You're exporting a photo from Photoshop, Lightroom, or your phone's camera roll and you hit the quality slider. Do you go 100% to be safe? 80% to save space? What does "quality 95 vs 100" even look like in practice? The numbers don't mean what most people assume, and picking the wrong one either bloats your file for no visual benefit or introduces ugly blocky artifacts you can't undo later.
Unlike PNG, JPEG actually has a real quality dial — but that dial doesn't behave the way most people expect. The relationship between the number you pick and what you actually get, in both file size and visual fidelity, is wildly non-linear. Below, you'll find the full 0–100 breakdown, a live tool to test it on your own photo, camera-specific settings for Canon and Nikon shooters, and how JPEG stacks up against JPEG 2000 and WebP. Once you understand how the scale really works, choosing the right setting takes seconds.
For most web images, a JPEG quality between 75% and 85% gives the best balance of visual fidelity and file size. Product or hero images can go up to 90–92%, thumbnails can drop to 60–70%, and print work should stay at 90–100%. Going above 90% rarely improves visible quality but can double your file size, since the JPEG scale is non-linear near the top — 90% and 100% are usually indistinguishable to the human eye, even though the file sizes are very different.
What does the JPEG quality setting actually do?
JPEG is a lossy format, meaning the compression process permanently discards some image data based on what the human eye is least likely to notice. The "quality" number — usually shown as 0–100, or as a 0–12 scale in Photoshop — controls how aggressively that discarding happens. It is not a measure of resolution, sharpness, or color accuracy on its own; it's a dial on how much detail gets thrown away during three specific stages of compression:
- Color space conversion. The image is converted from RGB to YCbCr, separating brightness (luma) from color (chroma), because the eye is far more sensitive to brightness differences than color differences.
- Chroma subsampling. Color resolution is reduced more than brightness resolution, since you can throw away a lot of color detail before anyone notices. This happens even at 100% quality unless subsampling is explicitly disabled — one reason "100%" isn't the same as "lossless."
- Quantization. The image is broken into 8×8 pixel blocks, transformed mathematically (a Discrete Cosine Transform), and the precision of that transformation is reduced — this is the actual step your quality slider controls, and where almost all visible artifacts, including the blocky look of over-compressed JPEGs, originate.
The critical thing to understand: this scale is not linear. Moving from 100% to 90% quality can cut file size by 50–70% with no visible difference at normal viewing distance, because the top of the scale spends enormous file size preserving detail the eye barely registers. Moving from 50% to 40% might only save a small amount further, while introducing very visible blocking. The sweet spot is almost never at the extremes.
The JPEG quality scale, 0 to 100, explained
"JPEG quality 0 to 100" is the most common way the setting is expressed, but the number itself isn't standardized by the JPEG specification — it's an encoder convention that different programs interpret slightly differently. Here's roughly what each range does, based on how mainstream encoders like libjpeg and mozjpeg behave:
| Quality range | What it looks like | Relative file size | When to use it |
|---|---|---|---|
| 90–100 | Visually near-identical to the source; artifacts essentially invisible even zoomed in | Largest | Print masters, archival copies, professional photography delivery |
| 80–89 | No visible loss at normal viewing distance; very minor softness under 200% zoom | Medium-large | Hero images, product photography, portfolio sites |
| 60–79 | Minor softness in gradients and skin tones may appear; still generally clean | Medium | General web content, blog images, email attachments |
| 40–59 | Visible blocking starts appearing around edges and in flat color areas | Small | Thumbnails, low-priority background images |
| 0–39 | Heavy blocking, color banding, and "mosquito noise" around edges — rarely acceptable | Smallest | Placeholder images only, or extreme bandwidth constraints |
Two numbers people search for specifically deserve a direct answer. Quality 90 vs 100: at normal screen viewing distance these are functionally identical, but 90% is typically 30–50% smaller in file size — there's almost never a reason to choose 100 over 90 for anything other than an archival master. Quality 95 vs 100: the gap is even smaller; most viewers cannot reliably tell them apart in a blind test, though 95% still saves a meaningful chunk of file size. If you're unsure, 95% is a safe "I don't want to think about it" choice for anything quality-sensitive.
Live demo: see JPEG compression at any quality level
Reading about compression only gets you so far — the fastest way to understand the 0–100 scale is to see it applied to a real photo, in real time. Drag the slider below to re-compress the sample image (or upload your own) at any quality level from 1% to 100%, and watch the file size and visual result change together. Everything happens locally in your browser; nothing is uploaded anywhere.
Why the right setting matters
Picking a JPEG quality setting isn't a cosmetic choice — it has direct, measurable consequences across performance, cost, and how your images actually look once they leave your editing software:
- Page speed. Photos are usually the heaviest assets on a web page. Exporting at 100% instead of 80% can double or triple page weight for zero perceptible gain, directly hurting Core Web Vitals like LCP.
- Storage and bandwidth costs. Photography sites, e-commerce catalogs, and CDNs pay per gigabyte. Multiplying every product photo's size by 2–3x for no visual benefit adds up fast at scale.
- Print quality. The opposite risk applies here — using a web-appropriate 75% quality setting for a large print job will reveal compression artifacts that were invisible on screen, especially in shadows and smooth gradients.
- Repeated editing. Every time a JPEG is opened, edited, and re-saved as a JPEG, it's compressed again on top of already-lossy data. Choosing the right quality and workflow up front avoids compounding this generation loss.
- Search visibility. Google's Core Web Vitals and image indexing both reward correctly sized, correctly compressed images — bloated JPEGs slow crawling and can hurt Largest Contentful Paint scores site-wide.
Step-by-step: choosing the right quality setting
- Identify where the image will be used. A thumbnail, a hero banner, a product photo, and a print file all have different tolerances for compression. Decide this before you touch a slider — it determines your entire target range.
- Start at 80% as your baseline. For the vast majority of photographic web images, 80% is the well-tested middle ground between visible quality and file weight. Export at this setting first before adjusting up or down.
- Adjust based on image content, not just use case. Images with smooth gradients, skin tones, or large flat color areas (sky, walls) show compression artifacts sooner — push these toward 85–90%. Busy, high-detail textures (foliage, fabric patterns, crowds) hide compression artifacts well and can often drop to 70–75% unnoticed.
- Zoom in at 100–200% to check for artifacts. Compression damage shows up first around high-contrast edges — text, logos, sharp object boundaries. If you see blockiness or color bleeding (mosquito noise) around these areas, increase the quality value by 5–10 points and re-export.
- Use Save for Web / Export tools, not generic Save As. Dedicated export dialogs (Photoshop's "Export As", Lightroom's Export panel) give finer-grained control and often strip unnecessary metadata automatically, unlike a basic overwrite save.
- Enable progressive encoding for web images. A progressive JPEG renders at low resolution first and sharpens as it downloads, improving perceived load speed on slower connections — most modern browsers support this with no downside.
- Always export from the original, never from a previously saved JPEG. If you need to re-edit later, go back to the source file (RAW, PSD, or an unedited copy) rather than re-opening and re-saving the compressed JPEG, to avoid compounding generation loss.
Camera JPEG settings: Fine, Normal, and Extra Fine
Most cameras — Canon, Nikon, Sony, Fujifilm, and smartphones alike — hide the 0–100 scale behind named presets rather than a numeric slider. Canon in particular labels its JPEG compression levels Extra Fine, Fine, and Normal (sometimes shown as Standard), and this is one of the most frequently searched camera-settings questions there is. Here's roughly how they map to the quality percentages covered above:
| Camera setting | Approx. quality | Typical file size | Best for |
|---|---|---|---|
| Extra Fine | ~95–100% | 10–15 MB+ | Professional delivery, large prints, heavy cropping |
| Fine | ~85–95% | 7–12 MB | Most everyday shooting where quality matters — the safest default |
| Normal / Standard | ~75–85% | 4–7 MB | Casual shooting, web-only use, burst/sports shooting to save card space |
Exact compression ratios vary by camera model and manufacturer, since "Fine" and "Normal" are implemented in each camera's own JPEG encoder rather than tied to a universal standard. A practical rule that applies across brands: if you plan to print, crop tightly, or edit heavily, use Fine or Extra Fine (or better yet, shoot RAW). If you're shooting casually for social media or a quick web upload, Normal is genuinely fine — the difference is very hard to see outside of pixel-peeping at 200% zoom. If your camera also offers RAW or RAW+JPEG, shooting RAW alongside a Normal-quality JPEG gives you a lightweight preview file plus full editing latitude if you need it later.
Common mistakes that waste quality or space
1. Always exporting at 100%
100% quality feels like the "safe" choice, but it's almost always wasteful. The algorithm spends a disproportionate amount of file size preserving detail that's invisible at normal viewing distance. Unless you're archiving a master file or sending to a professional print house, 100% is rarely the right call.
2. Assuming quality percentages mean the same thing across tools
Photoshop's 0–12 scale, Lightroom's 0–100% scale, and various online compressors don't map to each other 1:1. A "quality 80" export from one tool can look noticeably different from "quality 80" in another. Always judge by the visual result and file size in the tool you're actually using, not by assuming a universal standard.
3. Re-saving the same JPEG repeatedly
Opening, editing, and re-saving a JPEG over and over compounds compression artifacts each time — this is generation loss, and it's permanent. If your workflow involves multiple editing passes, work in a lossless format (PSD, TIFF, PNG) and only export to JPEG once, as your final step.
4. Using a single quality setting for every image on a page
A hero banner and a footer thumbnail don't need the same quality setting. Applying one blanket export quality across an entire site either over-compresses your most important visuals or wastes bandwidth on images nobody examines closely. Set quality per role, not per project.
5. Confusing quality with resolution
Lowering the quality percentage does not change an image's pixel dimensions or DPI — a 3% quality JPEG at 6000×4000 pixels is still a 6000×4000 pixel image, just a badly compressed one. If a photo genuinely looks blurry or low-res rather than blocky, the fix is resizing or re-sourcing a higher-resolution original, not raising the quality slider. See the DPI section below for the distinction.
Real-world quality examples
These are representative results from exporting the same source images at different JPEG quality settings, using a standard Save for Web workflow:
The pattern holds across most photography: smooth, low-detail areas like sky, skin, and flat backgrounds reveal compression artifacts first, while busy, high-frequency detail can absorb much more aggressive compression before anyone notices. This is also why two images at the identical quality percentage can produce very different file sizes — quality settings control compression intensity, not a fixed output size.
Quality setting by use case
Rather than one universal number, the right setting depends entirely on where the image is going and what it's showing.
| Use case | Recommended quality | Typical savings vs. 100% | Artifact risk | Notes |
|---|---|---|---|---|
| Thumbnails / icons | 60–70% | 70–85% | Low | Small display size hides most compression artifacts |
| General web photos | 75–85% | 50–70% | Low | The universal sweet spot for blog posts and content images |
| Hero / banner images | 85–90% | 35–55% | Low-medium | Large display size makes artifacts more noticeable |
| E-commerce product photos | 85–92% | 30–50% | Low-medium | Customers zoom in — detail and color accuracy matter |
| Email attachments | 70–80% | 55–70% | Low | Keeps attachments under client size limits |
| Social media uploads | 80–90% | 30–55% | Low | Most platforms re-compress on upload anyway — start high to survive it |
| Print / professional photography | 90–100% | 0–25% | N/A | Print reveals artifacts screens hide — prioritize fidelity |
| Archival / master files | 100% (or lossless PNG/TIFF) | 0% | N/A | Keep one untouched master; generate everything else from it |
JPEG vs. JPEG 2000 vs. WebP
JPEG 2000 (file extension .jp2) is a completely different, newer codec that happens to share JPEG's name. Instead of the block-based DCT compression standard JPEG uses, it uses wavelet-based compression, which degrades more gracefully under heavy compression — soft blurring instead of hard blocky edges — and it can do true lossless compression in the same file format, something standard JPEG cannot do at all.
In direct quality-per-file-size comparisons, JPEG 2000 generally wins, particularly at high compression ratios. It also supports 16-bit color depth, alpha transparency, and region-of-interest encoding, where a specific area of the image can be preserved at higher quality than the rest — useful for things like satellite imagery or medical scans. Despite those technical advantages, JPEG 2000 has never achieved mainstream web adoption: browser support is essentially limited to Safari, so it's mostly confined to specialist fields like digital cinema (DCI), medical imaging, and library/archive scanning rather than everyday web or camera use.
| Format | Compression method | Lossless mode | Browser support | Best for |
|---|---|---|---|---|
| JPEG | DCT, block-based | No (rarely used) | Universal | General photography, maximum compatibility |
| JPEG 2000 | Wavelet-based | Yes | Safari only | Medical imaging, digital cinema, archival scans |
| WebP | Predictive + VP8 | Yes (separate mode) | All modern browsers | Everyday web delivery — usually 25–30% smaller than JPEG at equal quality |
For nearly everyone reading this — website owners, photographers exporting for clients, anyone emailing or uploading images — the practical choice is between JPEG and WebP, not JPEG 2000. Use WebP where you control the delivery pipeline and can confirm browser support (or let your CDN auto-negotiate format per browser); use JPEG when compatibility has to be guaranteed, such as email attachments or files a client will open in unknown software. JPEG 2000 is worth knowing about, but it's rarely the right pick outside its specialist niches.
Quality vs. resolution: DPI and pixel dimensions
A surprisingly common point of confusion: JPEG "quality" and image "resolution" are controlled by two completely separate settings, and fixing one won't fix a problem caused by the other. Quality controls compression (how much detail is thrown away during encoding). Resolution — measured in pixel dimensions (e.g. 3000×2000px) and DPI (dots per inch) — controls how much detail exists in the first place and how large it can be printed or displayed sharply.
- For web use, DPI is largely irrelevant — browsers render based on pixel dimensions, not DPI metadata. What matters is matching pixel dimensions to the largest size the image will actually display at (often 1–2x the container width for retina screens).
- For print use, the standard target is 300 DPI at final print size — a photo meant to print at 8×10 inches needs roughly 2400×3000 pixels to hit that target, regardless of JPEG quality setting.
- A low-quality-but-high-resolution JPEG will look blocky but sharp in outline; a high-quality-but-low-resolution JPEG will look smooth but blurry or pixelated when enlarged. They're different problems with different fixes.
If you're prepping images for both web and print from the same source, export two versions from your original master rather than trying to find one setting that serves both: a resized, 75–85% quality version for the web, and a full-resolution, 90–100% quality version for print.
Compress your JPEG right now — free
The Rebrixe JPEG Compressor runs entirely in your browser. Your images are never uploaded to a server — compression happens locally, and you can preview the exact visual result and file size at any quality setting before downloading. No account, no file size limit, no watermarks.