I let this free tool re-encode my Plex library overnight, and my media drive is half empty again

Adding new content to your Plex library is so easy that you don’t realize how quickly you’ve accumulated terabytes’ worth of movies and shows that are just rotting on your drives. I have about 7TB of data in my Plex library, including more than 100 shows with over 8,000 episodes and 453 movies. But my collection has now reached the point where my drives don’t have enough space for my photos and other documents. I could add more storage, but that costs money, and adding another drive to my existing setup is a bit cumbersome. So I let Tdarr work through the library overnight, and by morning, it had freed up close to 3.5 TB without removing anything from Plex.

What Tdarr Does

Tdarr is a conditional transcoding application that monitors a media library and processes files according to rules you define. It uses FFmpeg or HandBrake for the actual conversion, while Tdarr manages the queue, checks each file, and tracks whether the job succeeded. You can tell it to convert H.264 video to H.265, leave existing HEVC files untouched, preserve selected audio and subtitle tracks, and reject outputs larger than the original.

A typical installation has a Tdarr Server and at least one Tdarr Node. The server hosts the web interface, library database, and job queue, while the node performs the conversion. Both can run on the same machine, or you can install additional nodes on more powerful computers and let them share the workload. My Ugreen DH4300 supports compatible ARM64 Docker containers, so I can host Tdarr there and add another node if the NAS processor isn’t fast enough for the entire queue.

You can also consider Unmanic and FileFlows. Both offer similar features, and I’ve used Unmanic before — it gets the job done almost as well as Tdarr. I’d also recommend FileFlows, but you need to pay to get the most out of it because the free version is limiting.

Scanning the Library

Tdarr’s initial scan showed that most of my library was made up of 720p and 1080p H.264 files. The tool reads each file with tools such as FFprobe and MediaInfo, then records its codec, bitrate, resolution, container, and available audio and subtitle streams. Around 6,700 episodes and movies matched the rules I had created.

In my library, older TV shows made up most of the queue. Many of these episodes used H.264 at relatively high bitrates, even though the resolution was limited to 720p or 1080p. I skipped files already encoded in HEVC or AV1, and also excluded 4K HDR content because I didn’t want to risk changing its color information or image quality. My flow targeted the H.264 video stream while keeping the original resolution and frame rate. I copied compatible audio and subtitle tracks into the new file without being re-encoded.

Tdarr can’t look at a video and decide that it is wasting storage. It follows the codec, bitrate, file-size, and resolution conditions you give it. For example, a grainy film and a clean digital animation can respond very differently to the same encoder settings, even when their original files look similar on paper.

Running the Overnight Queue

I used my M5 Mac for encoding and kept the media on the Ugreen NAS. I installed a Tdarr Node on the Mac, mounted the NAS library over the network, and used the Mac’s internal SSD as the temporary transcode cache. I used two workers to convert H.264 video to H.265 through FFmpeg’s hevc_videotoolbox encoder, which uses Apple’s hardware video engine. Tdarr copies the original audio and subtitle streams into the new MKV files. Each output also has to pass a health check and come in at least 10% smaller than the original.

I moved the largest movies and TV seasons to the front of the queue, so the overnight run processes the biggest files first. Of course, the Mac had to remain plugged in with sleep disabled throughout the run.

Results and Quality

I’ll admit hardware encoding made this practical, although re-encoding compressed video always carries some risk of quality loss. I checked dark scenes, fast movement, and older grainy episodes after the conversion. Most looked unchanged from a normal viewing distance, while some fine grain appeared softer when I compared paused frames.

I wouldn’t use Tdarr on the only copy of an irreplaceable video. For replaceable Plex media, recovering around 3.5 TB was worth leaving the Mac busy overnight.