AI looks at a thumbnail of each photo and suggests a short name, a descriptive phrase, and tags in your language. You choose the naming pattern, manually correct anything you want, and download the renamed batch with a CSV matching old and new file names.
Searching for a photo by its file name is impossible when they are all named IMG_ followed by a number: the name the camera gives tells nothing about what's inside the photo.
For each photo in the batch, a single call is made to the vision model. What travels is a reduced copy of the photo —the thumbnail, cropped to 512 pixels on the long side and recompressed—, never the original file. The model returns four things at once: a short name of two to four words covering the subject, environment, and moment; a complete descriptive phrase intended as alt text; three to six single-word tags; and a category from a closed list (landscape, portrait, food, document, screenshot, animal, city, event, or object). The model's instructions prohibit it from using people's proper names, identifying anyone, inventing specific places, or copying text from sensitive documents.
With these pieces, the file name is constructed according to the naming pattern you choose: date + description + sequence number, description + number, description + size in pixels, or category + date + original name. You can also build a custom pattern by activating and deactivating segments and choosing the separator, underscore (_) or hyphen (-). The date comes from the photo's EXIF (or from Google Takeout's JSON if provided). The description segment is converted to lowercase without accents or symbols so that the name works on any system. The sequence number is padded to two digits —three if the batch exceeds one hundred photos— and if two photos end up with the same name, a correlative suffix is added. The full name is truncated to ninety characters.
Nothing is downloaded blindly: before downloading the batch, you can open the list of name changes, search for a specific photo, view the description and tags proposed by the AI, and rewrite the name manually; what you write overrides the pattern. The ZIP includes a CSV with the original name, the new name, the capture date, the size before and after, the operations applied, the description, and the tags. This same CSV can be downloaded separately. If requested, the descriptive phrase is also saved inside the photo, in the EXIF description field; however, this is incompatible with comprehensive metadata cleaning, and if you activate both, the tool will warn you and prioritize the cleaning.
Renaming proceeds photo by photo, so a difficult photo does not ruin the batch: if the AI fails or the quota for AI names is exhausted, that photo is named by its date, and the process continues with the others. It also chains with the other Google Photos tools in AiToolsForApps —reduce file size, privacy, crop, and PDF album—: you request everything in the same pass, and the server determines the correct processing order. There is no OAuth, no Google Photos API, and no gallery permission: you upload the files yourself, and both originals and downloads are deleted automatically after two hours or sooner if requested.
Use it free without an account. If you want to save your result, you can sign up afterwards.
Save / Sign upSearchable names
instead of IMG_20240815_WA0023, you get 2024-08-15_playa-atardecer-ninos_01, with the date at the beginning so that alphabetical order is also chronological.
Four patterns and one custom
date + description + sequence number, description + sequence number, description + size in pixels, or category + date + original name; and if none suits you, you choose the segments and the separator one by one with a live example on a photo from your batch.
The final say is yours
the renaming list shows you original and proposed names, lets you search, and allows you to manually type the name before downloading; your name overrides the pattern.
Description and tags included
the same call returns a phrase for alt text and three to six tags per photo, useful for accessibility or for a post caption, and you can save the description within the photo's EXIF.
The name map in a CSV
original name, new name, capture date, size before and after, applied operations, description, and tags, in a CSV that Excel opens in columns. If you change your mind, you’ll know exactly what each photo was.
The batch never gets stuck
if a photo cannot be described or the AI naming quota is exhausted, that photo is named by its EXIF date, you are notified on its detail card, and the rest of the batch continues.
The model only receives a thumbnail
the AI sees a reduced copy of 512 pixels on the long side, never the original, and only if you activate AI naming.
All in one go
renaming combines with reducing file size, removing location, cropping, and PDF album creation for the same batch, no account needed, and without touching your Google Photos.
Try it without signing up: 10 photos per batch and 5 photos with AI-generated names. Register to expand batch limits and save your naming pattern and settings for next time.
In the features that use AI, responses are generated by artificial intelligence and are not a substitute for professional advice.