be opt-in. - **What happens to my uploaded photo?** It stays in your browser until you make a result. Then it is used to make that result and nothing else, and removed from the tool's storage after seven days along with the result. A result you keep in your library stays as long as you keep it. Runs on GPT Image 2.5. DreamBench is an independent platform; its tools come with every plan (https://dreambench.ai/pricing). Copy last checked 2026-10-22.