Resize vs crop — know which you need
Resize changes the dimensions of the entire image — everything gets proportionally smaller or larger. Crop removes parts of the image to change its size or aspect ratio.
If you need a 1200×630 pixel image for a social media card but your original is 2000×1500, you need to both resize (scale down) and crop (change the aspect ratio). Most resize tools handle both, but it is worth knowing what you need before you start.
The algorithm that matters
When you scale an image down, the tool has to decide how to represent multiple original pixels as one new pixel. Different algorithms make this trade-off differently:
- Nearest neighbor: fastest, but produces blocky results — each pixel becomes a larger block. Good for pixel art, bad for everything else.
- Bilinear: averages surrounding pixels. Produces smooth results but can look slightly blurry on text and sharp edges.
- Bicubic / Lanczos: the best quality algorithm — considers more surrounding pixels to preserve sharp edges and fine detail. Slower but produces the sharpest results.
For photos, bilinear is usually fine. For screenshots with text, logos, and graphics, use bicubic or Lanczos if the tool offers it. The free image resizer uses high-quality downsampling for clean results.
Standard sizes for common use cases
| Use case | Width × Height (px) |
|---|---|
| Blog post featured image | 1200 × 628 |
| Twitter/X card image | 1200 × 675 |
| Facebook post image | 1200 × 630 |
| Instagram square post | 1080 × 1080 |
| LinkedIn post image | 1200 × 627 |
| YouTube thumbnail | 1280 × 720 |
| E-commerce product (square) | 1000 × 1000 minimum |
| Email inline image | 600 px wide maximum |
Maintaining aspect ratio
When resizing, always lock the aspect ratio unless you specifically need to distort the image. A distorted image looks unprofessional and is obvious to viewers.
If you need a 1:1 square from a 4:3 landscape photo, crop first to a square, then resize to the target dimensions. Stretching a 4:3 photo to 1:1 will look wrong.
Upscaling: when it works and when it doesn't
Making an image larger (upscaling) is harder than making it smaller. Standard resizing adds pixels by interpolating between existing ones, which produces a blurry result at high upscale ratios.
AI-based upscalers (like Real-ESRGAN) use a neural network to intelligently add detail when enlarging. They produce dramatically better results than standard upscaling — but they are slower and most free tools have resolution limits.
As a rule: if you need to upscale more than 2×, AI upscaling is worth it. For less than 2×, standard bicubic upscaling produces acceptable results.
After resizing: always compress
Resizing alone does not always reduce file size proportionally. A photo resized to half dimensions may only reduce file size by 60–70% if compression settings remain the same. After resizing, run the image through the free image compressor to minimize the file size further.
Summary
Use the free image resizer to hit exact pixel dimensions. Crop before resizing if you need to change the aspect ratio. Use bicubic or Lanczos for sharp results on text and graphics. Compress after resizing for smallest file size.