pixo

Pure Go image processing library. Drop-in replacement for nfnt/resize and disintegration/imaging.

Install

go get github.com/agentine/pixo

Features

• Resize with 7 interpolation filters (NearestNeighbor, Box, Linear, CatmullRom, MitchellNetravali, Lanczos2, Lanczos3)
• Fit and Fill with aspect ratio preservation
• Rotate (arbitrary angle + 90/180/270), Flip, Transpose, Transverse
• Crop with anchor-based positioning (9 anchor points)
• Adjust brightness, contrast, gamma, saturation, sigmoid, hue
• Effects: Gaussian blur, sharpen, grayscale, invert, 3x3/5x5 convolution
• Compose: paste, overlay with alpha blending
• I/O: Open/Save/Encode/Decode with format detection, EXIF auto-orientation
• Formats: JPEG, PNG, GIF, BMP, TIFF (read/write), WebP (read)
• Compatibility packages for painless migration from nfnt/resize and disintegration/imaging

Usage

Resize an image

package main

import (
    "log"
    "github.com/agentine/pixo"
)

func main() {
    img, err := pixo.Open("input.jpg")
    if err != nil {
        log.Fatal(err)
    }

    // Resize to 800x600
    resized := pixo.Resize(img, 800, 600, pixo.Lanczos3)
    pixo.Save(resized, "output.jpg", pixo.JPEGQuality(90))

    // Fit within 800x800, preserving aspect ratio
    fitted := pixo.Fit(img, 800, 800, pixo.Lanczos3)
    pixo.Save(fitted, "fitted.jpg")

    // Create a 200x200 thumbnail
    thumb := pixo.Thumbnail(img, 200, 200, pixo.Lanczos3)
    pixo.Save(thumb, "thumb.jpg")
}

Transform

rotated := pixo.Rotate90(img)
flipped := pixo.FlipH(img)
cropped := pixo.CropCenter(img, 500, 500)

Adjust colors

bright := pixo.AdjustBrightness(img, 20)   // +20%
contrast := pixo.AdjustContrast(img, 30)   // +30%
gray := pixo.Grayscale(img)
sharp := pixo.Sharpen(img, 1.0)
blurred := pixo.Blur(img, 2.0)

Compose images

result := pixo.Overlay(background, foreground, image.Pt(100, 100), 0.7)

Migration from nfnt/resize

Change your import path:

// Before
import "github.com/nfnt/resize"

// After
import "github.com/agentine/pixo/compat/resize"

The API is identical:

// Before
m := resize.Resize(800, 0, img, resize.Lanczos3)

// After — same code, just different import
m := resize.Resize(800, 0, img, resize.Lanczos3)

Available interpolation functions: NearestNeighbor, Bilinear, Bicubic, MitchellNetravali, Lanczos2, Lanczos3.

Migration from disintegration/imaging

Change your import path:

// Before
import "github.com/disintegration/imaging"

// After
import "github.com/agentine/pixo/compat/imaging"

All functions have the same signatures:

// Before
dst := imaging.Resize(src, 800, 600, imaging.Lanczos)
dst = imaging.Blur(src, 3.5)
dst = imaging.AdjustBrightness(src, 20)
err := imaging.Save(dst, "output.jpg")

// After — same code, just different import
dst := imaging.Resize(src, 800, 600, imaging.Lanczos)
dst = imaging.Blur(src, 3.5)
dst = imaging.AdjustBrightness(src, 20)
err := imaging.Save(dst, "output.jpg")

Filters

Filter	Description
NearestNeighbor	Fastest, pixelated
Box	Fast, averaging
Linear	Bilinear, good balance
CatmullRom	Bicubic, sharp
MitchellNetravali	Bicubic, balanced
Lanczos2	High quality, a=2
Lanczos3	Highest quality, a=3

Benchmarks

Run benchmarks:

go test -bench=. -benchmem

License

MIT