Image Resizer
Resize each original by percentage or locked ratio with smooth or pixel sampling, use a common canvas, or try local 2×/4× AI upscaling.
Your files
Files stay on your deviceDrop files herePick a file to start
PNG, JPEG, WebP (static) · up to 20 MiB per file
Limits & supported formats
ZIP or folder: up to 8 supported images, 40 MiB expanded total. Nested archives and unsupported entries are rejected.
Input: up to 8 files · PNG, JPEG, WebP (static) · 20 MiB each, 40 MiB combined.
Static images: up to 2,048 px per side. Animated inputs are not supported here.
Still image output: up to 3,072 px per side and 9,000,000 pixels. No animation is exported.
Larger output is available for PNG/JPEG only: up to 3,072 px per side and 9 million pixels, within the memory estimate. Input, AI and other output formats keep their existing limits. Enlarging an ordinary resize does not recover detail.
Detailed budgets: 40,000,000 decoded pixels, an estimated 192 MiB working allocation and 64 MiB per ZIP. These estimates help reject large jobs; they do not guarantee a browser process memory peak.
Files are processed one at a time with shared settings. A failed file does not stop the others.
Add a file to get started.
Three steps. One moving picture.
01Add one image or a batch
02Choose per-original pixels or percentage, a common canvas, sampling, or an AI factor
03Check the new dimensions and download
Image Resizer
Scale each original by percentage or pixels with its own locked ratio, or use one common output canvas. Smooth sampling interpolates alpha; Pixel art uses nearest-neighbor sampling. Other output formats retain the 2048 px side limit. Choose Contain for padding, Cover for a filled crop, or Stretch when distortion is intentional. The optional ESRGAN Slim model accepts images up to 512 × 512 px and estimates new detail at 2× or 4×; it can invent artifacts and cannot recover the original missing information. Alpha is resized separately. Ordinary PNG/JPEG output can reach 3,072 px per side and 9 million pixels within a separate memory estimate; input and AI limits remain smaller.
Choosing your settings
Enlarging does not add real detail. Per-original mode uses each image’s dimensions; common canvas mode uses shared dimensions; original metadata is not copied.
If the result needs another pass
Choose PNG to preserve transparent padding, or JPEG with a background color. Check portrait and landscape inputs separately.
Good to know
Can enlarging recover missing detail?
Ordinary enlargement interpolates pixels. AI upscaling estimates patterns learned from other images; neither can guarantee recovery of the real missing detail.
How does a mixed portrait and landscape batch scale?
Per-original percentage and locked ratio use each source image’s dimensions. Common canvas uses one width and height with Contain, Cover or Stretch. Choose PNG to preserve transparent padding.