A raw, uncompressed 4K video would fill hundreds of gigabytes per minute — clearly not what happens on your phone. So how does a one-minute clip shrink from a theoretical 50 GB to a real 400 MB, and then again to 120 MB after compression? The answer involves two big ideas: redundancy (video repeats itself constantly) and perception (your eye can't see everything anyway). Here's how it actually works, in plain English.
To appreciate compression, picture what video looks like without it. A single 4K frame is about 8.3 million pixels. At 8 bits per color channel and three channels (RGB), that's roughly 25 MB per frame. At 30 frames per second, one second is 750 MB. One minute is 45 GB. Obviously, no phone could store that.
The reason real iPhone 4K footage is "only" ~170 MB per minute (not 45 GB) is that the camera already compresses it during recording, using HEVC. And then a tool like VidZip compresses it further to ~50 MB by trimming the headroom. Both rounds of compression lean on the same principles.
Look at any single frame of video. Most of it is similar. A blue sky is one shade of blue repeated across millions of pixels. A wall is mostly the same color. A face has large patches of similar skin tone. Storing every pixel individually is wasteful — the encoder can describe a block of similar pixels with one number instead of one number per pixel.
This is spatial compression, and it's essentially what JPEG does for photos. The frame is divided into blocks (typically 8×8 or 16×16 pixels), and each block is described with a handful of numbers representing its average color, brightness, and variation. Smooth areas need very little data; detailed areas need more. Result: a single frame shrinks dramatically.
Spatial compression alone gets you to about photo-level efficiency. The real magic of video compression comes from the time dimension. Watch a video frame by frame: between one frame and the next, most of the picture doesn't change. A person talking has the same background, the same clothes, mostly the same face — only their mouth moves.
So instead of storing 30 full frames per second, the encoder stores:
A typical video might have an I-frame every second and P/B-frames for the other 29 frames. The P/B-frames are tiny because they only describe motion and changes. This is why a video can be 10× smaller than the equivalent stack of JPEGs — it's not storing 30 pictures per second, it's storing one picture plus 29 sets of changes.
VidZip uses these same principles (modern codec, smart bitrate control) to shrink your videos 70-85% with no visible quality loss. You don't need to understand the math — just pick a preset.
Download VidZip Free →Even with redundancy removed, the encoder has to make a choice: keep every detail perfectly, or throw some away to shrink the file further?
Lossless compression preserves every pixel. The decompressed video is bit-for-bit identical to the original. But because most video is already compressed when recorded, lossless recompression saves very little — usually 5-10%. Not worth it.
Lossy compression takes advantage of the fact that your eye can't see everything. The encoder throws away detail it judges invisible: subtle color differences the eye can't distinguish, fine high-frequency texture that disappears at viewing distance, brightness variations below the perception threshold. Done well, the loss is invisible to humans — but the file size shrinks dramatically (50-95%). Almost all the video you watch and shoot is lossy.
This is the key insight behind "lossless-looking" compression: you're not actually preserving every bit, you're throwing away bits the eye couldn't see anyway. The result looks identical at normal viewing distances.
If you push compression too far (bitrate too low for the content), the encoder runs out of data to describe each block of pixels. Blocks become visible as actual squares — a phenomenon called artifacting or "macroblocking."
It shows up first in:
Well-tuned compression avoids this by picking a bitrate high enough to describe the content without exposing the blocks. VidZip's presets are calibrated to stay on the safe side of this threshold for typical iPhone footage.
The principles above — spatial redundancy, temporal redundancy, lossy discarding — are implemented by specific algorithms called codecs. H.264, H.265 (HEVC), AV1, and VP9 are all different implementations of the same underlying ideas, with different tradeoffs in efficiency, complexity, and licensing.
Newer codecs (HEVC, AV1) use more sophisticated versions of these techniques — bigger block structures, better motion prediction, more efficient math — to squeeze more quality out of each bit. But the core ideas haven't changed since the early days of digital video. Our codec guide compares them in detail.
All these techniques — spatial, temporal, lossy — operate under one master constraint: bitrate. The bitrate is how many bits per second the encoder is allowed to use. The encoder's job is to spend that budget wisely: enough to describe important detail, not so much that the file bloats.
This is why bitrate is the single most useful number in video compression. Lower the bitrate and the file shrinks (but may eventually artifact). Raise it and quality improves (but the file grows). Tools like VidZip essentially manage bitrate for you — you pick a target quality or file size, and the app picks the right bitrate to hit it.
When you compress a video with VidZip, here's the simplified chain of events:
None of this requires you to understand the math. You pick a preset; VidZip picks the right bitrate and codec settings to hit the quality/size tradeoff you want; the encoder does the rest.
Video compression exploits two kinds of redundancy. Spatial redundancy: within a single frame, neighboring pixels are often similar (a blue sky is mostly the same blue), so the encoder describes blocks of pixels once instead of pixel-by-pixel. Temporal redundancy: across frames, most of the picture stays the same — only the moving parts change. The encoder stores full keyframes occasionally and describes other frames as changes from nearby frames, dramatically cutting the data needed.
Lossless compression preserves every pixel perfectly — what you decompress is identical to what went in. Savings are small (5-10%). Lossy compression throws away data the encoder judges invisible to the eye, achieving 50-95% size reduction. Done well, lossy is visually indistinguishable from the original. Almost all video you watch and shoot is lossy.
Because of time. A photo is one frame; compression can only exploit spatial redundancy (similar neighboring pixels). Video adds temporal redundancy — most of the picture doesn't change from frame to frame, so the encoder describes changes rather than repeating the whole picture 30 times per second. That extra dimension is where the big savings come from.
I-frames (intra-coded) are full pictures stored completely, like a JPEG. P-frames (predictive) describe only what changed since the previous frame. B-frames (bidirectional) describe changes by referencing both previous and upcoming frames. A typical video uses mostly P and B frames with an occasional I-frame for reference, drastically reducing total data.
Video codecs divide each frame into blocks (often 8×8 or 16×16 pixels) and describe each block with a few numbers. When bitrate is too low, the encoder runs out of numbers to describe fine detail, so blocks become visible as visible squares — especially in dark gradients, fast motion, or fine textures like hair and grass. This is called artifacting.
VidZip handles all the math so you don't have to. Pick a preset, tap Compress, get a smaller file that looks the same.
Get VidZip →