# TumblThree: archiving Tumblr, Twitter and Bluesky blogs to disk > TumblThreeApp/TumblThree is a C# WPF desktop application that downloads posts, media and metadata from Tumblr, Twitter and Bluesky blogs into local folders. It is a full rewrite of TumblTwo and remains a Windows-only tool with no headless mode. **TumblThreeApp/TumblThree** — A Tumblr and Twitter Blog Backup Application - Repository: https://github.com/TumblThreeApp/TumblThree - Website: https://TumblThreeApp.github.io - Stars: 742 · Forks: 87 - Language: C# - License: MIT - Published: 2026-10-21 · Updated: 2026-10-21 · Language: en - Canonical page: https://hysenlabs.com/projects/tumblthreeapp-tumblthree ## A desktop archiver for blogs, not a general purpose scraper TumblThree downloads photo, video, audio and text posts from a given Tumblr or Twitter blog and stores them locally. The scope is deliberately narrow: you point it at blogs and it mirrors them. The application is written in C# using WPF and the MVVM pattern, built on the Win Application Framework, and the project describes itself as the code rewrite of TumblTwo, its own predecessor. That rewrite history explains the feature list. TumblTwo was a simpler tool, and TumblThree keeps the basic job while adding a queue, collections, a clipboard monitor and multiple network sources in the same window. The rewrite is also why the repository carries a second licence file, LICENSE-3RD-PARTY, next to the main MIT LICENSE, and why the repository is split into src, lib, scripts and docs directories. The application is not a headless tool. There is no command line interface described, no configuration file format, and no service mode. Everything happens through the main window and its settings panel. The project keeps an active release cadence, with v2.23.0 published on 2026-10-14 after v2.22.0 on 2026-09-06 and v2.21.0 on 2026-07-11, and the last push was on 2026-10-18. ## What gets pulled down for a Tumblr blog The per-blog feature set is where the real work is. A blog download handles photo, video, text, audio, quote, conversation, link and question post types, and it stores meta information for photo, video and audio posts alongside the media. Posts that embed media, such as photos inside a question and answer post, are handled with the inlined photos and videos downloaded too. On image sizes, the documentation is honest about the ceiling. All image sizes possible are downloaded, with the API variant available only for newer blogs, and higher resolution simply is not possible for old blogs. That limit comes from the source side, not from the application. Beyond plain posts there is a set of narrower modes. You can restrict the download to original content and skip reblogged posts, filter to tagged posts, limit the download to specific blog pages, define a time span, or download hidden blogs that need a login, and password protected blogs of non-hidden blogs are also supported. Linked files hosted elsewhere are covered as well, with support for Imgur, Gfycat, Webmshare, Uguu and Catbox. A separate option skips a file if it has already been downloaded before in any currently added blog, which is what makes overlapping backups cheap. ## Likes, searches, communities and other network sources Beyond direct blogs the application downloads several other kinds of content, each with its own login requirement. There is a downloader for liked-by pages, such as a user's liked-by page, and for the likes page, both of which need a login. Tumblr search, Tumblr tag search and Tumblr Communities each get their own downloader, and the tag search and Communities modes are marked as requiring a login while plain search is not. All of them can be limited to a defined time span and download all possible image sizes. Twitter blogs and Bluesky blogs are handled as first class sources with photo, video and text posts, meta information for photo and video posts, the option to keep only original content, tagged post filtering and time span limits. Downloads of Bluesky user feeds are listed as a feature alongside the other sources. Newer additions sit on top of that. Download of Tumblr Communities is marked as recent, and there is an option to get Gif images instead of the WebP or Gifv variants that Tumblr serves by default. The feature list also includes a preview of photos and videos, an image viewer with slideshow mode, and a file rename function. ## The queue, the throttler and the scheduled start The download engine is organised around a queue. A download queue is a listed feature, the queue list is autosaved, and it can be saved, cleared and restored, which is what makes a long backup survive a restart. Multiple concurrent downloads of a single blog and of different blogs are supported, and the number of concurrent downloads is one of the settings you can change. A bandwidth throttler is included, and the application uses SSL connections and the Windows proxy settings, so it inherits the proxy configuration of the machine it runs on. The `https_proxy` style environment variable is not part of this project's documented surface, the Windows proxy settings are. There is a scheduled start time, described as being able to set a start time for an automatic download, for example during nights. That is aimed at desktop users rather than servers, and it is one of several features that assume somebody is sitting in front of the machine. An option to download a URL list instead of the actual files is worth calling out, because it changes what a backup is. It gives you the manifest of what is out there without pulling the media down. ## Clipboard monitoring and blog collections Two workflow features deserve their own mention because they change how the tool is used day to day. The clipboard monitor detects blogname.tumblr.com URLs in the clipboard, which means copying a blog address and pasting it anywhere is enough for the blog to be added to the bloglist automatically. That removes the tedious step of adding a source by hand, and it is the kind of small automation that makes a backup application something you keep running. Blog grouping is the other one. Blogs can be grouped into collections, so a large archive can be organised by subject, by follow type or by whatever scheme suits you. Settings of multiple selected blogs can be changed at once, which means a change to concurrency or naming does not have to be repeated blog by blog. The settings panel itself covers the download location, turning the preview on or off, the number of concurrent downloads, the image size of downloaded pictures, download defaults and portable mode. Portable mode matters for anyone moving a large archive between machines, since it keeps configuration alongside the application rather than in the user profile. ## What this costs you in platform and maintenance terms There are real constraints, and they are structural rather than fixable. The application is built on WPF, and the topic list includes windows and wpf, with the CI configuration in appveyor.yml. A WPF desktop application does not run on Linux or macOS, and there is no console mode described in the documentation. For a large archive scheduled to run nightly, that rules out the deployment most people would want. The source side is the other risk. Several download modes need a login, including hidden blogs, liked-by pages, likes, tag search and Communities, and those logins are not automated by the application as far as the documentation describes. When a platform changes its endpoints or its login flow, a desktop tool like this needs a code change and a new release. What you get in exchange is a local, browsable archive with media files on disk rather than a database dump, plus a slideshow viewer and a preview panel for checking what arrived. Internationalization is supported with several languages available, and the project ships its own releases, so no separate build step is required to get a current binary. ## Conclusion Use TumblThree if you want an offline copy of blogs you follow, especially Tumblr blogs whose media would otherwise disappear, and if you are happy working on Windows with a desktop window and a settings panel. Do not pick it for a headless server, a scheduled job on Linux, or a multi-user archive, because the project ships as a WPF desktop application with a graphical queue and slideshow viewer. Verify first the release you download, since the latest published version is v2.23.0 from 2026-10-14, and check whether the sources you care about are still reachable without a login, because hidden blogs and several search modes require credentials you must supply yourself. ## FAQ ### Where do I download TumblThree and is there an installer for Linux? Latest versions are published on the GitHub Releases page of the project, and the project builds a WPF desktop application for Windows, with AppVeyor CI and a Windows platform topic. The documentation does not describe a Linux or macOS build. ### Does TumblThree need a login to download a blog? Plain blog downloads do not require one, but the liked-by and likes downloaders are marked as login required, as are tag searches and Tumblr Communities. Hidden blogs also require a login, and password protected blogs of non-hidden blogs are supported. ### Can TumblThree resume a large download queue after a restart? Yes. The download queue list is autosaved, and the documentation lists saving, clearing and restoring the queue list as supported operations, so an interrupted backup can be picked up again. ## Sources - [License: MIT](https://github.com/TumblThreeApp/TumblThree/blob/master/LICENSE) - [Project website](https://TumblThreeApp.github.io) - [README](https://github.com/TumblThreeApp/TumblThree/blob/master/README.md) - [Releases](https://github.com/TumblThreeApp/TumblThree/releases) - [TumblThreeApp/TumblThree on GitHub](https://github.com/TumblThreeApp/TumblThree) --- Hysen Labs editorial analysis, written from the project's own repository and release notes. Cite the canonical page: https://hysenlabs.com/projects/tumblthreeapp-tumblthree