How Meta introduced AV1 to Reels

• We’re sharing how we’re enabling manufacturing and supply of AV1 for Fb Reels and Instagram Reels.
• We imagine AV1 is essentially the most viable codec for Meta for the approaching years. It affords larger high quality at a a lot decrease bit charge in contrast with earlier generations of video codecs.
• Meta has labored intently with the open supply neighborhood to optimize AV1 software program encoder and decoder implementations for real-world, global-scale deployment.

As folks create, share, and devour an ever-increasing quantity of on-line movies, Meta is working to develop essentially the most bandwidth-efficient methods to transcode content material whereas sustaining cheap compute and energy consumption ranges. Selecting essentially the most acceptable video coding codecs — the algorithms for compressing and decompressing the file — is essential. Over the previous twenty years, researchers have developed video coding requirements with ever-higher compression effectivity, together with AVC, HEVC, and VVC, developed by MPEG/JVET, and VP9 and AV1, developed by Google and the Alliance for Open Media (AOM). Newer-generation commonplace usually can cut back the bandwidth by about 30 p.c to 50 p.c in contrast with its predecessor whereas sustaining comparable visible high quality. On the identical time, nonetheless, every new commonplace has consumed considerably extra power and compute than the final, whereas necessitating encoders that have been many instances extra complicated.

We imagine AV1 would be the most viable codec for Meta over the subsequent a number of years. AV1 is the first-generation royalty-free video coding commonplace developed by AOM, of which Meta is a founding member. It delivers about 30 p.c higher coding effectivity than VP9 and HEVC — permitting individuals who use our apps to take pleasure in high-quality video at a lot decrease bandwidth, and enabling us to maximise storage effectivity and cut back egress site visitors, CDN prefetching/caching, and community congestion. AV1 additionally has a a lot richer characteristic set than different video coding requirements and might help most of Meta’s typical manufacturing usages. AV1 is royalty-free, and each the encoder and decoder implementations are open sourced, with very lively improvement and good help.

Over the previous few years, Meta has labored intently with the open supply neighborhood to optimize AV1 software program encoder and decoder implementations for real-world, global-scale deployment. Our purpose is to enhance playback from what we at the moment provide with AVC and VP9. We wish to make sure that as we roll out AV1, it delivers actual worth to the individuals who use our apps.

Discovering the suitable AV1 encoders and decoders

A number of open supply and closed-source encoder implementations are prepared for manufacturing, all virtually as environment friendly because the AV1 reference encoder. In a paper, “Towards much better SVT-AV1 quality-cycles tradeoffs for VOD applications,” collectively printed with Intel eventually 12 months’s SPIE convention, we benchmarked a number of open supply encoders — together with x264, x265, libvp9, libaom, SVT AV1, and VVC reference encoder (vvenc) — for a video on demand (VOD) use case. The graph beneath illustrates the trade-off between encoder high quality (vertical axis) and complexity (horizontal axis). Each level on the graph corresponds to an encoder preset. The y-axis represents the common BD-rate relative to libaom cpu-used=0; decrease values point out higher coding effectivity. The x-axis represents the encoding time in seconds in logarithmic scale.

Just a few highlights from this graph:

• SVT-AV1, the productization encoder for the AV1 coding commonplace, maintains constant efficiency throughout a variety of complexity ranges. With a complete of 13 presets, SVT-AV1 can cowl a complexity vary that extends from the upper high quality AV1 to the upper speeds AVC presets comparable to greater than 1000x change in complexity. This complexity vary covers all open supply software program encoders utilized in manufacturing techniques.
• At any given level on the x-axis, SVT-AV1 can maximize coding effectivity in contrast with another manufacturing encoder. For instance, the M12 preset has comparable complexity efficiency to the x264 veryfast preset, however M12 is about 30 p.c extra environment friendly.
• At any given level on the y-axis, SVT-AV1 can maximize encoding pace in contrast with another manufacturing encoder. For instance, the M8 preset is about as environment friendly as libvp9 preset 0, however M8 is sort of 10 instances quicker.

SVT-AV1 affords 13 presets, permitting a fine-grained trade-off between high quality and pace. Extra importantly, SVT-AV1 now features a “-fast-decode” possibility, which accelerates software program decoding — with solely a slight drop in effectivity — by robotically limiting or disabling the usage of AV1 coding instruments that aren’t software-decoder pleasant. SVT-AV1 additionally offers thread administration parameters to steadiness density and pace — important for large-scale manufacturing — probably enabling a one- or two-second delay for dwell video streaming. Many parameters will be adjusted to enhance coding effectivity or to help sure manufacturing eventualities. Some AV1 coding instruments that have been proposed to be used circumstances in deployment, comparable to reference body scaling, tremendous decision, movie grain synthesis, and swap frames, are additionally supported in SVT-AV1.

Our greatest problem will likely be client-side decoding of AV1. Many {hardware} distributors, together with Intel and NVIDIA, have begun to help AV1 {hardware} decoding on PC. Nevertheless, we’re serving video primarily to cell phones, most of which don’t embody AV1 {hardware} decoders. For now, we should rely totally on software program decoders. Two main open supply software program decoders are suitable with a number of platforms: dav1d was developed by VideoLAN and the open supply neighborhood and might function an app-level decoder, whereas Google’s libgav1 is built-in into the Android SDK.

After extensively benchmarking the decoders’ efficiency, specializing in sides comparable to useful resource necessities, crashes and responsiveness, and body drops, we determined to combine dav1d into the participant for each iOS and Android platforms. We’ve been working intently with the open supply neighborhood to optimize dav1d’s efficiency. Within the final 12 months, we additionally labored with Ittiam to conduct a benchmark test on Android telephones. dav1d can help 720p30 real-time playback on a lot of the units in our pattern, reaching 1080p30 on sure mid-range and high-end fashions.

Some Android telephones, such because the Google Pixel 6 Professional and Samsung Galaxy S21, already help {hardware} AV1 decoding. Within the close to future, we count on {that a} rising variety of high-end Android fashions will help AV1 {hardware} decoding, with mid-tier units following ultimately.

Deploying AV1 encoding on Fb Reels and Instagram Reels

Early in 2022, we deployed AV1 encoding for Fb and Instagram Reels. When somebody uploads a video, the platform generates a number of bit-rate encodings tailor-made to the video’s projected watch time. To stop stalling attributable to modifications in bandwidth, purchasers can choose the model that most closely fits their connection pace — a method known as adaptive bit rate (ABR) streaming. For movies with excessive projected watch time, we use superior ABR encoding primarily based on the convex hull dynamic optimizer algorithm. For every uploaded video, we produce a number of down-scaled variations and encode every with a number of quantization parameters (QPs) and Fixed Fee Elements (CRFs). For instance, for a 1080p video, we would create seven resolutions and 5 CRFs, for a complete of 35 encodings. After encoding, the system upscales decoded movies to the unique decision and calculates the standard rating.

Within the graph of charge distortion (RD) curves beneath, the x-axis represents the encoding bit charge and the y-axis the standard rating, expressed in FB-MOS items on a scale of 0 to 100.

From these 35 RD factors, we calculate the convex hull, a curve that connects the RD factors on the higher left boundary. (Theoretically, if we may use all attainable encoding resolutions and CRFs to provide a a lot denser plot, any level on the convex hull would be the most optimum encoding possibility for this video by way of decision and CRF worth.) As illustrated above, we will then choose one of the best encoding for supply primarily based on the goal high quality or bit charge.

We’ve simplified this difficult course of. In previous studies, we discovered that we may use the high-speed preset for first-pass encoding and to provide the convex hull, after which take a second cross to encode the chosen (decision, CRF) factors with the high-quality preset. Despite the fact that this method requires further encoding, it’s quicker as a result of the primary cross will be finished far more rapidly. (Coding effectivity drops solely barely.) This method works even when the primary and second passes use totally different encoders. For instance, we will use AVC or VP9 within the first cross and AV1 within the second. We will additionally leverage the {hardware} encoder in our internally designed ASICs to speed up this course of.

In the long run, we selected a two-stage hybrid {hardware}/software program ABR encoding method. {Hardware} AVC encoding is triggered at video add time; for this stage, we retailer solely the standard and bit charge data however not encoded bitstreams. When projected watch time of the video exceeds the brink, second stage encoding is triggered with software program AVC, VP9 or AV1 encoder primarily based on the chosen (decision, CRF) on the convex hull.

We will simply add AV1 as one of many second-stage encoders; it’s already deployed for Fb Reels. We’ve carried out an identical heuristic-based method for Instagram Reels. For one instance video proven within the graph above, three encoding households with AVC, VP9, and AV1 have been produced. Their RD curves intently comply with the convex hull from the first-stage encoding. For this explicit video instance,  the best-quality AV1 encoding rivals these of the opposite two requirements, however with a bit charge 65 p.c decrease than AVC’s and 48 p.c decrease than VP9’s. As well as, AV1 achieves the specified high quality inside a really slim bit charge vary, so we will additional cut back compute and storage prices by producing fewer encodings in the course of the second stage. Because of this, individuals who use our merchandise can take pleasure in high-quality video at a lot decrease bandwidth.

AV1 decoder integration and testing

It was comparatively simple to allow AV1 decoding and playback on the iOS units. After only a few rounds of exams, we began supply. To combine the dav1d decoder on iOS, we discovered that two to 4 threads would meet most of our manufacturing wants; any further threads would waste reminiscence and energy with out boosting efficiency.

dav1d has two modes: synchronous and asynchronous. In synchronous mode, dav1d decodes one body at a time however permits low-latency decoding for every body. In asynchronous mode, dav1d decodes a number of compressed frames in parallel, suspending rendering till all frames are decoded. In idea, asynchronous mode offers larger throughput and quicker decoding. For now, we undertake synchronous mode on iOS because it suits the prevailing participant stack, however we’re wanting into migrating to asynchronous mode sooner or later.

To help the decoding of 10-bit AV1-encoded HDR video, we constructed a single dav1d binary that helps each 8- and 10-bit decoding and ensures that coloration data is preserved within the transcoding course of.

The Android platform offered greater challenges. First, as a result of folks have interaction with our apps on an enormous variety of Android fashions, we needed to run native and large-scale A/B exams on numerous units to seek out the optimum decoder configurations. To assist debug and triage issues from the AV1 decoder library, we added in depth logging that propagated again error messages from all through the participant stack. This important step helped us rapidly determine and resolve points within the integration course of.

Second, as a result of we’re utilizing app degree software program decoders, we used the {hardware} VP9 decoder and software program AV1 decoder collectively when taking part in the identical video stream, to accurately help combined codec manifest and in-stream ABR lane swap. We wanted to verify they interacted with the render engine accurately.

We additionally wanted to help units with low efficiency and show decision. (This was not an issue with iPhones.) Though AV1 can encode high-resolution movies at a a lot decrease bit charge than VP9, bit charge discount is smaller for low-resolution movies. That makes it troublesome to indicate enchancment in top-line supply metrics for low-performance Android telephones. We responded by utilizing higher-quality encoding presets to spice up coding effectivity in low-resolution ABR lanes.

One other problem was that reminiscence allocation and thread creation elevated the decoding latency of the primary few video frames, prolonging the software program decoder begin time, delaying participant startup, and inflicting in-play stalls. This was most difficult with Reels, as a result of folks usually scroll throughout a number of Reels movies in fast succession. To enhance scrolling efficiency, we prefetched a number of Reels movies earlier, earlier than they have been performed.

Earlier than we conduct a large-scale A/B supply check, now we have to examine whether or not the top machine is highly effective sufficient for real-time decoding and playback of AV1 bitstreams. Nevertheless, there isn’t any simple solution to classify Android telephone efficiency. We can not check each mannequin that exists, as there are millions of them. And traits comparable to core counts, chipset distributors, RAM dimension, and 12 months and mannequin are usually not ample indicators of functionality. We ultimately determined to run a small benchmarking check to measure efficiency and provides every telephone a efficiency rating. This benchmarking check consisted of primary compute operations, together with Gaussian blur, reminiscence allocation, reminiscence copy, and 3D rendering. With this method, we may assign scores to any present or upcoming cell phones and group them primarily based on these numbers. Our A/B exams then recognized the fashions that might help 720p, 1080p, and 10-bit HDR playback.

After the preliminary Android rollout, we began to allow AV1 {hardware} decoding for the few Android telephones that help it. We count on {hardware} decoding to enhance AV1 efficiency, and we plan to carry out large-scale exams when a bigger variety of succesful telephones grow to be obtainable.

Newest supply standing

We began the AV1 supply for Fb Reels on iPhone in early 2022 and noticed the advantages throughout the first week of the rollout. 

The next graph reveals the week-over-week common playback FB-MOS for all Fb Reels movies performed on iPhones. Playback FB-MOS improved by about 0.6 factors after we deployed AV1. 

This second graph reveals the common bit charge for all Fb Reels movies performed on iPhones. AV1 diminished the common bit charge by 12 p.c. 

This final graph reveals the watch time of various codecs for Fb Reels on iPhone. AV1 watch time rose to about 70 p.c in the course of the first week of rollout.

We’ve continued to allow new options for iPhone, together with 1080p30 8-bit AV1 supply for iPhone 8 and past, 10-bit HDR supply as much as 1080p30 for fashions of iPhone X and past that help HDR show, and 1080p60 8-bit AV1 supply for iPhone 11 and past. AV1 encodes a excessive share of the Fb Reels and Instagram Reels movies watched on iPhones. We’ve additionally enabled 8-bit AV1 supply to pick out midrange to high-end Android telephones. The watch time share on Android for AV1 is comparatively small however rising.

What’s subsequent for AV1 at Meta?

AV1 delivers actual worth to the individuals who use our merchandise. It affords larger high quality at a a lot decrease bit charge in contrast with earlier generations of video codecs. For instance, within the video beneath, there’s an apparent distinction in high quality between AVC, VP9, and AV1 at roughly the identical bit charge.

Going ahead, we are going to proceed to broaden AV1 supply for Android telephones and allow {hardware} decoding in new units that help it.

For low-end Android telephones, it stays difficult to play again high-resolution AV1 bitstreams. To handle this, we’re at the moment experimenting with combined codec manifest help. On the server facet, the ABR supply algorithm generates a combined codec manifest that accommodates a number of video adaptation units with bitstreams encoded utilizing totally different codecs, comparable to VP9 and AV1. It additionally specifies which AV1 and VP9 lanes the machine ought to select from primarily based on its efficiency rating. For instance, a low-end telephone can play AV1 as much as 540p and swap to VP9 for larger decision lanes. 

With increasingly more {hardware} distributors implementing AV1 decoders in cell SOCs, we count on the variety of AV1 succesful units to proceed to develop within the subsequent few years, permitting extra finish customers to take pleasure in the advantages of AV1.

Acknowledgements

This work is a collective effort by the Video Infra staff and Instagram staff at Meta, together with exterior companions, together with the Intel SVT staff, VideoLAN, Ittiam, Two Orioles, and the open supply neighborhood. The authors wish to thank Jamie Chen, Syed Emran, Xinyu Jin, Ioannis Katsavounidis, Denise Noyes, Mohanish Penta, Nam Pham, Srinath Reddy, Shankar Regunathan, David Ronca, Zafar Shahid, Nidhi Singh, Yassir Solomah, Cosmin Stejerean, Wai Lun Tam, Hassene Tmar, and Haixiong Wang for his or her contributions and help.