Who we are
Autocompressor is a free service that compresses videos, images, audio, and GIFs to a target file size. Depending on what file type you upload, it will be handled differently such that the result serves a similar function to what you uploaded (e.g. videos are transcoded to videos, audio files to audio files, images to images, and animations to animations as opposed to videos). Autocompressor intends to compress files more optimally than other online services using AUTO-REZ™ media compression technology.
What AUTO-REZ™ technology is
AUTO-REZ™ is the file compression backend that works to optimize your videos, images, audio, and GIFs. When you use AUTO-REZ™, you can trust that it will compress your files just as well as you could have done it using advanced tools manually on your computer in the same amount of time, thanks to our advanced algorithms for video resolution, audio bitrate, image quality, and codec settings. AUTO-REZ™ is highly efficient and features optimizations that waste less time while obtaining better results compared to other online services. AUTO-REZ™ has unique features that replicate what video encoding professsionals do when handling files, such as copying compatible audio tracks if they are a reasonable bitrate.
What we're using
AUTO-REZ™ is proprietary software and there is no tool available that can automatically replicate AUTO-REZ™'s results. Behind the scenes, the algorithms are choosing settings for:
FFmpeg (ffmpeg.org), a complete video and audio decoding, processing, and encoding solution
ImageMagick (imagemagick.org), a complete image decoding, processing, and encoding solution
Gifsicle (lcdf.org/gifsicle), a GIF optimizer and preprocessor
MozJPEG (MozJPEG GitHub), a higher compression JPEG encoder/lossless optimizer
ECT (ECT GitHub), an extremely efficient lossless PNG optimizer
AUTO-REZ™ performance vs other compressors
AUTO-REZ™ performs better than the leading online compressors.
In this test, we cut the first 5, 10, 15, 20, 30, 45, 60, 80, 100, 120, 150, and 180 seconds from Rick Astley's Never Gonna Give You Up, as well as encoding the full 212-second video.
The source is YouTube's H.264 1080p video encode and their 128 kbps Opus audio encode. Download source. Segments were cut losslessly and without transcoding using MKVmerge. VMAF was measured using FFmpeg.
VMAF is a metric developed by Netflix that approximates visual quality on a scale of 0-100.
AUTO-REZ™ was tested using H.264/AAC "Normal", 8mb.video was tested using "Extra Quality", FreeConvert.com has no settings aside from target size 8 MiB.
None of the services overshot the target size of 8 MiB while encoding these test clips.
You can see that at 5 and 10 seconds, AUTO-REZ™ passed through the video track resulting in perfect quality, while re-encoding the Opus audio to AAC for compatibility with MP4. The other services re-encoded the video track and actually made it bigger, even though it was already H.264. FreeConvert naively uses the source resolution of 1080p, resulting in rapid quality loss at lower bitrates. 8mb.video naively uses 720p for all videos, resulting in lower quality than FreeConvert at high bitrate but a less steep drop-off. Both services would clearly benefit from AUTO-REZ™'s resolution selection technology, since the gap widens at low bitrate as they both use too high of a resolution for the bitrate. Important note: 8mb.video recently added a feature that increases the output resolution to 1080p when the output bitrate is more than 4 Mbps, which would make its performance at ≤15 seconds similar to FreeConvert. This does not affect its performance on longer lengths.
BD-Rate refers to the relative amount of data required to represent a given quality. A negative BD rate means that it requires that much less data to get the same quality. At 1 minute, AUTO-REZ™ has a -23% BD-rate adavantage against 8mb.video XQ and a -37% BD-rate advantage against FreeConvert. At 3 minutes and 32 seconds, the length of the music video, the difference is very dramatic: -29% against 8mb.video XQ and a whopping -61% against FreeConvert, meaning it's as good at 3:32 as FreeConvert is on a 1:23 video.
AUTO-REZ™ will never output a file that is larger than the requested target size. We achieve that by adjusting the output until it reaches the target size if the first attempt is not successful. Because target size applications are very strict, we'll throw an error rather than output an oversized file if push comes to shove. If you find a file that causes Autocompressor to produce an oversized file, join our support Discord and send it over, and we'll figure out why.
Automatic Output Format Selection
Autocompressor can handle video, audio, image, and animation inputs. In addition to that, we only need to know the category of formats that you want as output. We separate all media files into four different types: "video" (like .mp4), "audio" (like .ogg), "image" (like .jpg), and "animation" (like .gif), and handle those types differently. For example, if you select the MP4 category and upload a GIF, we will output an APNG, if you upload a PNG you will get a PNG or JPEG depending on which is small enough, and if you upload a WebM you will get an MP4. Each category is its own tier of compatibility with hardware and software, and for most purposes you can leave it on MP4/M4A/PNG/JPEG/APNG and get the results you want.
Widest Video Quality Range
AUTO-REZ™ can encode videos in an ideal way throughout the entire useful quality range for videos. The combination of resolution selection and logistic audio bitrate means that what AUTO-REZ™ does makes perfect sense no matter what the target bitrate is. This is in contrast to other sites, which have logic that breaks down at both extremes of quality. It's important to note that no website can increase the quality of your upload, and only an 8K source can take advantage of 8K encoding.
We use a low-cost innovative method to determine how complex your uploaded video is to compress. In general, the higher the complexity, the lower the resolution of the output should be, given the same bitrate. For example, a live action, high motion video is significantly more complex to compress, and will therefore be given a lower resolution, than a PowerPoint presentation. The aim with complexity based resolution selection is that all videos of the same length and filesize get the same "quality per pixel". This method usually does a better job of selecting the correct resolution than even video encoding professionals and movie piracy groups. Our complexity based resolution selection does cost some processing power, so for the "Fastest" preset, we disable it and use a very simple bitrate-based method. For both resolution selection methods, we take into account the format and compression level to adjust the resolution of the output. Correct selection of the resolution significantly improves the quality against naive methods like using a fixed resolution or the source resolution.
Whenever you upload a video file to Autocompressor, we analyze its tracks to determine if they are already suitable for the output format, and if they are, we simply copy them over. We don't just look at the codec, we also consider its level of compatibility with other software. For example, if your video contains an HE-AACv2 audio track at a low bitrate and you're outputting to an MP4 file, we won't copy the track, instead re-encoding it to the more compatible AAC-LC. Suitability is also determined by the bitrate that we would have used otherwise. In general, if you upload a video to Autocompressor that was previously compressed to the same size and format by Autocompressor, you'll get the same file back. Most of the time, Media Passthrough helps for phone and camera videos that tend to use low-bitrate AAC that need not be re-encoded.
Autocompressor has the best handling of audio in a video of any website in existence. Most video conversion websites use a fixed percentage of the total bitrate for audio, perhaps with a bitrate cap. However, Autocompressor uses a logistic function to decide the audio bitrate, which means the audio bitrate rises quickly with the total bitrate at low bitrates and then tapers off at high bitrates when the audio becomes indistinguishable. We adjust the parameters of this function based on the audio codec being used. For example, Opus is not useful below 6 kbps and becomes fully transparent for stereo audio at 192 kbps, while AAC-LC is useful down to 2 kbps but becomes transparent for stereo audio at 256+ kbps. In addition to that, we support 5.1 and 7.1 surround audio and use a piecewise function to determine when to enable them for the output. At the other end of the spectrum, we automatically downsample and downmix the lowest bitrate audio to improve efficiency. For any video source, we aim to emulate as closely as possible what video encoding professionals and movie piracy groups choose for the audio configuration.
Smart Audio Adjustment
Smart Audio Adjustment (SAA) is a new AUTO-REZ™ feature that combines the best of our Resolution Selection and Audio Handling. In SAA, we compare the result of the complexity based resolution selection to the result of the bitrate based resolution selection. If the complexity based resolution is higher, that means the video is low complexity compared to a typical video, which often indicates that audio needs to be prioritized. For example, a PowerPoint presentation will be flagged by this test, and in that scenario, voices need to be clearer. That's why SAA can boost the audio bitrate by up to 60% when the video is low complexity. Because of this, your presentations, songs with thumbnails, talk shows, and screencasts will have much clearer audio at the expense of a small amount of video quality. It's important to note that SAA can only work when complexity based resolution selection is used. The "Fastest" preset, which uses only bitrate based resolution selection, cannot take advantage of SAA technology.
Unique Target Size Encoding Capabilities
It's easy to find a website that can "compress a JPEG", "convert video to GIF", "convert JPEG-XL to AVIF", or "convert M4A to OGG", but often much harder to find one that can do that conversion with a target size. We've seen some sad excuses for online target size GIF converters before, so we aimed to fix that. Autocompressor can encode to every major format with a target size and do it well. For example, you can directly paste in a Reddit video link, select "GIF", and quickly get a perfect 25MB GIF to post in the memes channel. We use Gifsicle when encoding GIFs on the "Normal" preset to maximize quality. Our PNG+JPEG encoding mode will try to optimize your PNG before encoding a JPEG, and when it does encode a JPEG, it uses MozJPEG technology to maximize quality. In addition, Autocompressor supports target size encoding to M4A and Opus OGG, as well as WebP, AVIF, APNG, Animated WebP, and even Animated AVIF. Those formats were all very difficult or even impossible to find an online target size converter for, until we made one. All images are encoded to an approximately correct resolution based on our image compression research so you always know you're getting the best possible quality.
Discord Sticker and Emote Encoding
Discord emotes, and especially stickers, can be a huge pain to compress properly. That's why we created special cases to compress your animated emotes to a 256 KB GIF up to 128x128 and your still emotes to a highly optimized 128x128 PNG. Additionally, we couldn't find any websites that can automatically compress the APNG stickers you need, so that's why we created a tool to instantly convert them to a 512 KB APNG at 320x320 or 160x160. It's the easiest possible way to get good quality stickers for Discord.
New Transcode Servers
We've just got a new server with more processing power than the web server. We handle transcodes in the most secure way possible: at no point is your video sent to the web server, and at no point does the transcode server have access to the web server's database. The client asks the web server to open a transcode slot on the transcode server, and then the web server checks the database to make sure that you're allowed to use the settings you specified. Then, the web server authorizes a transcode on the transcode server with the settings, and the client uploads the video to the transcode server. The client does not directly tell the transcode server what settings to use. It is also impossible to "pirate" Autocompressor by using DevTools or an extension to expose premium settings, because the server will reject their use. If an extension claims to be able to get premium features for free, it is a scam!
Unfortunately, no video conversion service can survive without serving ads. We use Google AdSense to serve ads. We promise to avoid, as much as possible, the serving of ads that push scams like search extensions, bad file converters, fake download buttons, and straight up malware. We're aware of how media conversion websites are targeted by such advertisements. If you find ads that violate these guidelines, send us a screenshot of the ad as well as its destination URL at email@example.com and we'll ban that advertiser from our platform. We have absolutely zero tolerance for these types of scams. Note, however, that we're not referring to ads for services that offer things like video encoding for your streaming service. These ads are completely acceptable and Autocompressor users may be particularly interested in them.
The Pricing page currently contains tentative prices for premium service that are subject to change at any time. We need to develop a system to integrate PayPal into our service in order to serve premium services. We're working on our accounts before we do this.
The AUTO-REZ™ API will allow you to integrate AUTO-REZ™ video, image, audio, and GIF compression into any application, whether that's a Discord bot, a CDN, or even another online conversion website. This API will be available to Professional and Enterprise tier users and it will include features not available via the standard web interface, such as lax filesize limits (useful for %-of-original-size and similar applications), forced conversion to an image, audio, animation, or video instead of the source type, and the ability to prevent special cases like Emote encoding. Using the API will also allow you to specify that we POST your result to a specific target, so that your application can easily process the result without waiting for it. If you run a conversion website that's broad in scope and want to avoid having to implement target size encoding for various formats, the AUTO-REZ™ API can help lighten your workload.
Do you have questions about how Autocompressor works, our goals, or other information? Email us at firstname.lastname@example.org and we'll try to answer as soon as possible.