AVIF Converter

The performance budget meeting goes the same way everywhere. Someone in a design lead seat declares that the new hero image must weigh less than 150 kilobytes, because mobile. You have a 1,600×900 photograph of an actual sunrise, and the math is not on your side. JPEG at quality 70 squeezes it to 196 KB and the sky starts to look like a paint-by-numbers. WebP at quality 75 lands at 138 KB and holds together. You're in the meeting, you're nearly home, and then the resident performance engineer asks the question that changes the project: what about AVIF?

AVIF at quality 55 encodes that same sunrise at 89 KB. Not only smaller than the JPEG target — smaller than half of it — with a sky that gradients smoothly from horizon to zenith because the format stores ten bits per channel instead of eight. The 150 KB budget stops being a constraint and becomes headroom. That's not a marginal improvement over WebP; it's another category of outcome, and it's why the format's adoption curve is the steepest in imaging right now.

This page converts in both directions — your JPGs, PNGs, and WebPs into AVIF for shipping, and AVIF files back into conventional formats when something on the wrong side of the support matrix needs to open one. Native browser encoders do the work where they exist and a WebAssembly build of the codec covers everything else, which means one tool, any operating system, zero uploads. Below: what AVIF is, what the quality and effort sliders really buy, the encode-time bill nobody warns you about, and the browser matrix that decides whether you can ship it yet.

AVIF's origin story starts with a fight. By 2015, streaming video needed a next-generation codec, and the patent-licensing maze around HEVC had the industry looking for exits. The Alliance for Open Media — Google, Amazon, Apple, Microsoft, Mozilla, Netflix, and the rest of the industry's who's-who — formed to build AV1: a video codec deliberately engineered to be royalty-free, published in 2018. AVIF, standardized the following year, is that codec's still-image face: AV1 keyframe coding, dropped into the same HEIF-derived container structure that HEIC uses, minus the licensing anxiety.

The family resemblance to WebP is structural. WebP is VP8's keyframe path; AVIF is AV1's. Each format is a video codec's intra-frame machinery repurposed for stills, and AV1 is roughly two generations of codec research beyond VP8 — a decade of prediction, transform, and filtering improvements stacked between them. The visible consequence at the file level: AVIF routinely halves what JPEG produces at equal perceptual quality, and beats WebP by a further fifth.

• 8, 10, or 12 bits per channel — a billion-plus colors, and smooth gradients that 8-bit formats band
• Full alpha — transparency alongside deep color, in lossy or lossless coding
• Animation in the container — sequences ride the same compression as stills
• Film grain synthesis — the codec can strip real grain, compress the clean image, and re-synthesize convincing grain at decode time
• Royalty-free end to end — the entire point of the alliance that built it

Most format tools point one way. This one points both, because AVIF's situation today genuinely requires both directions — you compress into it to ship modern pages, and you convert out of it whenever a file lands somewhere the format hasn't reached yet. The direction switch at the top of the tool decides which encoder spins up.

Going in — JPG, PNG, or WebP to AVIF — you're buying size and quality. The quality slider behaves like every modern codec's (75–85 is the photographic sweet spot, and AVIF's scale runs more generous than JPEG's), while the effort control is unique to this format: how hard the encoder searches for optimal coding decisions. More effort, smaller file, longer wait. We'll put exact numbers on that trade further down.

Coming out — AVIF to JPG or PNG — you're buying compatibility. An AVIF that some portal, kiosk, or older Safari install refuses to open becomes an ordinary JPG in seconds, decoded by the WebAssembly build so it works even in a browser that can't display the format natively. PNG output preserves everything the decode recovers; JPG output is the smaller, sharing-friendly option, exactly as on our other converters. If you arrived here with a file you can't open rather than a page to speed up, skip straight to the steps, pick the reverse direction, and you're done.

1. 1

   Pick your direction

   To AVIF for shipping smaller files; from AVIF for compatibility rescue. The switch changes which encoder and controls appear.

2. 2

   Drop in the files

   Batches are welcome — the queue tracks each file's size, dimensions, and progress. Photographic content is AVIF's home turf, but graphics and screenshots convert happily too.

3. 3

   Set quality and effort (to AVIF)

   Start at quality 75 with medium effort. That combination hits the sweet spot most sites ship. For a single hero image you'll serve millions of times, push effort to high and let it grind.

4. 4

   Set output format (from AVIF)

   JPG for photographs headed to people and portals; PNG for anything with transparency or an editing future.

5. 5

   Read the report

   Each file shows input size, output size, savings, and encode time. The encode-time column matters here in a way it doesn't on JPEG tools — check it before you commit to a five-hundred-file queue.

6. 6

   Inspect and download

   Zoom the preview into gradients and fine texture — the places AVIF visibly outperforms its elders — then take the files individually or as a batch.

Every block-based codec does the same three things: predict, transform, entropy-code. JPEG skips the first step entirely — each 8×8 block is coded blind, from nothing. VP8 predicts each sub-block from its neighbors above and to the left. AV1 predicts from a far larger neighborhood — above, left, above-left, above-right — across dozens of directional modes, and it can choose between 64×64 and 128×128 superblocks, letting flat regions of sky code as enormous single units while detail stays at fine granularity. The encoder spends its decisions where the picture actually has structure.

After prediction and transform, AV1 adds the refinements that a decade of streaming-video research bought. A constrained directional enhancement filter sharpens edges along their own angle without amplifying noise. A loop restoration filter can apply a separable Wiener filter — effectively a learned, signal-adaptive blur that removes compression artifacts — across whole regions. And on top of all of it sits the film grain machinery: the encoder estimates the image's natural grain, removes it so the clean structure underneath compresses dramatically better, transmits the grain's statistical description, and the decoder synthesthesizes fresh grain over the result. Old film scans and high-ISO photographs sometimes compress at sizes that look like misprints.

The last mile is precision. AVIF codes and decodes at 10 or 12 bits per channel as a first-class citizen, and the whole pipeline maintains that precision end to end. That's the technical basis for the format's famous resistance to banding: a sunset sky that stair-steps visibly in 8-bit JPEG holds a continuous ramp in 10-bit AVIF, because there are 1,024 steps available where the older format had 256. Sources shot in 8-bit still benefit, since the extra precision protects the encode-decode round trip even when the input never had it.

Here's the same 2,400×1,600 photograph through every codec this site discusses, encoded on the same machine back to back:

Two patterns are worth pulling out of that table. First, quality is the big lever and effort is the small one — dropping from q75 to q60 saved 53 KB while upgrading effort saved 23 KB, and only the quality change is visible if you squint at fine texture. Second, time scales with effort far more than output size does. High effort spent six seconds to remove thirteen percent — a trade that pays for a hero image served a million times and nowhere else.

That's the meeting-ending number from the opening story, now with the arithmetic attached. And the batch version: a 40-image gallery page at JPEG q80 weighed 5.9 MB and served in 5.9 MB; the AVIF version weighs 2.4 MB and took 74 seconds to encode once, on a developer's laptop, in a browser tab. The seconds are paid at build time. The megabytes are saved on every pageview, forever.

AVIF's remaining rough edges are known and enumerable, which is the best kind of problem:

• Browser support is solved at the top and ragged at the bottom. Chrome 85 (August 2020), Firefox 93 (October 2021), and Safari 16 (September 2022) all decode AVIF. Everything before those versions doesn't, and older Safari in the wild is the gap most sites still measure.
• Encode cost is real. Seconds per image at reasonable effort, minutes at high. Batch conversions need planning that JPEG workflows never did, and the effort slider exists precisely so you can opt out of the grind.
• Animated AVIF exists but is rare tooling-wise. The container supports sequences; most converters — this one included — handle stills only.
• Extreme dimensions hit codec levels. AV1's level system caps pixel dimensions and total pixel counts; very large panoramas may exceed what encoders accept without tiling. 4K and 8K photographic sizes work fine.
• Color profiles are simplified. Wide-gamut and HDR AVIFs exist in the wild; this converter's output path works in sRGB, so exotic color spaces are mapped to the standard profile on the way through.
• Some very old hardware decodes slowly. First-generation phones without AV1 hardware blocks decode AVIF in software — fine for one hero image, noticeable in a fifty-image gallery. The picture-element fallback pattern covers this audience too.

How do you know whether your own audience crossed over? Two minutes of checking beats a year of guessing. Your analytics packages tell you the browser versions actually visiting; cross-reference the oldest Safari and Chrome versions with meaningful share against the support dates above. And your server logs can answer the question empirically — serve AVIF with a fallback for a month, then count how many requests fell through to the JPEG. When that number approaches zero, the training wheels come off.

The format crossed from experiment to infrastructure around 2023, and the names doing it at scale tell the story. Netflix, one of the alliance's founding members, ships AVIF where its own measurements say it wins. News organizations with image-heavy front pages adopted it for heroes and thumbnails, where the size-to-attention ratio is highest. WordPress added native AVIF support to core in version 6.5, which quietly made it a checkbox rather than a project for a third of the web.

• Performance teams, for whom AVIF is the current endgame of image weight on hero and above-the-fold content
• News and media sites, converting once at publish time and saving the bytes on every article view
• E-commerce platforms, where product-image weight translates directly into conversion-rate math
• Portfolio and agency sites, where a fast first paint is both UX and a selling point in itself
• Streaming and social platforms, using AVIF for artwork and thumbnails at enormous scale

There's also a quieter adoption channel most sites never notice: CDNs now convert cached images to AVIF automatically at the edge, for visitors whose browsers accept it. If your traffic runs through one of those, part of your audience is already receiving AVIF without anyone on your team deciding anything. Converting deliberately, with a tool like this one, matters for everything the CDN doesn't touch — self-hosted assets, upload pipelines, email-adjacent systems, and the places where you want to control the quality setting rather than inherit a default.

Notice who's absent: people emailing images to each other, or printing them, or feeding them to decade-old software. AVIF is a delivery format for the web, full stop. The moment a file's destination is anything other than a modern browser, convert it out — that's why the reverse direction on this tool isn't an afterthought but half the product.

You don't choose a format in a vacuum; you choose it against an audience and a workflow. The honest decision matrix, accumulated from everything above:

<picture>
  <source srcset="hero.avif" type="image/avif">
  <source srcset="hero.webp" type="image/webp">
  <img src="hero.jpg" alt="Sunrise over the coastal trail" width="1600" height="900">
</picture>

That markup is the whole strategy in five lines: every browser takes the best format it understands, the JPEG at the bottom guarantees nobody gets a broken image, and you never have to choose between progress and safety. Encode the AVIF with this tool, the WebP with the JPG to WebP Converter or PNG to WebP Converter, and the JPEG is the original you already had. When a hard size cap matters more than the format itself, the Image Compressor targets kilobytes directly.

AVIF is the newest format this site's tools handle, and it's the one where a browser can least be trusted to have the pieces built in. So the page brings its own: the encoder and decoder are compiled to WebAssembly and arrive with the page itself, running in the same sandboxed tab you're reading. Where the browser exposes a native AVIF encoder, the tool hands off to it for speed; everywhere else, the WASM build does identical work. Your images enter through the file picker and leave through the download tray, with no network stage anywhere in between.

That architecture also explains the pricing, such as it is: there isn't any. Server-based AVIF converters charge because AVIF encoding is expensive CPU time on rented machines. An encode that runs on your processor costs nobody anything, so the tool can be free, unmetered, and watermark-free — and your unreleased product photos, your client's campaign imagery, and your family archive all stay exactly where they started, on your machine.

AVIF work sits at the end of a pipeline that usually starts with older formats and sometimes has to return to them. The neighboring pages:

• JPG to WebP Converter — the workhorse step for photographic libraries at batch scale
• PNG to WebP Converter — the lossless and transparency path for graphics and screenshots
• Image Compressor — the format-agnostic option when a hard kilobyte cap rules the task