简体中文

Image Converter & WebAssembly Compressor

Batch-convert and compress JPEG, PNG, WebP, AVIF, SVG, TIFF and camera RAW files with pinned local codecs, workers and measured results.

选择格式,以及“质量”实际要付出什么代价

Four raster formats cover most web delivery, while sanitized SVG output preserves safe vector source. Picking the right output matters more for file size than any slider.

JPEG is built for photographs. It throws away detail your eye is unlikely to miss, which is why a photo saved as JPEG is a fraction of the size of the same photo as PNG. It cannot store transparency, and it is lossy at every quality setting - there is no such thing as a lossless JPEG here.

PNG 专为平面图形而设计:屏幕截图、徽标、图表、任何具有锐利边缘或大面积单一颜色的内容。它的编码器是无损的,因此它准确地存储给定的像素,并且支持透明度。将照片另存为 PNG ,您通常会得到比 JPEG 大几倍的文件,但没有明显的增益。

WebP usually beats JPEG for a similar visual quality and can preserve transparency. AVIF can be smaller again, but its local encoder takes longer. Both outputs are lossy in this studio.

The quality slider applies to JPEG, WebP and AVIF. Around 80 is a useful starting point for photographs, but the honest way to choose is to inspect the measured output, SSIM and PSNR values, then adjust.

  1. Add as many as 50 files, up to 100 MB each. Everything is read locally; nothing is uploaded.
  2. Pick PNG, JPEG, WebP, AVIF or safe SVG output. The panel tells you whether that operation is lossy and why.
  3. Resize by percentage, exact pixels or a social preset, with the aspect ratio locked unless you deliberately unlock it.
  4. Keep Privacy Clean enabled to emit a fresh encoded file without source metadata.
  5. Inspect real before and after bytes, pixel metrics and the comparison view, then download one result or stream the batch as a ZIP where File System Access is available.

为什么重复编辑会降低图像质量,以及如何避免它

每次有损图像被解码并再次保存时,编码器会在已经烘焙到像素中的近似值之上生成一组新的近似值。这就是所谓的代际损失,而且它会加剧。打开 JPEG,保存它,打开它,再次保存它,然后重复:经过足够多的轮次后,结果明显斑驳,尤其是在文本和锐边周围。

这很重要,因为它很容易意外发生。在一个应用程序中裁剪照片,在另一个应用程序中调整其大小,然后从第三个应用程序导出它,可以将三轮 JPEG 编码放入一个只需要一轮的文件中。

避免这种情况的方法是保留原始内容并始终从中重新编辑,而不是从上次导出的内容中重新编辑。如果需要中间工作文件,请使用PNG,因为它的编码器是无损的,不会增加另一轮损失。然后在最后从原始文件中进行一次有损导出。

将 JPEG 转换为 PNG 不会撤消任何操作。 JPEG 丢弃的细节消失了,而 PNG 将忠实地将降级的像素保留在更大的文件中。 PNG 是您仍在编辑的内容的工作副本的正确选择,而不是已压缩内容的修复工具。

调整大小也值得理解。缩小图像会对其进行重新采样,即使输出格式无损,也会改变像素值,并且无法通过再次放大来恢复丢弃的细节。超出原始尺寸的放大永远不会增加真正的细节;它只是插入已经存在的内容。

常见问题

我的图像是否上传到服务器?
No. Pinned codecs, the RAW decoder, resizer and batch workers run on this device. File bytes, names and conversion settings are not sent to analytics or an image service.
这里的任何转换都是无损的吗?
PNG encoding is lossless, provided you do not also resize. Safe SVG cleanup keeps vector geometry but removes active and external content. JPEG, WebP and AVIF outputs are lossy, and any resize resamples pixels regardless of output format.
为什么我的文件变大而不是变小?
通常是因为您将照片转换为 PNG,它精确存储每个像素,并且无法与 JPEG 或 WebP 在照片内容上竞争。如果您将质量提高到高于原始编码的水平,也可能会发生这种情况。无论哪种方式,该工具都会报告实际测量的尺寸,而不是隐藏增长。
我应该使用什么质量设置?
For photographs, around 80 is a useful starting point. Compare measured bytes and perceptual metrics at a couple of settings rather than trusting one rule, because the right answer depends on the image and output codec.
当我导出到 JPEG 时,透明度会发生什么变化?
JPEG 无法存储 Alpha 通道,因此透明区域将填充您在编码前选择的背景颜色。默认为白色。如果需要透明度,请使用 PNG 或 WebP。
我可以转换 GIF 动画吗?
GIF 被接受作为输入,但仅使用其第一帧并且不保留动画。不提供 GIF 作为输出格式,因为生成带有 .gif 扩展名的静态图像会产生误导。
我一次可以转换多少张图像?
A queue is limited to 50 items and each file is limited to 100 MB. Processing is local, so high-resolution RAW batches on a phone can still be slow. The tool applies a device-aware pixel budget before allocating an unsafe canvas.
这会删除 EXIF 元数据吗?
Yes when Privacy Clean is enabled. Raster exports are freshly encoded from decoded pixels, and sanitized SVG exports remove editor metadata, scripts, event handlers and unsafe external references.
Which RAW files can I open?
The local LibRaw worker accepts DNG, CR2, CR3, NEF, ARW and RW2 inputs, plus TIFF. Camera variants outside those families can still be unsupported, and very large RAW files are subject to the same 100 MB and device-memory limits.

该工具不会做什么