Using bicubic to scale a screenshot of text up by exactly 2x, where nearest neighbour would have stayed perfectly crisp.
Resolution and scaling
Bicubic interpolation
Also called bicubic, bicubic resampling or bicubic smoother. Here is what it means, when it changes what you export, and what people get wrong.
Resolution and scalingRuns on your deviceUpdated 10 August 2026
The short answer
Bicubic interpolation works out each new pixel by fitting a smooth cubic curve through a 4 x 4 block of sixteen surrounding pixels. It is the default resizing method in most image editors because it keeps edges reasonably firm while avoiding blockiness. It costs more than bilinear and can ring slightly at hard edges.
Updated
The short version
Bicubic interpolation at a glance
| Quick fact | Detail |
|---|---|
| Neighbourhood | 4 x 4, sixteen pixels |
| Against bilinear | Four times the samples, firmer edges |
| Typical role | Default photo resize in most editors |
| Variants | Smoother for enlarging, sharper for reducing |
| Known artefact | Ringing or haloes at high-contrast edges |
Slide the table sideways to see every column.
Why sixteen pixels instead of four?
Bilinear interpolation reads a 2 x 2 block and draws straight lines between the samples, so every edge comes out as a flat ramp and detail flattens with it. Bicubic reads a 4 x 4 block and fits cubic polynomials in both directions, matching not only the values of the neighbouring pixels but the rate of change between them. The curve therefore continues through an edge rather than clipping it flat, which is what keeps the result looking like a photograph after a resize. The cost is four times as many samples per output pixel, which barely registers on modern hardware.
What are bicubic smoother and bicubic sharper?
Same maths, different tuning of the curve. Bicubic smoother biases towards a gentler curve and is offered for enlarging, where it produces fewer haloes and a cleaner gradient at the cost of some crispness. Bicubic sharper biases the other way and is offered for reducing, where a resize naturally softens and a little extra contrast is welcome. On strong edges it can push values past black or white and clip them. Photoshop's automatic setting simply picks between them based on whether you are going up or down.
Where will you see ringing?
A cubic curve fitted across an abrupt jump in brightness overshoots on the way in and undershoots on the way out. The visible result is a light halo just outside a dark edge and a darker line just inside it, repeated faintly further out. It shows up most on black text over white, on logos and line art, and on a cut-out product sitting against a pure white background, where a faint grey outline can appear after resizing. Lanczos rings more strongly; bilinear does not ring at all but is markedly softer.
If a halo appears around a cut-out, resize the layered file before flattening, or resize a little less aggressively, rather than trying to sharpen it away.
What people get wrong about bicubic interpolation
Each one is a real failure mode, not a style preference.
Blaming a faint white outline around a cut-out product on the background removal when it is ringing from the resize.
Assuming bicubic sharper is always the better option, when on an enlargement it exaggerates noise and edge overshoot.
Questions people ask
Bicubic interpolation, answered
The follow-up questions people search for once they have the definition.
Is bicubic better than bilinear?
For photographs, yes. Bicubic reads sixteen neighbours rather than four and keeps edges noticeably firmer. Bilinear is cheaper, which is why it is used for live previews, video scaling and anything that has to run at frame rate, where the softness is an acceptable trade.
Should I use bicubic or Lanczos for shrinking?
Lanczos usually holds more fine detail when reducing, since it reads a wider neighbourhood. Bicubic is the safer choice when the image has hard, high-contrast edges, because Lanczos produces more visible haloes around them. Try both on one image and compare at 100 percent.
Why does my resized logo have a faint outline?
That is ringing. The cubic curve overshoots where brightness changes abruptly, leaving a light halo outside the shape and a dark line inside it. Reduce the resize amount, use bilinear for that asset, or export vector artwork at the size you need instead.
Does bicubic add detail to an image?
No. Every output pixel is a weighted blend of pixels already in the file, so no new information enters. Bicubic simply distributes existing detail more convincingly than simpler filters do. Adding genuine-looking detail requires a learned super-resolution model.
Same cluster
More from the glossary
Neighbouring entries, so every term in the set is one hop from every other.
Bit depth
Bit depth is how many bits an image uses to store each colour channel, which sets how many distinct levels that channel can hold.
Banner image
A banner image is the wide letterbox artwork across the top of a shop, profile or channel page.
BMP
BMP is Microsoft's raster image format, which stores each pixel's colour value directly with little or no compression.
From the studio
Tools that do this
Each one runs on your own device, so the photo stays where it is.
Keep reading
Elsewhere on Photomop
Platform image rules
Image size, file type and background rules for 22 selling platforms, each one sourced.
Read itConvert image
JPG, PNG, WebP and AVIF, with an honest answer about HEIC and alpha.
Read itAll the tools
The whole toolbox in three families, with what each one needs before it can run.
Read itOn-device privacy
The pipeline in five steps, what leaves your device, and how to check it yourself.
Read it
Now do it to a photo
Photomop is a photo studio that runs on your own device. Resize, crop, compress, convert, batch edit, remove a background and change one all work in the browser tab you are reading this in, at full resolution and with no watermark. The on-device page shows you how to check that the photo stays put.