Clear answers about image quality, resampling filters, WebP, JPEG, PNG, file size, and browser-based processing. Every answer is shown here—nothing is hidden behind a collapsed accordion.
What is the difference between image resizing and image compression?
Resizing changes the pixel dimensions of an image, for example from 3000 × 2000 to 1500 × 1000. Compression changes how the resulting pixel data is stored so the file can use fewer bytes. They solve different problems: resizing reduces the number of pixels, while encoding determines how efficiently those pixels are represented. Compressy lets you control both in one export.
Which resampling filter is best for photographs?
Lanczos-3 is a strong starting point for detailed photographs, especially when reducing dimensions. It can retain crisp texture, although very sharp, high-contrast edges may show faint ringing. Mitchell–Netravali is a useful alternative when you prefer a smoother result; Catmull–Rom can look a little sharper. Compare the actual output at its intended display size.
Which filter should I use for pixel art?
Use Nearest Neighbor. It selects existing source pixels instead of blending neighboring colors, keeping deliberately hard pixel boundaries intact. For the cleanest enlargement, use an integer enlargement where available, such as 200%. Bilinear, bicubic, and Lanczos filters blend neighboring samples and can soften pixel-art edges.
What is the difference between Catmull–Rom and Mitchell–Netravali?
Both are cubic reconstruction filters, but they make different trade-offs. Catmull–Rom is an interpolating cubic that tends to emphasize edge sharpness. Mitchell–Netravali uses a balanced cubic setting (B = 1/3, C = 1/3) intended to balance sharpness, smoothness, and artifacts. Choose Catmull–Rom for a crisper look and Mitchell when a slightly gentler reconstruction is preferable.
What is ringing or a halo around an image edge?
Ringing is a faint light or dark echo that can appear beside a sharp transition, such as dark text on a white background. Filters with stronger edge emphasis can produce this effect around high-contrast boundaries. It is not necessarily a defect in the original file; it can be a side effect of reconstruction. Try Mitchell–Netravali or Bilinear when a Lanczos or Catmull–Rom result looks too pronounced.
Does enlarging an image create real detail?
No. Upscaling creates additional output pixels by estimating values from the pixels already present. It can make an image fit a larger space, but it cannot recover facial detail, text, or texture that was never captured or has already been discarded. For the best result, start with the highest-resolution original available and avoid enlarging it more than necessary.
When should I choose WebP?
Choose WebP for many website images when you want a compact file and need to retain transparency. It is a strong default for product photos, article imagery, thumbnails, and general web graphics. In Compressy, the quality control is used for lossy WebP output; use PNG when you need lossless storage of the exported pixels or especially clean edges in text and diagrams.
When is JPEG still the right choice?
JPEG is a practical choice for photographs when broad compatibility with older software, upload forms, or publishing systems matters. It does not support an alpha transparency channel, so transparent areas are flattened against a light background in the JPEG export. JPEG is lossy, which means repeated editing and re-encoding can gradually introduce visible artifacts.
When should I keep an image as PNG?
Use PNG for screenshots, interface captures, diagrams, charts, logos, line art, and images that need transparency with lossless encoding. PNG preserves the pixel values supplied to its encoder, but converting an already-compressed JPEG to PNG cannot restore detail that the JPEG discarded. PNG files can also be larger than lossy photo formats.
Does 86% quality mean the file will be 14% smaller?
No. The quality number is an encoder setting, not a promised percentage of file-size reduction or a universal visual-quality score. The resulting size depends on the image content, dimensions, format, and browser encoder. A detailed photograph, a flat illustration, and a screenshot can produce very different file sizes at the same setting.
Does 100% quality make JPEG or WebP lossless?
No. A 100% setting asks the lossy encoder for its highest quality setting; it does not turn JPEG or Compressy's quality-controlled WebP export into a lossless archive. For lossless output of the resized pixel data, choose PNG. Keep the original file as well when exact source preservation matters.
Will PNG use the quality setting?
No. PNG is encoded losslessly, so the quality selector is disabled when PNG is selected. Resizing can still change the pixel data because new dimensions require resampling, but PNG will store the resulting raster without adding lossy compression.
Does Compressy preserve transparent backgrounds?
WebP and PNG can preserve transparency. JPEG cannot store transparent pixels; Compressy flattens transparent areas onto a light background when producing JPEG. For logos, cut-outs, product images with transparent backgrounds, and interface assets, choose PNG or WebP instead of JPEG.
What happens when I select 100% resize?
The output keeps the source pixel dimensions, so no resampling is needed. The image is still exported in the selected output format, which means it may be re-encoded and its file size can change. Choose 100% when you want to convert or recompress without changing width and height.
How much smaller will an image become when I resize it to 50%?
At 50% scale, both width and height are halved. The output therefore contains about one quarter as many pixels, a 75% reduction in pixel count. File size will not necessarily fall by exactly 75%, because compression also depends on visual complexity, format, and quality.
What dimensions should I use for a website image?
Start from the largest size at which the image will actually be displayed, and account for high-density screens when needed. For example, an image shown at 600 CSS pixels may need around 1200 source pixels wide for a 2× display. Avoid uploading enormous originals when the layout only displays a small thumbnail, but do not upscale a small source just to reach an arbitrary number.
Can I process multiple images with the same settings?
Yes. Add up to 25 images, choose the resize percentage, resampling filter, output format, and quality once, then process the batch. Images are processed sequentially. Each source image and its output must stay within the 40-megapixel safety limit, and practical capacity also depends on the browser and available device memory.
Which image types can I convert?
The file picker accepts JPEG, PNG, WebP, AVIF, and BMP images, subject to what the browser can decode. The available export formats are WebP, JPEG, and PNG. Compressy is designed for still raster images rather than vector artwork, PDFs, or animated-image editing.
Are my images uploaded or stored?
No image contents are sent to a file-processing server. Decoding, resizing, and encoding happen locally in your browser. The page itself still downloads normal website assets, such as scripts, styles, and fonts. Compressy does not provide cloud storage for your files.
Why can the converted image be larger than the original?
A larger output is possible when you upscale, choose a high quality setting, convert a simple image to a format that represents it less efficiently, or re-encode an already compressed source. PNG can be especially large for photographic content. Compare the original and output totals, and try a smaller scale or a lossy WebP/JPEG setting when file size is the priority.
Does changing the format improve image quality?
Changing the format does not recreate missing detail. A lossy format can reduce file size by discarding information, while PNG stores the exported pixel data losslessly. The best choice depends on the image: photos usually tolerate lossy compression better than screenshots, diagrams, or text-heavy graphics.
Does resizing crop or stretch the image?
No. Compressy applies the selected scale to both dimensions, keeping the image's proportions instead of cropping it or forcing it into a new aspect ratio. The final width and height are rounded to whole pixels. Use a dedicated crop tool when you need to change the framing or aspect ratio.
Are DPI and pixel dimensions the same thing?
No. Pixel dimensions describe how many pixels an image contains; PPI or DPI describes a print-density relationship between those pixels and a physical size. A 2400-pixel-wide image printed at 300 pixels per inch is about 8 inches wide. Compressy is focused on raster pixel dimensions, so check the print provider's required dimensions and density for print work.