Smaller images.Ready for what’s next.

Add an image, choose Target mode, set the maximum file size to 50 KB, and choose a format your destination accepts. Compress, then check the measured output bytes and the target status before downloading. Allowing smaller dimensions can help reach the ceiling, but always check the portal’s dimension rules too.

100 KB is a maximum of 100,000 bytes here, not a request to pad the file to exactly that length. A result is marked Target met only when its actual encoded byte count is at or below the ceiling. Some images cannot reach the limit with your chosen format, quality floor and dimensions.

Image encoders do not produce every possible file size, so the search keeps a measured candidate that fits. Adding padding would make the file larger without improving its image detail. Use the exact byte count in the result rather than a rounded KB label when checking an upload limit.

JPG, WebP and AVIF use lossy encoding here, so compression can change visible detail. The quality value is a codec setting, not a percentage of retained visual quality, and equal settings across formats are not equivalent. PNG optimization preserves the working pixels unless you enable lossy palette reduction; cropping, resizing and conversion to 8-bit sRGB can still change the source.

Inputs include JPEG/JPG, PNG, WebP, AVIF and HEIC/HEIF, plus GIF and BMP when the browser can decode them. Animated files use only the first/default still frame, and HEIF collections use the primary image. Outputs are JPEG via MozJPEG, PNG via OxiPNG, WebP and AVIF via local WebAssembly codecs; a verified browser encoder may be used if a codec fails, with a notice.

JPG is widely accepted for photos but cannot keep transparency. PNG preserves working pixels without lossy encoding when palette reduction is off, though it can be larger for photos. WebP and AVIF support transparency and can make efficient web images, but AVIF takes more processing and some upload portals reject newer formats. Choose the format your destination accepts and compare the actual output.

Yes: browse, drop or paste multiple images, and add more without replacing the queue. The current safety limits are 50 files and 200 MB of source files per batch, with a separate 200 MB budget for retained previews and outputs. Shared settings apply to the batch, while crops are set per image; completed outputs can be downloaded individually or together as a ZIP.

Keep Lock aspect ratio enabled when using width and height: they become a bounding box rather than a forced shape. Percentage and longest-side resizing also preserve the image’s proportions. Use the crop control if you need a different aspect ratio, and leave upscaling off unless you deliberately want larger dimensions.

No image processing upload is made: file bytes are decoded, resized and encoded in local browser workers. This tool route does not load ads, analytics, chat or translation scripts; it requests the page and its static application, worker and codec assets. If you explicitly use Share, the selected result is handed to your device’s share service and the app you choose.

The workspace adapts to phones and tablets in current browsers with Web Workers, WebAssembly and the required image APIs. Before going offline, use the offline controls near the tool to download the required assets and check their reported status; a first visit alone is not a guarantee. Browser storage can be cleared or evicted, and private browsing or unsupported service workers can prevent offline use. Large images and AVIF encoding can still be demanding on a phone.

Yes: PNG, WebP and AVIF outputs can preserve transparency. PNG has optimization effort controls, not a quality slider; optional palette reduction is explicitly lossy. JPG needs you to approve flattening transparency onto a chosen background color, and Auto excludes JPG for transparent images unless you allow that conversion.

Yes: HEIC/HEIF inspection and a local decoder are loaded on demand, with native decoding used where available. Add the photo, choose JPG and inspect the output before downloading. Unsupported HEIF variants, corrupt files or resource limits can still cause a clear failure; the tool does not preserve extra collection images, HDR or the original wide-gamut profile.

KB uses 1,000 bytes and MB uses 1,000,000 bytes. You can select binary KiB, which uses 1,024 bytes, or MiB, which uses 1,048,576 bytes. Target success is checked against the measured output bytes, so use the convention specified by your destination.

A small ceiling, preserved dimensions, detailed content, a high minimum quality or lossless PNG can prevent a passing result. The tool shows the closest observed attempt as a target miss, not as proof that no smaller result is possible. Allow more resizing, reduce the minimum quality for a lossy format, or try WebP if your destination accepts it, then explicitly compress again.

Read the per-image error, then retry with a smaller source, fewer queued images or another output format. Corrupt files, unsupported image variants, codec-loading failures and memory or time limits can stop a job without producing a download. If a codec could not load, reconnect before retrying; use a trusted editor to resize an input that exceeds the decode limit. Your original file is not modified.

ImageFixer by ByteVerse is free to use, with no signup, watermark or paid feature locks. File, pixel and queue limits protect browser resources rather than enforce a paid quota. There is no guaranteed completion time or promised percentage reduction: results depend on the image, settings and device.

Source metadata is always discarded during re-encoding, including EXIF, GPS and camera metadata; there is no keep-metadata switch. A local scan can flag location or camera-serial markers, but an incomplete scan or no warning does not prove the source had none. Working pixels are 8-bit sRGB, and original ICC, wide-gamut and HDR information is not retained. Metadata removal does not hide personal information visibly present in the image.

Auto measures eligible JPG, WebP and AVIF candidates. Balanced prioritizes passing candidates with larger pixel area, then lower sampled decoded pixel error, then fewer bytes; Smallest tests the selected minimum-quality setting and prefers fewer measured bytes. The sampled error is not a perceptual quality score, and the search is not an exhaustive guarantee of the best possible image. Compare the result yourself before downloading.

Each input must be non-empty and no larger than 50 MB, with a normal ceiling of 50 megapixels and 16,384 pixels on either side. Devices reporting very low memory use a lower 12-megapixel ceiling; worker concurrency also scales down with available cores, memory and aggregate pixels. A job has a two-minute safety timeout, and the active limits appear in the workspace. These checks can reject a file before a resize is possible, so reduce oversized sources in a trusted editor first.

Individual downloads and ZIP export are available for completed results. Save to folder appears only where the File System Access API is supported, commonly in desktop Chromium browsers, and asks you to choose a destination. Share appears only when the device can share that output file; handing it to another app does not confirm delivery. Normal downloads remain the alternative when these browser features are unavailable.

Images and results are held in memory for the current queue, not saved to localStorage or IndexedDB by the tool. Custom presets you explicitly save, the theme preference and cached offline application assets can persist; the workspace does not automatically restore a previous image session. Settings links contain validated settings, not image data, source filenames or per-image crops. Downloads and files you explicitly save are separate and remain wherever your browser or chosen folder stores them.

Add the PNG, choose JPG under Convert to, and select Compress images. JPG cannot keep transparent pixels, so if the picture has transparency the tool asks for your permission to flatten it onto a background colour you choose. Use Quality mode if you only want to change the format without a file-size limit, then check the result before downloading.

Add your image, choose WebP under Convert to, then compress. WebP keeps transparency and is often smaller than a comparable JPG or PNG, but results depend on the image and settings, so compare the exact byte counts. Check that the website or upload form you are using accepts WebP before replacing an existing file.

Output formats are JPG, PNG, WebP and AVIF, or Auto, which measures JPG, WebP and AVIF and keeps the best result under your settings. HEIC, HEIF, GIF and BMP can be added as inputs when your browser can decode them, but they are not offered as output formats. The tool does not create PDF, SVG or animated GIF files.

Choose Width & height under Exact size & resize method, enter the pixels, and compress. With Lock aspect ratio on, the size works as a bounding box so the image is never stretched; turn it off only if you deliberately want to force both values, which can distort the picture. Use Crop first when you need a different shape, such as a square 1080 × 1080 image.

Resizing changes the number of pixels, for example from 4,000 to 1,200 pixels wide. Compressing changes how those pixels are encoded to make the file smaller. Using both usually gives the smallest file: resize to the dimensions you actually display, then compress. Resizing smaller removes detail permanently from the downloaded copy, but never from your original.

Often, but not always. A PNG photo converted to JPG or WebP is commonly much smaller, while a small graphic converted to a different format can end up larger. The result panel shows the original and output bytes side by side and warns when an output is larger, so you can try another format or lower quality before downloading.