4K Video Upscaler
한국어

Guide

How to Upscale Video to 4K Free in Your Browser (No Upload)

최근 업데이트:

Drop a video on the home page — a clip of any length — pick a size (1080p, 2K or 4K) and a quality (Fast, Pro or Ultra), press Start, compare before and after on the same frame, download. On a computer, several videos or a whole folder go in one batch, upscaled one after another. Everything runs on your own computer's or phone's graphics chip through WebGPU, and the page measures your device before each run, so the time estimate is yours, not an average.

1. How to use it

  1. Choose a clip.

    MP4, MOV, M4V, MKV, WebM or 3GP of any length — any size on a computer, up to 2 GB on a phone or tablet; AVI, MPG, VOB, WMV, FLV, DV, a camcorder's MTS and the other old formats open too, as do ProRes and 10-bit H.264 — the page turns them into an MP4 first. The browser reads the clip's resolution, frame rate and length on your device and opens the workbench in place of the drop box. Once it is open, the page takes 16 moments spread across the clip and recognises what it holds — whether the picture is 2D animation, live action, 3D animation or a screen recording, whether it shows faces, and whether the words on screen are subtitles, a watermark or danmaku — measures how sharp, noisy and blocky it is, and reads whether the file carries an AI-generated label: the basis for choosing the upscale's experts and tuning it to the content. A phone or tablet whose graphics chip supports half-precision shaders (shader-f16) recognises the kind of picture and the faces, without telling subtitles, a watermark or danmaku apart; any other phone or tablet measures the picture by its rules only. The recognition runs on your device in the background and stops the moment you press Start. A file this browser cannot open or decode is named the moment you choose it, with what to do next: the browsers that open it as it is, or a conversion to MP4 (H.264). If a run later finds the file can no longer be read, the panel asks for it again: with Replace, or on an Android phone — whose gallery can stop letting the browser read a video — from Files.

    The home page: a drop box with a Choose video button, and the four trust points above it
    Drop a clip on the box or press Choose video.
  2. Pick a size and a quality, preview, start.

    A platform button (YouTube, TikTok · Shorts · Reels, Instagram, X) fills in the shape, size, file type and frame rate that platform wants, and Strength, under Quality, starts on Auto: the page picks how hard Fast and Ultra work for this clip on this device. "Preview this frame" runs a short test on the frame you paused on and shows it before and after, with the time estimate; under Start the panel names the setting it recommends for this clip, and its time, one tap away. Then press Start; you can pause and resume, a long run , and a long job offers a short trial first — as much of the clip as your device upscales in about 1 min — whose finished view starts the whole clip in one tap.

    The workbench: a blocky 480p clip on the left with the preview window on the eyes; on the right the Upscale panel, with Pro's before and after in the preview at its top, then Preview this frame, Start with its estimate and the setting recommended for the clip, and the template, size and quality buttons below
    Pro on a blocky 480p download: the preview at the top of the panel shows the marked window before and after; Start, its estimate and the recommendation for this clip sit right under it.
  3. Compare and download.

    The original sits on the left and the result on the right — same frame, same moment — and a Detail window on the picture shows one spot of both at the result's own pixels, opening where the upscale sharpened most. Download as MP4, MOV, MKV or WebM with the sound, copied or re-encoded for the format — where this browser cannot carry it into the format you chose, the panel says so before you start. The file size (the video bitrate) is a setting: Smaller, Standard or Larger — about 17.1 / 34.2 / 54.7 Mbit/s for 4K H.264 — or a rate you type (1–100 Mbit/s), and the panel shows the megabytes before you start. Once a run is done, "Export again" rewrites the file from the frames you already have, with no second upscale: another file size (bitrate) — Smaller, Standard, Larger or a rate you type — or another format, container or frame rate; a change of container alone copies the picture over in seconds. The result stays in this browser for 24 hours: within that time, "View" under "Choose video" on the home page opens it here again, to play, compare and download. Open the tool when you are ready.

    The finished view: the original on the left, the Pro result on the right, a gold axis between them, and a Detail window in the corner showing the face in both at 1:1; the panel lists the output size, codec and file size, a Download button, how long the result stays in the browser, and a card asking how the result compares with the original, with four answers from Much better to Worse
    Done: original left, result right, the axis draggable; the Detail window shows both at 1:1.

Screenshots: Komekami! Girl's PV (2026), official channel — CC BY 4.0 via Wikimedia Commons, re-encoded at 220 kbit/s.

Any length — long videos welcome, saved as it goes

  • Any length

    no cap on the clip — a two-hour film runs the way a ten-second one does, and on a computer the file can be any size

  • 15 s

    of graphics work before the first finished piece is on disk, then one every 30 s

  • 24 h

    an unfinished run waits for you — the home page offers to continue it from the last saved piece, frame for frame

The result streams into your browser's private storage while the run goes on, in finished pieces — which is what lets a long video, a full movie or an hour-long recording, run through. A job the estimate puts past 90 seconds gets its first piece after 15 seconds of graphics work, and one more every 30 seconds after that; once that first piece is saved, the panel says the run can be continued after a reload. Close the tab — or come back to a browser the phone has closed — within 24 hours: the home page lists the job, picks it up from the last saved piece frame for frame, and joins the pieces into one file at the end. Shorter jobs run straight through in one piece. You can take the whole clip, a window of 5 seconds to 10 minutes dragged on the filmstrip, or one that ends where you type; the estimate next to Start gives the time on your own device, a long job's panel says how much room it needs and offers a short trial first, and where the browser measures its free space, the page checks the result fits before any graphics work starts; when it does not, the panel names a result kept from an earlier run and deletes it in one tap if that makes room. If the device runs out of space partway through a long job, the panel says how much the job had stored and how much free space finishing needs: free that much and press Start, and the run carries on from the save. With no saved progress to continue from, it offers a smaller Output that fits, with the room that size needs.

Several videos or a whole folder at once

On a computer, in any browser that runs the upscaler, you can pick several videos in the window "Choose video" opens, drag several videos or a whole folder onto the page, or press "choose a folder" under "Choose video" — videos in its subfolders come too. One batch takes up to 200 videos. You set Output and Quality once, the list shows what each video becomes and about how long it takes, and the videos are upscaled one after another on your graphics chip. A video too small for the Output you chose goes to the largest size it can reach, and a Quality this computer can't run for a video steps down to one it can, as its row says. AVI, MPG, VOB, a camcorder's MTS and the other files the page converts first are upscaled one at a time: choose each of those on its own. A phone or a tablet upscales one video at a time.

Under "Where the results go", "A folder (recommended)" — in Chrome and Edge — saves each video into a folder you pick as soon as it is done, beside the files already there and never over one. "This browser", in every browser that runs the upscaler, keeps the results here for 24 hours, and "Download all" saves them one after another. Close the page halfway, and "Your recent videos" on the home page keeps the batch: come back in the same browser within 24 hours and choose the same videos again to run the rest. A batch that was saving into a folder asks for that folder again.

What the download is called, and what it keeps

A download is named after the original, its size and its quality: holiday.mp4 upscaled to 4K on Pro is holiday_4K_Pro.mp4, and a 30-second trial of it holiday_4K_Pro_trial-30s.mp4, so a Fast and a Pro version of the same clip sit side by side in your Downloads folder. A video an Android phone's gallery hands to the page under a number is named by the time it finished. The file keeps the original's title, description, comments, author, copyright and keywords — and, from an MP4 or MKV, its sound's language and name. It also keeps when and where the original was shot and the camera or phone it was shot on, written where photo libraries look for them, so your gallery files the upscaled video next to the original; MKV and WebM keep the date and time, as they have no standard place for GPS coordinates or a camera model. To leave all of that out, turn off "Keep date, location and camera info" under More options — your browser remembers the choice. The encoding tool names 4K Video Upscaler and its version, and one line at the end of the comments says how the file was made: the original and new frame size, the quality, the codec, the date and the site's address. Whose account the original came from and its cover art are never carried over, and nothing about your device is written into it. An original's AI-generated label carries over whatever the settings: when the original says it was made with AI — Content Credentials (C2PA), an IPTC digital source type, or the generator's own label in its metadata — an MP4 or MOV result declares it too, in IPTC's standard form, and a label in the metadata goes over word for word in MP4, MOV, MKV and WebM alike. The C2PA manifest itself is signed over the original's bytes, so it stays with the original. A turned or mirrored video comes out the way your player shows the original.

Old formats: AVI, MPG, VOB, WMV, FLV and camcorder files

Camcorder files open as they are. The MTS / M2TS of an AVCHD camera and TS recordings are rewrapped as MP4 inside the page before the workbench shows them — the pictures copied, not re-encoded, and a Dolby Digital (AC-3) sound track turned into AAC (Opus in Firefox) so every browser plays it; a bar in the drop box shows how far it is, and nothing is uploaded. A recording joined from several parts into one file — MTS, TS, VOB or FLV — opens whole, every part in order. Most camcorder and TV footage is , and the panel says so: it is upscaled with its fields as they are, so moving edges can show fine comb lines. Try a short window first, or deinterlace it in HandBrake (step 3 below) for the cleanest result.

DVD rips and older downloads come in containers and codecs that browsers do not read — AVI (often DivX or Xvid), MPG and VOB (MPEG-2), WMV, FLV and DV, an MKV holding MPEG-2 or Xvid, and the Motion JPEG of old cameras' AVI and MOV files or the MPEG-4 and H.263 of old phones' 3GP. Two pictures from editing and filming take the same path: Apple ProRes, which no browser decodes, and the 10-bit or 4:2:2 H.264 of cameras' high-quality modes where the browser cannot decode it. The page converts such a file into an MP4 inside your browser with FFmpeg's own decoders, fetched the first time you choose one, deinterlacing interlaced pictures on the way, and nothing is uploaded: the beginning opens within seconds — enough to preview it and try a short run — while the rest is prepared behind the workbench, and a run of the whole video starts once it is ready; an AVI or FLV that holds H.264 has its video copied into the MP4 untouched. The converted video carries the file's first sound track, re-encoded in AAC (Opus in Firefox) — 5.1 stays 5.1 where the browser's encoder takes it, else it becomes stereo; its other sound tracks, subtitles and chapters stay in the original file. In a browser without these formats still need a conversion first, and the page says so the moment you choose one. Convert such a file once to MP4 (H.264) at its own size, then upscale the MP4:

  1. Open the file in HandBrake — free and open source, for Windows, macOS and Linux (for a DVD, open its VIDEO_TS folder).
  2. Keep the preset it opens with, Fast 1080p30: it writes an H.264 MP4 and keeps a video smaller than 1080p at its own size. For footage at 50 or 60 frames per second, set Video → Framerate to Same as source.
  3. For a DVD, a tape or a camcorder recording, check that Filters → Deinterlace is on: interlaced video shows combed stripes on anything that moves, and an upscale would sharpen them.
  4. Start the encode, then choose the MP4 on the home page.

A file that plays in one browser and not in another — an iPhone's HEVC in Firefox — needs no conversion: the page names the browsers on your device that run the site and open it as it is. An MKV opens in Safari too, rewrapped the same way. Every container and codec is listed on Supported formats.

When the page can't open a file, its message says why; why a video won't open in your browser explains each message and what to do.

For DVD and tape footage, how to upscale old videos says how big it can go, which tier to pick and how long an hour takes.

Which settings for which video

Starting points from our own test footage; the preview decides for your clip. Every row opens the tool with those settings.

Your videoTryWhy
Phone or camera footage, 720p–1080p, cleanFast → 2K or 4K open with these settingsFast crispens edges and leaves the look alone; the repair tiers spend minutes per minute of clip to change little.
Blurry, shaky or dim phone clip, any sizeFast → 1080p or 2K openBlur and shake are not fixed by any tier. Pro only if flat areas (sky, walls, skin) show small squares, like a bad livestream.
Anime, motion comics, line art, UI, screen recordingsFast → the largest size offered openLine content gains the most from Fast: crisper edges, no bright rims, nothing invented. A 640 × 480 episode reaches 2K, not 4K.
Downloaded anime with blocks, white rims or noisePro → 2K (on a phone too: 480p episodes) openPro gives the cleanest lines on animation. A DVD's VOB or MPG is deinterlaced as the page converts it; an interlaced MP4 or MKV rip is cleanest deinterlaced first (HandBrake's deinterlace filter, or ffmpeg's yadif).
AI-generated videoFast → 4K; Pro when it looks smeared and is ≤ 720p openGenerated clips are clean but soft. Never Ultra on them: it hardens the soft look into wax.
Downloaded or re-compressed 480p–720p live footage with visible blocksPro → 1080p or 2K openPro removes blocks and light noise most cleanly; on this footage Fast looks the same as a plain enlargement.
Old film, heavy grain, the worst 360p–480p sourcesUltra → 1080p or 2K, on a computer; preview first openUltra is the one tier that wins on heavy grain and on footage Pro leaves blocky. A clip that is smeared flat rather than blocky cannot be brought back by any tier — Ultra makes it worse.
1080p with compression blocksFast → 4K openOn a computer's graphics chip Pro and Ultra take a 1080p clip too, and the time next to Start says how long it will run.
Faces close upFast; for Pro or Ultra, check the skin in the preview — with Ultra, set Strength to Light open Ultra · LightFast does not restore skin and does not make it plastic; a 4× repair can over-smooth it, and Ultra at Light keeps more of the skin's own texture.
Night and dark scenesFast; with visible noise, Pro open ProFast leaves noise as it is; Pro cleans light noise best; Ultra hardens fine detail. The stripes in a dark sky (banding) stay at every tier.
Old tapes with stripes on moving things (interlaced)An AVI, MPG, VOB, WMV, FLV or DV: choose it as it is. An MP4, MKV or MTS: deinterlace first (HandBrake or ffmpeg yadif), then upscaleThe page deinterlaces the old formats as it converts them. The upscale keeps an MP4's or a camcorder file's fields as they are, and would sharpen the stripes.
Already 4KPro → 4K repair, on a computerThe panel opens on 4K repair: Pro cleans the picture at the clip's own size. It needs a graphics chip with half-precision shaders; anywhere else the panel says the video is already 4K.

How long it takes

Time is frames × time per frame: one minute of clip at 30 fps is 1,800 frames. The estimate next to Start is measured on your own device — the settings the clip opens with, every other setting converted from that measurement by the work it takes — and is the one to trust. Under it the panel recommends a setting for the clip: Pro at the largest size it runs, when that finishes within 8× the clip's length — or within 5 minutes for a short clip — and Fast otherwise; Ultra is never the recommendation, and a phone's is Fast. A setting past 5 minutes names a quicker one that saves at least 30 % of the time, and the recommended one, each one tap away. For scale, on a Mac with Apple silicon:

JobGraphics chip, per frameGraphics chip, per minute of clip
Fast, 1080p → 4K21.1 ms≈ 51 s
Fast, 720p → 4K (two 2× steps)50.7 ms≈ 2 min
Pro, 720p → 4K404 ms≈ 16 min
Ultra, 720p → 4K777 ms≈ 31 min

Fast's rows are at Strong, the largest of its three networks; from 1080p to 4K, Medium takes 10.6 ms a frame there and Light 3.1 ms. The table is the graphics chip's time alone; a whole run also decodes and encodes the video.

A phone runs Fast at Medium by itself (Strength on Auto), which looks like Strong on most footage: on a flagship Android phone with an Adreno 7-series graphics chip that is about 2.5 min per minute of 1080p clip to 4K, against 4.5 min at Strong, and a phone slows further as it warms up. A mid-range phone takes a few times more; where Medium would run past 8× the clip's length, Auto takes Light. A computer with an NVIDIA RTX 30–50-series graphics card is in the Mac's range or quicker; built-in graphics take a few times longer, and Intel built-in graphics (10th-gen Core or older) far longer. A browser with graphics acceleration switched off runs on the processor, far slower: the panel's device line recommends the quickest setting, and Auto takes Light.

Asking an AI about your settings

The head of the Upscale panel: the summary line, the device line, and the ChatGPT, Gemini, Claude and Copy icons after 'Can it run?'

The icons after "Can it run?" open ChatGPT, Gemini or Claude with your device, clip, settings and estimate already typed in (the Copy icon puts the same text on the clipboard for any other assistant); the text links this guide, so the answer comes from the long version, not a guess.

Links that carry settings

A link to this site can name the settings to start from — https://4kvideoupscaler.com/?quality=pro&size=2k opens the page with Pro and 2K ready (the rows above and an assistant's answer link this way); what the device cannot run falls back to what it can, and a link never starts a run.

ParameterValuesMeaning
qualityfast / pro / ultraThe quality tier.
size1080p / 2k / 4kThe output size (longer side).
templateyoutube / shorts / instagram / xA platform template: its shape, size and format, as the panel's own buttons set them.
strengthauto / light / medium / strongHow hard the network works: Fast's size, Ultra's denoise; auto lets the page choose.
steadyon / offThe Steady filter.
denoise1 / 0.5 / 0Ultra's strength by its older name, in that order Strong / Medium / Light.
fps60 / 30 / 24The output frame rate, only below the clip's own.
reasonblocky / noisy / soft / fast / platform / archiveThe main problem the settings answer — a label for our statistics, not a setting.
via—Where the link came from (an assistant, the guide, a share).
noteup to 60 charactersOne line shown with the settings.

2. Quality tiers

Three , two kinds of network. Fast is a small 2× network that crispens edges. Pro and Ultra are 4× repair networks that rebuild the picture and remove compression blocks and noise: Pro is the cleaner choice for almost everything, Ultra the heavier one for old film and heavy grain. Under Quality, sets how hard Fast and Ultra work — Fast's network size, Ultra's denoise — and Auto chooses for the clip and the device. Tier names stay; the network behind a tier may be updated.

Fast

Makes edges crisper, changes nothing else · the tier every device starts on

Doubles the size and firms up edges and lines on top of a plain enlargement. It adds crispness, not picture information: the look, colour and grain are left alone. Strength picks the network: Light the smallest and quickest, Medium about as sharp as Strong in about half the time or less, Strong the largest.

Comparable to
An Anime4K-class 2× network (a few thousand to about a hundred thousand parameters).
Per frame
≈ 21.1 ms per 1080p → 4K frame at Strong (Medium ≈ 10.6 ms, Light ≈ 3.1 ms); ≈ 50.7 ms per 720p → 4K frame (two steps)
Best for
Clean footage, anime, line art, screen recordings, AI-generated video — and everything on a phone.
Not for
Compression blocks, noise, old tapes: there it looks the same as a plain enlargement.
Phones
Yes — at Medium by itself, Light where Medium would take too long, Strong when you choose it.

Pro

Repairs blocky, noisy footage · the cleanest lines on anime · half of Ultra's time

Rebuilds the frame at four times its size while removing compression blocks and light noise — the cleanest de-blocking of the three, and the cleanest lines on downloaded anime.

Comparable to
A Real-ESRGAN-class compact 4× network, the lighter member of that family.
Per frame
≈ 0.4 s per 720p frame — about 8× the time of Fast on the same job
Best for
Downloaded or compressed anime; blocky or lightly noisy live footage up to 720p.
Not for
Clean footage, where it costs minutes for little change.
Phones
Yes, up to 854 px on the longer side (480p) — up to about two seconds a frame, so a short window. At 720p, a budget phone works about 29 minutes on a 30-second clip.

Ultra

The heaviest repair · for old film and heavy grain · can smooth skin

The deepest repair: removes heavy grain and noise and sharpens smeared texture. Its Strength, under Quality, is its denoise in three steps — Strong (its plain model, which Auto runs), Medium, Light — and Light keeps grain and skin texture. About twice the time of Pro.

Comparable to
A Real-ESRGAN-class compact 4× network, the deeper member of that family.
Per frame
≈ 0.8 s per 720p frame — about 15× the time of Fast on the same job
Best for
Old film, heavy grain, the worst 360p–480p footage when Pro leaves blocks behind. Preview faces first, and try Strength Light on them.
Not for
Clean or AI-generated clips (it hardens them), phones.
Phones
On computers. Even a flagship phone would work about 50 minutes on a 30-second 720p clip, and a run that heavy can make the browser switch the graphics chip off, so phones are not offered it.

The repair tiers on a phone. Pro and Ultra are 4× restoration networks, and every frame is a long stretch of work for the graphics chip. On a phone that adds up: a 30-second 720p clip keeps the graphics chip working for about 29 minutes with Pro on a budget phone and 50 with Ultra even on a flagship, and runs that heavy can make the browser switch the graphics chip off. That is why a phone gets Pro on clips up to 480p and Ultra runs on computers; Fast is unaffected.

No tier interpolates frames (frames are only dropped), deinterlaces (the page does that to the old formats as it converts them, before any tier runs), restores faces or changes colour and exposure; each frame is processed on its own, and the optional "Steady" filter under More options only settles grain that flickers in still areas.

3. Before and after

Three real runs of this page's own engine, one per tier, from low-resolution sources: the same window of the same seconds, the source on the left the way a player shows it, the result on the right. Drag the axis.

Before · 480pFast → 2K
Fast, a clean 480p clip → 2K: lines and lashes come out crisp, nothing is invented and the look stays the same.Komekami! Girl's PV (2026), official channel — CC BY 4.0 via Wikimedia Commons
Before · 480p, blockyPro → 2K
Pro, a blocky 480p download → 2K: the blocks go and the lines are drawn back.Komekami! Girl's PV (2026), official channel — CC BY 4.0 via Wikimedia Commons; re-encoded by us at 220 kbit/s to stand in for a poor download
Before · 360p, film grainUltra → 1080p
Ultra, a grainy 1933 film at 360p → 1080p: the grain goes, outlines and the window's lattice are drawn back — the footage Ultra is for.Ugokie Kori no Tatehiki (1933) — Public Domain Mark 1.0 via Internet Archive

4. The engine: SR Engine

Every tier of 4K Video Upscaler runs on SR Engine, our own WebGPU : 62 KiB of code (gzip), every compute kernel written for this family of networks, and model files of 4 KB to 2.4 MB, each fetched the first time it runs. Measured by us against ONNX Runtime Web, a general-purpose runtime for machine-learning models in the browser, with the same weights and frames: Fast is 4–14× faster in GPU time per frame on a Mac with Apple silicon and a budget Android phone with a Mali graphics chip, Pro and Ultra 1.6–2.4× faster on the Mac, and Pro 1.8× faster on that phone. On the same machine and network, SR Engine is 3.0× faster than the open-source engine we ran before.

  • 4–14×

    faster in GPU time per frame than ONNX Runtime Web with Fast — the same networks and frames, on a Mac with Apple silicon and

    a budget Android phone with a Mali graphics chip
  • 62 KiB

    of engine code (gzip) before the first frame, models aside; ONNX Runtime Web's WebGPU files are 3.84 MiB even with brotli, 6.39 MiB with gzip

  • 3.0×

    faster than the open-source engine we ran before — the large 2× network (Fast at Strong), the same machine: 21.1 ms against 63.3 ms per 1080p → 4K frame

Arena memory

112 Bper pixel for the large 2× network — 368 B before

One pool for every intermediate result

All the intermediate results of a frame live in one GPU buffer, sliced by how long each layer's output must survive; a slice is reused the moment its last reader is done. The heaviest step of 720p to 4K takes 546 MB for the whole frame, the input, the output and the weights included, and fits a 128 MiB budget in 4 .

720p → 4K, second step (2560 × 1440 in): 112 B × 3.686 MP is the 413 MB of feature maps, and the 546 MB whole-frame figure adds the weights, the input texture, the output pixels and one readback. Megabytes here are 10⁶ bytes; the 128 MiB budget is binary, because the limit it comes from — the storage binding every WebGPU device must allow — is defined that way.

Row bands

4.6%extra rows on the heaviest repair network — 20% before the planner derived each layer's rows from its readers

Big frames in bands, identical output

When a frame will not fit the memory budget, the engine cuts it into horizontal bands, each carrying the rows of context its own layers need, and the output is the same as a whole-frame run down to the last bit. Carrying that context is the only cost, and the planner works every layer's rows backwards from what reads them instead of assuming the worst: 4.6% extra rows on the heaviest repair network, 1.1% on the Fast second pass. Rows the planner stopped recomputing are wall clock on a phone: Fast runs 3.23× faster there with the medium network and 2.02× with the large one, and the heaviest tier fits in 3 bands instead of 4. A device with a 128 MiB binding limit still delivers 4K, and 96 MiB is all a phone is given.

The same second step in 4 bands under the 128 MiB budget: 42.2 ms per frame, output equal to the whole-frame run. The phone figures are one session on a flagship Android phone with an Adreno 7-series graphics chip.

Half precision

3.5×faster with shader-f16 than the 32-bit path

16-bit floats where the chip supports them

Where the graphics chip offers 16-bit shader arithmetic (), the whole network runs in 16-bit floats, otherwise in 32-bit: the heaviest repair pass takes 0.78 s per 720p frame, against about 2.7 s on the 32-bit path. Against the PyTorch reference, 16-bit output measures 54–70 dB and 32-bit output 93–100 dB.

The tests fail below 50 dB for 16-bit output and 60 dB for 32-bit.

Frame-sized submits

88 msto answer a tap on a phone while a repair job runs — 1024 ms when each frame went to the graphics chip in one go

The page answers while the chip works

A phone's browser paints the page on the same graphics chip the networks run on, and only once the work already handed to it is done: a frame given in one go holds the page for as long as the frame takes. So the engine hands a frame over in pieces of about 12 ms of GPU time — one display frame, measured on the device as it runs — cutting a longer layer into blocks of rows that compute the same pixels, and the page paints between two pieces. On a computer the pieces are 50 ms and the next one already waits in the queue, so the chip never idles: the large 2× network still runs a 1080p → 4K frame in 21.1 ms, against 63.3 ms on the open-source engine we ran before, which submitted once per layer. Between bursts of frames the page also pauses, so a whole run takes about 1.4× the network time and the device stays usable.

On a flagship Android phone with an Adreno 7-series graphics chip, running Pro on a 480p clip, the longest the page went without painting fell from 750 ms to 33 ms, for 4% more time per frame — the page painting again takes its turns on the chip.

Pixel-exact

100%of the pixels computed, and the result's mean colour within 1.5 of 255 of the source's

Every pixel computed, the colour unchanged

Every output pixel is the network's own, out to the frame's last column and row: the work covers the whole frame and its edges are padded with zeros, where the open-source engine we ran before worked in blocks of eight, left the last few columns and rows unprocessed and wrapped a quarter of the opposite edge's colour into the outermost ring. Frames reach WebGPU through the browser's 2D path, because importing a video frame directly in Chrome lifts the mid-tones by 4–8 out of 255, and an automated export test compares the result's mean colour with the source's. The exported file is written in BT.709, the colour standard players assume for HD video, and tagged as such: the export tests decode colour bars from it to check both.

Measured in Chrome with no network, each frame imported and written straight back: the mid-tones lifted in 5 of 7 test clips.

On the device itself

102checks on a flagship Android phone with an Adreno 7-series graphics chip, re-run on every real-device pass

Verified on real phones

A desktop is not evidence about a phone, so the networks run on one: every stored layer is read back on the device and compared with the CPU reference, layer by layer, alongside the operator probes and the 8-bit path. That is how the 3 compiler defects only phone graphics chips showed were caught — 2 on this phone, 1 on a budget Android phone with an Adreno 6-series graphics chip — and each is registered as a workaround that can be switched off, with a probe that re-asks on every run whether it is still needed, instead of the defect being rediscovered the day it stops holding.

The same graphs passed on all 3 desktop engines — Metal, the software renderer and WebKit — and only phones showed these 3.

Eight bits, on phones

3.22 sper 720p frame for the heaviest network on a flagship Android phone with an Adreno 7-series graphics chip — 6.24 s on the same phone in 16-bit float

The repair networks, in 8 bits

Where a phone's graphics chip can multiply four at a time, the engine quantises a repair network's weights to 8 bits and keeps the sums in 32-bit integers, with a calibration file beside the model deciding each layer's scale. Phones run Pro this way; Ultra stays on computers: even in 8 bits, a 30-second 720p clip would keep a flagship phone's graphics chip working for about 50 minutes. Desktops stay on 16-bit floats, where the integer path is emulated and slower.

Measured on a flagship Android phone with an Adreno 7-series graphics chip in three steps: 7.05 s in 16-bit float on the build this engine replaced, 6.24 s in the same precision here (the planner, 4 bands down to 3), then 3.22 s in 8 bits — 1.94× for the 8-bit step, 2.19× end to end. Correctness is checked on that class of device: 49.9–50.1 dB against the integer reference, where the test fails below 44, and a banded run matches a whole-frame one.

Separable resize

1.8–2.9×faster resize than the pass it replaced, same frames

Two passes beat one

The step that fits a network's output onto the size you asked for used to sample a square of pixels for every output pixel. It now runs as two one-dimensional passes — across, then down — with the horizontal result held in workgroup memory, so the work grows with the width of the filter instead of its square. The filter weights are computed once into a table the shader reads.

Against the CPU reference the resize now lands at 98–102 dB, where the previous one sat at 76–78.

Tuned on your GPU

18%faster on a Mac with Apple silicon once the engine picks the block shape

It measures the chip instead of guessing

The fastest block shape depends on the network and the chip together, so the first time a model, a precision and a device meet, the engine runs each candidate on a reduced frame and keeps the fastest; every later run takes the cached answer. A candidate has to win by more than 8% to displace the default. Where the wider block is known to lose — a flagship Android phone with an Adreno 7-series graphics chip runs the repair network 2.5× slower with it, on register spills — the wider block is never tried.

Not measuring the wider block on those chips takes first-run setup on the large 2× network from 1.7 s to 1.1 s, and saves 4.7 s on the repair network.

Low-spec baseline

16 KBof workgroup memory is all the baseline kernel asks for — the minimum WebGPU guarantees on every device

Built to fit old phones' limits

Every kernel is written within a low baseline — 4 storage buffers, 128 threads per workgroup, 16 KB of workgroup memory, no 16-bit shaders required — which the graphics chips of old and budget phones meet: a budget Android phone with a Mali graphics chip runs Fast at all three strengths, and Pro at 480p, in the comparison below, and a budget Android phone with an Adreno 6-series graphics chip, whose graphics chip has no half precision, runs Fast in 32-bit floats and matches the CPU reference. Where a device offers more, the engine detects it and steps up.

The engine's tests plan every shipped network at the sizes the product runs it, in 32-bit and 16-bit and at every block shape, and fail if a kernel asks for more than these limits.

Reference-checked

261GPU parity cases on each of 3 browser engines

Checked against a CPU reference

A plain TypeScript implementation of the networks defines what "correct" means, and PyTorch outputs of every shipped model are the fixtures. The GPU suite runs those on Chrome (Metal), on Chrome's software renderer and on the Safari engine, plus odd sizes, bands, device loss and misuse. The 894 shader texts the engine generates are pinned as well, so a kernel that changes has to say so.

A per-frame time more than 10% over the recorded baseline fails the build.

Model file

4 KBto 2.4 MB per model, fetched when a tier first runs

A new model is a file

Models ship as standard ONNX files: the network written in standard ONNX operators, the weights as 16-bit tensors, in the open format other model tools read. The engine reads the file with its own importer and plans memory, kernels and bands from the graph, so adding or swapping a tier is a converted weight file, and two weight sets of one network can be blended at load time.

The importer maps the whole family onto the engine’s own kernels: 3×3 and 1×1 convolutions, ReLU / LeakyReLU / PReLU / CReLU, dense concatenation, pixel shuffle.

SR Engine vs ONNX Runtime Web on WebGPU

We ran SR Engine and ONNX Runtime Web, a general-purpose runtime for machine-learning models in the browser, on the same weights and the same frames, taking turns on the same device, with ONNX Runtime Web on its WebGPU backend in half precision. On Fast's small networks SR Engine is 4–14× faster. On the heavier Pro and Ultra it is 1.6–2.4× faster on the Mac, and 1.8× faster running Pro at 480p on a budget Android phone with a Mali graphics chip.

SR Engine ONNX Runtime Web Compare the two bars within a row, not across rows

Job · where it ranTime per frameResult
Fast · GPU time per frame, upload included
Fast, Strong · 1080p input Apple silicon Mac · ChromeSR Engine 24.1 ms ONNX Runtime Web 158.5 msSR Engine 6.6× faster
Fast, Light · 1080p input Apple silicon Mac · ChromeSR Engine 6.6 ms ONNX Runtime Web 67.8 msSR Engine 10× faster
Fast, Strong · 1080p input Apple silicon Mac · WebKit, Safari's engineSR Engine 33 ms ONNX Runtime Web 289.5 msSR Engine 8.8× faster
Fast, Light · 1080p input Budget Android phone · MaliSR Engine 136.4 ms ONNX Runtime Web 1,915.5 msSR Engine 14× faster
Fast, Medium · 720p input Budget Android phone · MaliSR Engine 180.2 ms ONNX Runtime Web 1,415.7 msSR Engine 7.9× faster
Fast, Strong · 720p input Budget Android phone · MaliSR Engine 506.8 ms ONNX Runtime Web 2,043.7 msSR Engine 4.0× faster
Pro and Ultra · whole time per frame
Pro · 720p input Apple silicon Mac · ChromeSR Engine 467 ms ONNX Runtime Web 734 msSR Engine 1.6× faster
Pro · 720p input Apple silicon Mac · WebKit, Safari's engineSR Engine 552 ms ONNX Runtime Web 1,299 msSR Engine 2.4× faster
Ultra · 720p input Apple silicon Mac · ChromeSR Engine 934 ms ONNX Runtime Web 1,455 msSR Engine 1.6× faster
Ultra · 720p input Apple silicon Mac · WebKit, Safari's engineSR Engine 1,026 ms ONNX Runtime Web 2,459 msSR Engine 2.4× faster
Pro · 480p input Budget Android phone · Mali, SR Engine in 8 bitsSR Engine 5,876 ms ONNX Runtime Web 10,583 msSR Engine 1.8× faster

Median time per frame, measured by us: the same weights and frames on both sides, run in turns on the same device and browser, with SR Engine as the upscaler runs it and ONNX Runtime Web on its WebGPU backend in half precision.

Correctness · Fast, Light · Budget Android phone · Adreno 6 series

The frame from ONNX Runtime Web on a budget Android phone with an Adreno 6-series graphics chip: an anime eye and dark hair hidden behind dense vertical black stripesONNX Runtime Web · PSNR 5.6 dB
The same frame from SR Engine on the same phone: the green eye and the strands of dark hair drawn cleanlySR Engine · PSNR 99.5 dB
A correctness check on a budget Android phone with an Adreno 6-series graphics chip, whose graphics chip has no half precision, so both engines ran in 32-bit floats. One frame through Fast's Light network, the same weights on both sides: ONNX Runtime Web returned vertical stripes every time, a PSNR of 5.6 dB against the CPU reference (higher means closer to it), while SR Engine matched the reference at 99.5 dB. ReLIFE final PV (2018), TMS Entertainment / comico — CC BY 3.0 via Wikimedia Commons

Against the engine we ran before

PropertySR EngineThe open-source engine we ran before
Memory per pixel, large 2× network112 B368 B
720p → 4K second step, feature maps413 MBor 128 MiB in 4 bands≈ 1.36 GB
1080p → 4K, large 2× network21.1 ms / frame63.3 ms / frame
1080p → 4K, small 2× network3.1 ms / frame8.4 ms / frame
8-bit integer path on phonesYes — where the chip has an integer dot productNo
4× repair networksYesNo — 2× only
Precisionf16 where offered, f32 otherwisef32 only
Frame edgesEvery pixel computedLast columns skipped, edge colour wrapped
How a frame is handed to the GPUIn slices of GPU timeabout 12 ms each on a phone, 50 ms on a computer, measured as it runsOne submission per layer
First-run setup, large 2× network1.1–1.7 sonce per device: the block shape is measured on your GPU and keptFixed block shape
Model filesStandard ONNX, 4 KB to 2.4 MB per tierThe same networks, as JSON weights

Both engines measured by us on a Mac with Apple silicon in Chrome, same networks.

FAQ

Which quality should I pick?
Fast for clean footage, anime, screen recordings, AI-generated video and most clips on a phone: it crispens edges and changes nothing else. Pro for downloaded, re-compressed or lightly noisy sources up to 720p: it removes the blocks and the noise. Ultra only for heavy grain and old film. Under Start the panel recommends a setting for your clip: Pro when it finishes within 8× the clip's length on your device, otherwise Fast — never Ultra, and Fast on a phone. Strength, under Quality, sets how hard Fast and Ultra work: Auto takes Fast's strongest network that finishes within the same 8× the clip's length (Medium at most on a phone) and Ultra's Strong, and Light is the quickest. When unsure, press "Preview this frame" and compare.
How long does upscaling take?
Frames × time per frame; the page measures your own device before it starts and shows the estimate next to Start. One minute of clip at 30 fps, 1080p to 4K with Fast, takes 1–2 minutes on an NVIDIA RTX 30–50 series card, 2–4 on a Mac with Apple silicon and 5–12 on an iPhone; Pro and Ultra take longer. There is no length limit: a long run saves its progress and the home page offers to continue it after a reload. To do only a part, drag a window of 5 seconds to 10 minutes on the filmstrip, or type where it ends; a long job offers a short trial first: 5 to 30 seconds, as much as your device upscales in about 1 min.
Why is 4K not offered for my clip?
The biggest size offered is 4× the clip's longer side: 4K needs a source of at least 960 px on that side, 2K at least 640 px, 1080p at least 480 px; a clip under 480 px still goes to 1080p — enlarged 4×, the rest filled by the final resize, and it starts on Pro, which repairs a blurry small clip. A clip that is already 4K gets 4K repair on a computer whose graphics chip has half-precision shaders: Pro cleans it at its own size. Anywhere else the panel says it is already 4K and offers no size. A phone is offered the same sizes as a computer.
Why don’t I see Ultra (or Pro)?
Pro and Ultra need hardware rendering. On a computer whose graphics chip has no half-precision shaders (shader-f16), Pro runs in full precision and takes longer, and Ultra is not offered; on a computer’s graphics chip both take a source of any size, and the time next to Start is measured on your clip. On a phone, Pro takes clips up to 854 px (480p) and Ultra runs on computers: a 30-second 720p clip keeps the graphics chip working for about 29 minutes with Pro on a budget phone and 50 with Ultra even on a flagship, and runs that heavy can make the browser switch the graphics chip off. The panel lists only the tiers your device runs. In Firefox on a computer the reason is usually the browser, not the device: Firefox on Windows runs Fast only, as does a Firefox without half-precision shaders, and Chrome or Edge on the same computer run Pro and Ultra.
Which browsers and devices work?
Chrome and Edge 113 or newer on Windows, macOS and ChromeOS; Chrome 121, Edge 153 and Samsung Internet 29 or newer on Android 12 or newer; Safari 26 on macOS, iOS and iPadOS; Firefox 141 or newer on Windows and 145 or newer on a Mac with Apple silicon. With no graphics chip to use — no WebGPU, or hardware acceleration switched off — Chrome, Edge, Samsung Internet and other Chromium browsers run the job on the processor: slower, and still on your device. Browsers inside apps — TikTok, WeChat, Instagram — usually have no WebGPU; open the same link in Chrome or Safari. Where an app's own browser does run it, an iPhone saves the result through the share sheet ("Save Video" puts it in Photos). An Android app's own browser, or an iPhone app's that can't share a file, can't save the finished video: there the page shows the two steps that open it in your phone's browser, with an arrow at the app's menu. On Android, a phone maker's own browser that can't run the upscaler is told to open the page in Chrome, Edge or Samsung Internet. If Chrome or Edge switch WebGPU off after the graphics chip stops responding, the panel says whether it comes back by itself or after a browser restart, and Start turns back on when it does.
What happens to an AVI, VOB or camcorder file?
The page makes an MP4 of it first, inside your browser. AVI, MPG, VOB, WMV, FLV and DV — and the Motion JPEG of old cameras or the MPEG-4 and H.263 of old phones — are converted with FFmpeg's own decoders and deinterlaced on the way: the beginning opens within seconds while the rest is prepared, and an AVI or FLV that already holds H.264 has its pictures copied untouched. A camcorder's MTS, M2TS or TS is rewrapped with its pictures untouched. Nothing is uploaded; "Old formats" above has the details.
Is the sound kept?
Yes, wherever the format you download can carry it: the sound is copied as it is, or re-encoded where the format needs its own codec (Opus for WebM). Where this browser can carry it into neither, the panel says so before you start and names the formats that keep it. A camcorder's Dolby Digital (AC-3) sound becomes AAC (Opus in Firefox) when the file opens, so every browser plays it, and a file the page converts — an AVI, a VOB, a ProRes MOV — keeps its first sound track.
What happens if I close the page or lock my phone?
A run the estimate puts past 90 seconds saves finished pieces as it goes: close the tab and, within 24 hours, the home page offers to continue it from the last saved piece. While a job runs the page keeps the screen awake. On Android, switching to another app pauses the job and saves it at that frame: come back and it goes on from there by itself, even if the phone took back the graphics chip meanwhile; if the phone closed the browser, continue it from your recent videos on the home page. A run that comes back much slower restarts from that save on its own, and if it stays slow the panel tells you to restart the browser. On an iPhone or iPad, locking the screen or switching apps stops it, and the page continues it from its last save by itself once you are back — up to 3 times in a row.
Does it deinterlace?
The old formats, yes: AVI, MPG, VOB, WMV, FLV and DV files are deinterlaced as the page converts them, all the way through, even when the interlacing only begins partway in. The upscale itself keeps the fields as they are, so an interlaced MP4, MKV or camcorder file (MTS) is upscaled with them — the panel says so, and moving edges can show fine comb lines. For the cleanest result, deinterlace such a file first in HandBrake (Filters → Deinterlace), or try a short window to see how much it shows.
My file is over 2 GB — will it open?
On a computer, yes: a file of any size opens, and a long video runs the way a short one does. On a phone or a tablet one file can be up to 2 GB: open the page on a computer for a bigger one, or trim the part you need in the Photos or Gallery app and choose that clip. Length is never the limit.
Where is my upscaled video?
In this browser's private storage for the site until you download it: the Download button in the finished view saves it wherever your downloads go, named after the original, its size and its quality (holiday_4K_Pro.mp4); a batch in Chrome or Edge can save each video straight into a folder you pick. On an iPhone or iPad, Safari puts it in the Downloads folder of the Files app, not in Photos; once you tap Download, the line under it says how to add it to Photos. It stays there for 24 hours, as the finished view says — come back to the home page within that time: "View" under "Choose video" reopens it to play, compare with the original and download again, and × deletes it. While a result is not downloaded or a job is running, leaving the workbench — its close button, the logo, another video, the back gesture — asks first.
How do I get the cloud version?
It is invitation-only. Fill in the form at the end of this page (or email the cloud address with the subject "Cloud access"): what you tried locally and what was missing, the footage, your device. Requesting is free. A cloud version needs an account and, for some models, a fee — the invitation says what applies.

More answers: FAQ · Supported formats · System requirements

5. Cloud version

The cloud version is for what your own device cannot do in reasonable time: a whole episode with Pro or Ultra on an ordinary laptop, a batch that would keep your computer busy for days, a device without WebGPU, a phone whose graphics chip is not offered the repair tiers. Access is by invitation — tell us what you tried and what was missing, and we invite first the requests it can already serve. Requesting is free; the cloud version needs an account and, for some models, a fee, and the invitation says what applies before you send any footage.

Cloud access

Filled in from this browser when you start typing (graphics chip, browser, encoders, free storage) — edit or delete anything. Add the make and year if you like: "Dell laptop, 2020", "Samsung A52".
Email the requestopens your mail app with everything filled in

Or: Gmail · Outlook ·

What we will receive
Cloud access

Your email: —
What you tried locally, and what was missing: —
Footage: type · resolution · length: —

Page: https://4kvideoupscaler.com/how-it-works

Or write yourself, subject "Cloud access", to cloud@4kvideoupscaler.com

Your address is used to contact you about cloud access and for nothing else (privacy policy).