TIFF (Tagged Image File Format) is a raster image container that stores pixel data with minimal processing, and it's commonly used for print production, professional photography masters, archival storage, scanning, and OCR input because it preserves image quality through lossless or uncompressed encoding. The format supports layers, alpha channels, and Adobe's Image Source Data tag, which means a layered Photoshop document can be saved with full editing structure intact. Many photographers and designers prefer it because a TIFF round-trip doesn't degrade the file, while JPEG re-saves do.
What the TIFF Format Actually Stores Beyond Pixels
A TIFF is a container built from a header, one or more Image File Directories, and tagged fields that describe how the image data should be interpreted. The header is 8 bytes: a byte-order mark (II for little-endian, MM for big-endian), the constant 42, and an offset to the first directory. That small design choice is why a TIFF written on a Windows machine can be read identically on a Mac or a Linux server without conversion.
Each Image File Directory entry is 12 bytes: a 2-byte tag identifier, a 2-byte data type, a 4-byte count, and a 4-byte value or offset. Tags carry the information that makes TIFF self-describing β image width (tag 256), compression scheme (tag 259), photometric interpretation (tag 262), strip offsets (tag 273), and resolution (tags 282 and 283). A reader that finds an unfamiliar tag can skip it rather than failing, which is why the format has survived four decades of extension without breaking compatibility.
Beyond pixel data, TIFF files can hold private tags. Adobe's Image Source Data tag (37724) is the one that matters most to designers: it stores the full layered Photoshop document inside the TIFF. Open that file in Photoshop and your layers, masks, adjustment layers, and smart objects are still there. Adobe documents this behavior in its Photoshop User Guide under "Saving images in TIFF format," and it's the single feature most responsible for TIFF's persistence in design workflows.
The format also supports 8-bit and 16-bit integer channels, 32-bit floating-point samples, CMYK and Lab color modes, and multiple alpha channels. That range is why a scanned transparency, a raw-converted master, and a press-ready CMYK file can all live in the same extension. According to the Library of Congress's Sustainability of Digital Formats pages, TIFF 6.0 remains the recommended master format for still-image digitization because it's uncompressed or losslessly compressed and thoroughly documented.
TIFF vs JPEG vs PNG vs PSD: Where Each Format Owns the Job
Pick the format by what happens to the file after it leaves your hands. TIFF wins when the file must survive repeated editing without quality loss; JPEG wins when file size or universal compatibility matters more than fidelity; PNG wins for web graphics with transparency; PSD wins when you need Photoshop-native smart objects and layer comps with full application support.
| Dimension | TIFF | JPEG | PNG | PSD |
|---|---|---|---|---|
| Compression | None, LZW, ZIP, PackBits (all lossless); JPEG compression optional in TIFF 6.0 | Lossy DCT, quality 0β100 | Lossless DEFLATE | None or RLE (Photoshop proprietary compression) |
| Bit depth | 8, 16, 32-bit integer; 32-bit float | 8-bit only | 8 or 16-bit | 8, 16, 32-bit |
| Layers preserved | Yes, via Adobe Image Source Data tag | No | No | Yes, native |
| Transparency / alpha | Yes, multiple alpha channels | No | Yes, single alpha | Yes |
| CMYK support | Yes | Yes | No (RGB only) | Yes |
| Typical file size, 24 MP image | 70β140 MB uncompressed; 35β70 MB LZW | 8β15 MB at quality 90 | 25β45 MB | 80β300 MB with layers |
| Max dimensions | ~4 GB per file in classic TIFF; BigTIFF removes the limit | 65,535 Γ 65,535 px | 2,147,483,647 px per side | 300,000 Γ 300,000 px |
The size numbers above are arithmetic, not measurements: a 24-megapixel image at 8 bits per channel in RGB is 24,000,000 Γ 3 bytes = 72 MB raw, and LZW typically halves that on photographic content. The 4 GB classic TIFF ceiling is documented in the TIFF 6.0 specification from Adobe; files that exceed it need BigTIFF, which most current software supports but some print workflows still don't.
One row most comparisons skip: PSD's maximum document size is 300,000 pixels per side, while classic TIFF's limit is a byte ceiling rather than a pixel ceiling. For gigapixel stitched scans, that distinction decides the format before any quality argument starts.
Why Lossless Encoding Matters Across a Multi-Step Editing Session
Round-trip damage is the practical reason photographers reach for TIFF. Open a JPEG, adjust curves, save, close, and reopen β that's one generation of lossy compression on top of the original. Do it ten times across a retouching cycle and banding appears in smooth gradients, halos form at high-contrast edges, and 8Γ8 block artifacts become visible in flat skies. TIFF with LZW or ZIP compression stores every pixel exactly, so the tenth save matches the first save bit for bit.
Here's a worked example. A wedding photographer shoots 2,000 frames at 24 MP and delivers JPEGs. If each frame averages three save cycles during culling, retouching, and export, that's 6,000 lossy encodes across the gallery. At quality 90, each generation discards roughly 90% of the DCT coefficients' precision β cumulatively, the delivered file is several generations removed from the sensor data. Substitute TIFF in the retouching stage and the delivery JPEG is a first-generation encode from a pristine master. That's a computation from stated inputs, not a measured study.
Lossless doesn't mean free. LZW on photographic content typically compresses to 50β60% of raw size, ZIP to 40β60%, and neither touches the file the way JPEG's 10:1 ratio does. A 128 GB memory card that holds 4,000 JPEGs holds roughly 900 uncompressed 24 MP TIFFs. Photographers who shoot TIFF in-camera pay that storage cost on every frame.
There's a failure mode worth naming. Some cameras write TIFF with a proprietary tag layout that third-party software reads incorrectly β colors shift, or the file opens at the wrong bit depth. If a client sends a TIFF that looks washed out in your editor, check whether the embedded ICC profile survived the transfer; a TIFF stripped of its profile tag renders with the application's default assumption, which is usually sRGB even when the file is Adobe RGB.
Print Production, CMYK, and the Files Prepress Actually Accepts
Commercial printing is where TIFF's reputation was built. Prepress workflows want CMYK with a known dot gain, embedded output intent, and no surprises at the RIP. TIFF delivers all three. A CMYK TIFF with an embedded ICC profile carries the color conversion instructions directly, and a properly tagged file moves through a Fiery RIP or an Agfa Apogee workflow without the re-interpretation that PDF sometimes triggers.
Resolution expectations are concrete. Offset printing at 175 lines per inch wants image data at 350 pixels per inch at final size β the standard 2Γ rule. A 4Γ6 inch image at 350 ppi is 1,400 Γ 2,100 pixels, or 8.8 MB uncompressed in RGB. Scale that to a 24 Γ 36 inch poster and you need 8,400 Γ 12,600 pixels, which is 317 MB uncompressed RGB or roughly 158 MB in LZW. Those numbers decide whether you ship a TIFF or flatten to PDF.
Large-format printing relaxes the rule. Billboard and banner work at 1,200 mm viewing distance accepts 100β150 ppi, so a 2 Γ 3 meter banner needs far less data than the 350 ppi rule implies. Ask the print provider for their required ppi at final size before exporting; the answer varies by process and substrate more than any general guide admits.
Watch the byte ceiling on big jobs. Classic TIFF caps at 4 GB, and a 16-bit CMYK file at 300 ppi for a 24 Γ 36 inch poster exceeds it: 24 Γ 36 Γ 300Β² Γ 4 channels Γ 2 bytes β 6.2 GB. That's when BigTIFF becomes mandatory, and it's also when you find out whether your RIP supports it. Confirm with the prepress house before building the file.
Archival Masters, Scanning Standards, and Cultural Heritage Digitization
The Library of Congress lists TIFF as the preferred format for digitized still images in its recommended format specifications for textual and image content. The reasoning is documented and structural: the format is openly specified, widely implemented, and stores data without lossy transformation. Federal digitization guidelines such as the Federal Agencies Digital Guidelines Initiative (FADGI) Technical Guidelines for Digitizing Cultural Heritage Materials specify TIFF for preservation masters, with uncompressed or lossless-compressed encoding.
The FADGI guidelines set concrete targets. For a preservation master from a reflective original, the recommendation is typically 300β400 ppi at 24-bit RGB or 16-bit grayscale, stored as uncompressed or LZW-compressed TIFF. A 4 Γ 5 inch photographic print scanned at 400 ppi in 24-bit color yields 1,600 Γ 2,000 pixels, which is 9.6 MB raw β about 5 MB with LZW. Multiply across an archive of 50,000 prints and you're looking at 250β480 GB just for masters, before derivatives.
Archival practice usually keeps two copies: the TIFF master, untouched, and a derivative JPEG or PNG for access. The master is never edited. That discipline is why TIFF survives in memory institutions even as web delivery moved to WebP and AVIF. The format's extended metadata support β IPTC, XMP, and EXIF all fit in TIFF's tag structure β means a scanned image can carry provenance, rights, and technical metadata in the same file.
An edge case worth knowing: TIFF allows multiple images per file via successive Image File Directories. Scanners and fax systems sometimes use this for multi-page output, but most image editors only display the first page. If a multi-page TIFF opens as a single image in your app, that's why β the other directories are present but unrendered.
When In-Camera TIFF Beats Raw, and When It Costs You
Some cameras offer TIFF as an alternative to JPEG, and it's a real option for high-volume shooters who need a usable file straight from the card without a raw conversion step. The trade-off is fixed: you get baked-in white balance, picture style, and tone curve, but no highlight recovery and no shadow lifting beyond what 8 bits per channel allows. Raw keeps 12β14 bits of linear data; a TIFF straight out of camera is typically 8-bit, which means about 256 levels per channel instead of 4,096 or 16,384.
Storage math makes the choice concrete. A 45 MP full-frame body writing 14-bit raw at about 50 MB per frame, versus an 8-bit TIFF at roughly 135 MB uncompressed (45,000,000 Γ 3 bytes), means the TIFF costs almost three times the card space. Add a second card slot for backup and the cost doubles. For a wedding shooter burning 2,000 frames, that's 270 GB of TIFFs versus 100 GB of raws.
There's a workflow case where the TIFF wins: tethered capture into a retouching pipeline that immediately hands files to a designer who doesn't have raw support. A TIFF arrives color-correct and printable. A raw file requires conversion first. Studios that shoot product on set with a client reviewing live often standardize on TIFF for exactly this reason.
The failure case: TIFF from camera often arrives with an embedded JPEG preview that some asset management systems read as the primary image. If thumbnails look soft or wrong-colored in your DAM, check whether it's rendering the preview rather than the full-resolution data. Rebuild the previews before concluding the files are damaged.
Layers, Photoshop Interoperability, and the Workflow Trade-Off
Saving a layered Photoshop document as TIFF preserves the layer structure through the Image Source Data tag, and that's why design studios pass TIFFs instead of PSDs to clients and vendors who may not have Photoshop. The file opens in Affinity Photo, Capture One, and most professional editors with layers intact or at least flattened predictably, while a PSD depends on Photoshop-native parsing for full fidelity.
The trade-off is file size and speed. A layered 24 MP TIFF with ten adjustment layers, masks, and a smart object routinely reaches 300β500 MB, and saving takes noticeably longer than the equivalent PSD because the flattened composite must also be written. Designers working on magazine spreads with dozens of linked images will feel that on every save.
Compatibility has limits. Nested smart objects, layer comps, and some blend modes don't survive the TIFF round-trip cleanly in non-Adobe applications. If the receiving party uses Photoshop, PSD is the safer container for complex layer structures. If they use anything else, TIFF is the safer container for a flattened or lightly layered file. That split is the practical rule.
One naming discipline saves trouble: keep the layered TIFF and the flattened delivery file as separate documents. A client who opens your layered TIFF and edits a layer, then saves, can silently break your composite. Deliver a flattened copy for approval and keep the layered master internal.
Compression Choices, File-Size Budgets, and the 4 GB Wall
Choosing a compression scheme is a real decision with measurable outcomes. Uncompressed TIFF is the fastest to write and read but the largest. LZW is the most compatible lossless option across older software. ZIP (Deflate) usually compresses slightly better than LZW on photographs but is slower to encode. PackBits is fast and best on images with large flat areas, like scanned line art.
Here's a size budget for a concrete job: a 50 MP image at 16-bit RGB is 50,000,000 Γ 3 Γ 2 = 300 MB uncompressed. LZW typically brings that to 150β180 MB, ZIP to 140β170 MB. For a 500-image archive, uncompressed costs 150 GB, LZW costs 75β90 GB. At a typical cloud backup rate, that's a real monthly difference in storage billing.
The 4 GB limit is the wall most people hit without warning. A 100 MP 16-bit CMYK scan with alpha channels can exceed it in a single file. BigTIFF, specified by the Aware Systems extension and supported in current versions of Photoshop, ImageMagick, and libtiff, raises the limit by using 64-bit offsets. Confirm BigTIFF support with your output vendor before committing a large project to it.
An underused option: TIFF 6.0 permits JPEG compression inside the TIFF container. That gives you a TIFF extension with lossy data β useful when a system demands .tif but you need small files. Don't confuse it with a lossless master. The extension tells you nothing about the encoding, which is why checking the compression tag before archiving matters.
Common Misunderstandings That Cost People Time
The most persistent myth is that TIFF is always uncompressed. It isn't. TIFF 6.0 defines several compression schemes, including lossy JPEG, and a file's extension says nothing about which one was used. Verify by reading the compression tag (259) in a tool like ExifTool before treating a TIFF as an archival master.
A second misunderstanding: TIFF is a web format. Browsers largely don't render TIFF natively, and the ones that do via plug-ins are inconsistent. If you need an image on a web page, export to JPEG, PNG, WebP, or AVIF. Sending a TIFF to a client for web review guarantees a support call.
A third: bigger files mean better quality. A TIFF saved from an already-compressed JPEG carries the JPEG's artifacts at TIFF's file size. Converting format doesn't recover data that was already discarded. The quality ceiling is set at capture, not at export.
The last one bites designers regularly: multiple alpha channels in a TIFF don't travel predictably into page layout software. InDesign may import only the first alpha as a clipping path or ignore them entirely. Check the placed file rather than assuming the channels survived.
FAQ
Is TIFF better than RAW for photography?
No β raw holds 12β14 bits of linear sensor data per channel, while a camera TIFF typically holds 8 bits with processing baked in, so raw gives more highlight and shadow recovery. TIFF is better than JPEG for editing headroom. Choose raw when you'll grade the file, TIFF when you need a print-ready file immediately.
Can I open a TIFF file without Photoshop?
Yes. macOS Preview, Windows Photos (with the HEIF/RAW extension installed), GIMP, Affinity Photo, IrfanView, and Capture One all open standard TIFF files. Layered TIFFs may display flattened in some of these apps. Command-line tools like ImageMagick handle TIFF on servers where no GUI exists.
Why is my TIFF file so large compared to the JPEG version?
Uncompressed TIFF stores every pixel exactly: a 24 MP RGB image is 72 MB raw, while a quality-90 JPEG of the same image is usually 8β15 MB. That's roughly a 5β9Γ difference driven by JPEG's lossy DCT compression discarding detail the eye barely notices. Switching TIFF to LZW compression typically halves the TIFF size.
Does TIFF support transparency and layers?
Yes to both, with conditions. TIFF supports multiple alpha channels and, through Adobe's Image Source Data tag, full Photoshop layer structure. Non-Adobe applications may render layered TIFFs flattened. Transparency works reliably; layer editability depends on the receiving software.
What is the maximum file size for a TIFF?
Classic TIFF is capped at 4 GB per file by its 32-bit offset structure, which a 100 MP 16-bit CMYK file with alpha channels can exceed. BigTIFF uses 64-bit offsets and removes the practical limit, but confirm your RIP, DAM, or archive system supports BigTIFF before relying on it.
Is TIFF suitable for uploading to websites or email?
Poorly. Browser support is inconsistent, file sizes are large, and many email clients block or compress attachments over 25 MB. Export a JPEG or WebP for web delivery and keep the TIFF as the working master. A 72 MB TIFF won't survive most email gateways intact.
How does TIFF compare with PNG for graphic design?
PNG uses lossless DEFLATE compression and supports one alpha channel but no CMYK and no layers, so it suits web graphics and screenshots. TIFF supports CMYK, 16-bit depth, multiple alpha channels, and embedded layers, which is why print production standardizes on TIFF. For a logo on a website, PNG is the right pick; for a press-ready brochure, TIFF is.