HandBrake

Back around 2005 or 2006 I had a lot of scratched DVDs (I still have them). The disc drives in my PCs at the time were loud, slow and temperamental. I killed two of them. The cheap ones (usually bought in bulk from a large supermarket) lasted me maybe eighteen months before they started giving me read errors on clean discs.

That is the problem Handbrake solved for me. Eric Petit (who went by "titer" online) started writing it in August 2003 on BeOS (it was an operating system) of all things, then ported it to macOS. His one goal was making DVD rips to early iPods and Apple TVs painless. Twenty years later it is still one of the best tools.

Quick history - funny.

In April 2006 Petit committed his final SVN revision and just vanished. Gone. No admin access handed over to anyone. Meanwhile Apple dropped iPod firmware 1.2 which needed new H.264 profiles and nobody could push code to the main repo. Two devs named Rodney Hester and Chris Long got fed up, branched the code and called their fork "MediaFork" in early 2007. It was better than the original. A month later Petit reappeared, saw what they had done, said fair play and handed them the keys. That is how Handbrake got its current decentralised maintainer structure. Pure accident.

Another true story - not so funny (you can skip this part if you don't want any drama)

I am laughing about the Handbrake story but imagine there was even a more popular free, open-source project that had a similar story. Long story short - while it became one of the top 20 most downloaded free projects the developers started to go crazy on which features and roadmaps should they implement. Half created a fork that never managed to replicate the success of the original one and the project manager decided to abandon the project - just like that. I won't give names but I want to smash the keyboard - there's no one to blame here, except for a popular platform that acted as a parasite for decades while some people admired them. If only they knew.

Back to HandBrake

It took them thirteen years to reach version 1.0. Thirteen. But let me be honest - it is just a number and kudos for respecting the naming convention. They finally hit it in December 2016. No corporate parent, no foundation, no funding structure, just volunteers. Bradley Sepos, Scott (sr55) on the Windows UI, Damiano Galassi, John Stebbins. These names mean nothing to most people and they built one of the most used pieces of media software on earth. So, I think they deserve some credit. Full disclosure: I apologize upfront for missing the other people that put the effort - just let me know and I will update this list.

One thing people get wrong constantly. Handbrake is not "just an FFmpeg wrapper".

You see people on forums complaining that HandBrake is just a fancy GUI wrapper for FFmpeg. Yes, it uses libavformat to demux containers and libavcodec for initial decoding. But sitting on top of those is a core library called libhb which is where the actual engineering lives. libhb manages stream extraction, fixes broken metadata, handles audio and video sync, does variable to constant frame rate conversion, drops duplicate frames and patches corrupted ones. It is the reason Handbrake is famously bulletproof against weird source files that would crash a raw ffmpeg command.

This bulletproof design has one massive downside.

Handbrake cannot do stream copying. It cannot swap an MKV for an MP4 by simply changing the container. It forces every file through a full decode, filter and re-encode cycle. Want to put your MKV onto your smart TV? Handbrake will take four hours and burn your CPU to do it. This is the single biggest source of confusion in the user base. People expect a "convert container" button. There isn't one. If you only need to remux, use MKVToolNix (my favourite and an amazing tool) or raw ffmpeg.

Ignore anyone who tells you to "just use FFmpeg" instead.

These are the same people who tell you to "just use git commands" instead of a Git GUI. Technically correct and completely useless for 90% of actual humans. Writing a single ffmpeg command for a modern encode means juggling Pixel Aspect Ratio preservation, BT.2020 colour matrices, detelecine pullup filters, audio track mapping with downmix codecs and metadata flags, all strung together in a shell script that you will never want to touch again. Handbrake does this with three clicks. That is not a compromise, that is a useful abstraction.

The Rate Factor slider confuses everyone.

You open Handbrake, see a quality slider labelled RF, drag it to the right thinking you are getting better quality and end up with a 40GB output file. The RF scale is logarithmic. Lower number means higher quality and exponentially bigger files. Drop RF by six and your file size roughly doubles. Raise RF by six and it halves. The UI does not explain this at all.

For H.264 (x264) at 1080p or 720p, RF 18 to 22 is the visually lossless range. Standard definition DVD rips are fine around RF 20. For H.265 (x265) at 4K, RF 22 to 28 is fine and most people won't see the difference from the sofa. Do not compare RF values across codecs. RF 22 in x264 has nothing to do with RF 22 in SVT-AV1. Different codec, different maths, different output.

GPU reality check for newcomers.

Half the people coming to Handbrake have just bought a £1500 RTX 4090 and expect 4K encoding in ten minutes. They load up x265 on the slow preset and watch in disbelief as their 24-core CPU maxes out while the 4090 sits idle. The software encoders (x264, x265, SVT-AV1) are CPU only by design. They bypass your GPU because they are optimised for maximum compression efficiency, not speed. The GPU hardware encoders (NVENC, QuickSync, AMF, VideoToolbox) use dedicated silicon that does the opposite. Fast as hell but bigger files for equivalent quality.

Benchmarks for AV1 so you understand the tradeoff. Intel Core i9 running SVT-AV1 on the CPU hits about 27 frames per second on a 1080p source. NVIDIA NVENC AV1 on the same file hits 300 frames per second. Eleven times faster. But at a constrained 3 Mbps target, SVT-AV1 preset 3 gets a VMAF score of 75.7, while NVENC AV1 lands around 71. Hardware is fantastic for quick encodes where speed matters more than storage. Software is mandatory for proper archival.

Oh and the old forum line about NVENC being rubbish? That was true during the Kepler and Maxwell GPU generations. Not true since Turing. Modern NVENC from RTX 20-series onwards matches x264 medium preset quality at reasonable bitrates. Stop repeating wisdom from 2014.

If you think about Apple products

Just in case you're wondering if Apple products such as a MacBook Pro - in this case - I use an M3 Max, let me tell you that the Nvidia RTX 4090 is faster in almost all scenarios. Yes, it does the job well but it is not a match for the raw power and the hardware encoders of the RTX 4090. I didn't test the M5 Max as I have the Air version and it doesn't make any sense - Air is definitely slower.

One HEVC trap people fall into.

If you encode short clips at 1080p or lower, H.265 can actually produce bigger files than H.264. The codec has algorithmic overhead that only pays off at higher resolutions or longer runtimes. Do not blindly pick x265 for your phone video clips. 4K yes. Two minute TikToks no.

If you own DVDs or Blu-rays, read this carefully.

DVDs use CSS encryption. Handbrake used to have CSS decryption baked in but pulled it out years ago to avoid DMCA Section 1201 liability. You have to install libdvdcss separately from VideoLAN. On Windows it is libdvdcss-2.dll dropped into the Handbrake directory. On macOS it is libdvdcss.2.dylib. Legal status varies by country so check your own laws. The developers cleverly offloaded the legal risk onto the user.

Blu-rays are a different problem entirely. Handbrake cannot decrypt Blu-ray AACS or BD+ protection. There is no plugin. You cannot make it work.

An alternative and legal warning - please respect the laws in your country.

Alternative: You need MakeMKV. MakeMKV does a lossless bit for bit extraction of the disc, strips the encryption and wraps the HEVC stream, TrueHD audio and PGS subtitles into a huge MKV file, usually 50 to 80 GB for a 4K UHD disc. That file is your master. You then feed that MKV into Handbrake for actual compression. For a 4K HDR film the good archival settings are MKV container, H.265 10-bit, RF 22, slow preset, passthrough the primary lossless audio track (DTS-HD MA or TrueHD) and add a downmixed AAC 2.0 track for stereo playback. You end up with 18 to 25 GB per film, visually indistinguishable from the source disc.

Most reviews of Handbrake completely miss this. They tell you how to "rip Blu-rays with Handbrake" and leave users confused about why the software cannot read their disc. Handbrake on its own cannot rip a modern Blu-ray. Accept it and install MakeMKV.

Again, some users will use MakeMKV to extract the raw video then they use Handbrake to compress it which will also bypass the need for libdvdcss entirely.

There was an incident in 2017 worth knowing about (and it repeated with many free projects - that's why you need an external mirror such as dAppCDN)

Between May 2nd and May 6th of that year somebody compromised one of the Handbrake download mirrors and replaced the macOS installer with a trojanised version containing the OSX.PROTON remote access trojan. Anyone who downloaded Handbrake during those 96 hours and clicked past the Gatekeeper warning gave the attackers root access to their Mac. The malware harvested keychain data, browser passwords and encryption keys. Nasty. The team purged it within days, hardened their infrastructure, started publishing SHA256 checksums and signing releases with DSA signatures. A similar thing happened months earlier with Transmission (also originally authored by Petit) which got hit with KeRanger ransomware and Keydnap keylogger.

Remember!

Always verify your checksums when downloading installers. Always check the files using VirusTotal or Jotti's malware scan Download Handbrake from dAppCDN (as we integrate third-party malware scanners)

The 10-bit secret nobody tells you.

If you select the x265 10-bit encoder for a standard SDR Blu-ray (not HDR), the internal colour precision goes up dramatically. This kills the colour banding you see in dark scenes and wide skies, that ugly posterisation where gradients break into visible steps. Even better, because 10-bit algorithms are mathematically more efficient, your file is often smaller than the 8-bit version at equivalent quality. The only catch is that old phones and tablets from before 2018 might lack hardware 10-bit decoders and will have to decode in software, which destroys battery life and causes stuttering. For anything modern, always use 10-bit. HDR handling is where things get complicated and slightly dodgy.

HDR10 metadata passes through fine if you use any 10-bit encoder. Dolby Vision is where you hit walls. Handbrake parses Dolby Vision Profiles 5, 7.6, 8.1 and 8.4 through the x265 and SVT-AV1 software encoders. Profile 7, the one on commercial UHD Blu-rays, contains something called a Full Enhancement Layer. Handbrake strips the FEL and converts the stream down to Profile 8.1 base layer. Works on more devices this way, but videophiles who obsess over bit for bit accuracy hate it. If you want the FEL preserved you need different tooling. Hardware encoders are even more limited. They write HDR10 metadata into the container only, not into the bitstream itself, which can cause playback glitches on certain TVs.

Filters quickly because this is where Handbrake actually beats the competition.

For deinterlacing old DVDs and VHS captures you get Yadif (fast and old), Bwdif (modern improvement on Yadif with better detail retention during motion) and Decomb (which only applies deinterlacing to frames that actually have combing artefacts, leaving clean frames untouched). Decomb is the winner for mixed content. Yadif is fine if absolutely everything is interlaced. Bwdif is what you want for VHS captures where you are trying to preserve that 59.94 field motion fluidity. Detelecine is the filter to use when you have 29.97 NTSC content that came from 24 fps film with 3:2 pulldown. Handbrake uses the pullup algorithm to surgically remove the pulldown padding and restore the original 23.976 fps film cadence. This makes old DVDs of films look cinematic instead of judder-riddled.

Denoising matters because film grain costs you enormous amounts of bitrate if you try to encode it accurately. hqdn3d is the fast option that smears details into a plastic looking mess. NLMeans is the slow option that analyses pixel patches across multiple frames to find similarities and averages out random noise while preserving real detail. NLMeans will cut your encoding speed in half but the output is worth it for anything you care about.

If you hate Apple, skip this paragraph.

On macOS Handbrake can hook into Apple's proprietary CoreAudio AAC encoder which produces noticeably better audio quality at lower bitrates than the open source avcodec AAC implementation. This is the one legitimate reason to run Handbrake on a Mac over Windows for pure audio work. Do not re-encode already lossy audio either. If your source is 192 kbps AC3, passing it through to 128 kbps AAC will sound tinny and metallic and there is no fix. Either passthrough the original or encode to something higher.

Subtitle handling is awkward and it is not Handbrake's fault. Text subtitles (SRT, ASS, SSA) pass through to any container. Bitmap subtitles (PGS from Blu-ray, VobSub from DVD) work fine in MKV but the MP4 container refuses to embed them. If you put PGS into an MP4 output, Handbrake burns the subtitles permanently into the video frames. They cannot be turned off anymore. This is a container specification limit. Use MKV if you want toggleable subtitles.

The Live Preview feature is magic and almost nobody uses it.

You can select a 10 to 30 second slice of your source and run a full encode on just that slice. This lets you A/B test RF values, denoise strengths and deinterlacing parameters without committing to a six-hour render only to find out you picked the wrong settings. Do ten minutes of preview testing before starting a long encode. It will save you hours of wasted CPU time.

For power users and home server operators there is a JSON based API and the HandBrakeCLI command line tool with full feature parity with the GUI. Every setting you click in the graphical interface can be exported as a .json preset file and dropped onto a headless Linux server.

Finally

I am not sure if you've noticed but there is an entire industry of paid video converters that dominates the search results. I hope the AI adoption will change this - from this perspective let the search engines burn.

There are many and the only reason I don't list them here is to avoid a lawsuit. These companies pay for SEO aggressively and write articles about "the best video converter" (add the current year) that conveniently recommend their own products. Most of these tools throttle CPU usage, insert watermarks unless you pay, lock HEVC behind subscriptions and internally use older less efficient ffmpeg libraries than Handbrake does. I guess by my personal standards I could call them adware but maybe I am just exaggerating. Ok, let me say it: most paid converter apps are a graveyard of adware.

Handbrake is the free, open source, no watermark, no subscription. Do not pay for what is already free and better :)

If you have a collection of scratched DVDs, a hard drive full of old camcorder tapes or a 4K Blu-ray collection that deserves to live on your server, install Handbrake, install libdvdcss separately if you need DVD decryption, install MakeMKV alongside if you have Blu-rays, and start archiving. The project has no foundation and no corporate funding. Amazingly and as a comparison with the bad paid products - HandBrake doesn't allow donations. Following their public page and guidelines please consider donating to the VideoLAN non-profit organization and the Blender Foundation. What a team of fine gentleman!

No release notes available