RAW vs TIFF: What Converting a RAW File to TIFF Keeps and Loses, Tested on 17 Files

RAW vs TIFF: A RAW file holds the sensor's unprocessed data; a TIFF holds a finished, full-color image. In RevelRaw's November 2026 test of 17 camera files converted on a Mac, 16-bit TIFFs came out a median 4.8 times the RAW's size and 8-bit TIFFs 2.4 times. Keep the RAW for editing; export TIFF once a photo is finished.

Some RAW vs TIFF articles say a TIFF is about twice the size of a RAW, and none of the top-ranking pages we read measured more than one file. So we converted 17 public-domain files from 11 cameras and six brands to TIFF on a Mac, and measured size, compression, highlight recovery, shadow tones and metadata. One surprise: LZW compression made 16-bit TIFFs bigger than no compression in 15 of 16 files, while ZIP shrank them. Need to open the files first? See how to open RAW files on a Mac. Thinking of converting to DNG instead? Read our DNG vs RAW conversion test.

RAW vs TIFF compared: RAW, 8-bit TIFF and 16-bit TIFF

The Library of Congress, quoting the DNG specification, puts the difference in one line: "Unlike JPEG and TIFF formats which store images that have been processed by the camera, camera raw files capture unprocessed or minimally processed data directly from the camera sensor." A raw file stores one value per photosite (each light-sensitive point on the sensor), recorded through a color filter. Making a TIFF fills in the two missing colors at every pixel (demosaicing), applies the white balance, sets a white point above which everything clips, and applies a tone curve for display: a TIFF stores finished red, green and blue for every pixel. TIFF itself is an old, stable format: "The most recent version of TIFF is 6.0, published in 1992," the Library notes. Here is what the difference cost in RevelRaw's 17-file test.

FormatFile sizeHighlight headroomShadow bandingMetadataBest for
RAWThe original: 12.44 to 85.71 MB in the 17-file setKeeps sensor detail past the TIFF's white point: a −1 EV pull took 88.5% of clipped pixels off pure white on the most clipped file in the set87 levels after a +3 EV pushEverything the camera wrote, including 85 to 449 maker-note tags and its white-balance dataEditing, re-editing, your original
TIFF, 8-bit2.44× the RAW (1.35 to 4.41×)Fixed at conversion; clipped about as much as the 16-bit TIFF19.5 levels, about 4.5× fewer than the RAWNo maker notes or white-balance data; exposure basics keptA finished photo nobody will edit further
TIFF, 16-bit4.88× the RAW (2.69 to 8.83×); 4.83× from the Mac's own converterFixed at conversion: 0 of 1,334,832 clipped pixels came back on the most clipped file in the set84 levels, within 10% of the RAWSame as 8-bit; the Mac's TIFFs also kept lens and capture datePrint masters, handoff to a retoucher or client, an archive copy beside the RAW

Medians, with the smallest and largest ratio. Sizes: 16 files from a free, open-source RAW converter; the Mac's own converter wrote 16-bit only (17 files). Highlights: the most clipped file in the set, a Panasonic Lumix S5II frame. Shadows: distinct green tones (out of 256) left in the darkest tenth of the photo after brightening it three stops, 14 files; fewer levels means coarser steps.

Method in brief: 17 public-domain camera files from raw.pixls.us, converted on November 12, 2026 on macOS 27.0.1, with the Mac's built-in RAW support and with a free, open-source RAW converter. How we tested.

RAW vs TIFF file size: a 16-bit TIFF is about 5 times the RAW

In RevelRaw's November 2026 test of 17 camera files, every TIFF was bigger than its RAW unless it was 8-bit and compressed. The Mac's built-in converter wrote 16-bit TIFFs a median (the middle value) 4.83 times the size of the RAW, from 2.67 to 9.97 times, and none of the 17 came out smaller. The open-source converter's 16-bit TIFFs were a median 4.88 times the RAW (2.69 to 8.83 times, 16 files) and its 8-bit TIFFs 2.44 times (1.35 to 4.41). Altogether, 652.9 MB of RAW files became 2,986.6 MB (about 3 GB) of Mac TIFFs, 4.57 times the space.

16-bit TIFFs ran a median 4.0 to 5.1 times the size of their RAW; only 8-bit compressed TIFFs came close to it Median file size as a multiple of the original RAW, 16 camera files, November 12, 2026, LZW and ZIP with the horizontal predictor on: RAW 1.00; 8-bit TIFF with ZIP 1.17; 8-bit TIFF with LZW 1.24; 8-bit TIFF uncompressed 2.44; 16-bit TIFF with ZIP 4.01; 16-bit TIFF uncompressed 4.88; 16-bit TIFF with LZW 5.12. RAW (the original) 1.00× 8-bit, ZIP 1.17× 8-bit, LZW 1.24× 8-bit, uncompressed 2.44× 16-bit, ZIP 4.01× 16-bit, uncompressed 4.88× 16-bit, LZW 5.12× 0 2× 4× 6× Median size, times the original RAW RevelRaw · Nov 12, 2026 · n=16
Median TIFF size as a multiple of the original RAW: RAW 1.00, 8-bit ZIP 1.17, 8-bit LZW 1.24, 8-bit uncompressed 2.44, 16-bit ZIP 4.01, 16-bit uncompressed 4.88, 16-bit LZW 5.12. 16 camera files converted with a free, open-source RAW converter and recompressed losslessly, November 12, 2026. LZW and ZIP were run with the horizontal predictor on. The Mac's own 16-bit TIFFs, not shown, were a median 4.83 times (17 files).

Why a TIFF's size depends on the pixel count, not the RAW

An uncompressed TIFF is width × height × 3 colors × 1 or 2 bytes per value, plus a few kilobytes for the color profile and file information (at most 24.5 KB, never more than 0.0101% of the file). So a 24-megapixel photo makes a 72 MB TIFF at 8-bit and a 144 MB one at 16-bit, however the camera stored it. The Canon EOS R6 Mark II's 6000 × 4000 frame came out of the Mac at 144.01 MB from both its 29.40 MB RAW and its 16.47 MB C-RAW.

The ratio therefore depends on how the camera stored the RAW. In the Mac's TIFFs, uncompressed RAWs gave the smallest jump (Sony a7 IV 2.67 times, Fujifilm X-T5 2.79 times) and the most compact modes the biggest (Canon C-RAW 8.74 times, Fujifilm Compressed 8.52 times, Nikon High Efficiency 9.97 times). For what each RAW mode weighs to begin with, see our RAW file size study.

CameraFormat and modeRAWMac 16-bit TIFF
Canon EOS R6 Mark IICR3 RAW (lossless)29.40 MB144.01 MB (4.9×)
Canon EOS R6 Mark IICR3 C-RAW (lossy)16.47 MB144.01 MB (8.7×)
Canon EOS R5 Mark IICR3 RAW (lossless)54.44 MB268.59 MB (4.9×)
Canon EOS Rebel T7iCR2 RAW29.63 MB144.01 MB (4.9×)
Sony a7 IVARW Uncompressed73.46 MB196.46 MB (2.7×)
Sony a7 IVARW Lossless compressed (L)49.18 MB196.46 MB (4.0×)
Sony a7 IVARW Compressed (lossy)40.67 MB196.46 MB (4.8×)
Sony a7 VARW Lossless compressed44.20 MB196.46 MB (4.4×)
Nikon Z fcNEF 14-bit lossless27.82 MB124.02 MB (4.5×)
Nikon Z fNEF 14-bit lossless31.11 MB146.33 MB (4.7×)
Nikon Z50 IINEF lossless27.50 MB124.02 MB (4.5×)
Nikon Z50 IINEF High Efficiency12.44 MB124.02 MB (10.0×)
Fujifilm X-T5RAF Uncompressed85.71 MB238.91 MB (2.8×)
Fujifilm X-T5RAF Lossless compressed43.11 MB238.91 MB (5.5×)
Fujifilm X-T5RAF Compressed (lossy)28.04 MB238.91 MB (8.5×)
OM System OM-1 Mark IIORF 12-bit22.44 MB120.94 MB (5.4×)
Panasonic Lumix S5IIRW2 14-bit37.31 MB144.01 MB (3.9×)

MB is 1,000,000 bytes. Uncompressed 16-bit TIFFs from macOS 27.0.1's own converter, with each size as a multiple of the RAW. Each camera name links to the CC0 source file.

LZW or ZIP? LZW made 16-bit TIFFs bigger

In RevelRaw's November 2026 test, LZW made 16-bit TIFFs bigger than no compression in 15 of 16 files, while ZIP (also called Deflate) never did. Both are lossless: every LZW and ZIP file in the test, 128 of them (8 settings × 16 files), opened to exactly the same pixels as its uncompressed source. TIFF can also carry lossy JPEG compression, so check which kind an app means by "compressed".

On 16-bit TIFFs, LZW backfired. With the horizontal predictor switched on (an option that helps compression on photos), LZW files were larger than uncompressed in 15 of 16 (a median 1.07 times; the exception was the Sony a7 V, whose black border compresses easily); without it, in 16 of 16 (1.18 times). ZIP never made a file larger, 0 of 16, and with the predictor saved a median 16% (0.84 times). ZIP is slower: a median 3.85 seconds per 16-bit file against 1.42 seconds for LZW on our M5 Pro Mac.

On 8-bit TIFFs, both work if the predictor is on. LZW with the predictor shrank all 16 files, to a median 0.53 times uncompressed; without it, 6 of 16 grew, by up to 1.09 times.

Only 8-bit compressed TIFFs ever beat the RAW, and by a clear margin only from uncompressed RAWs. The Fujifilm X-T5's uncompressed RAF became an 8-bit TIFF 0.64 times its size with ZIP and 0.66 times with LZW; the Sony a7 IV's uncompressed ARW 0.82 and 0.92 times. Two lossless RAWs, from the Sony a7 V and Nikon Z50 II, came out just under the RAW's size with ZIP (0.993 and 0.995 times), each with a caveat explained in the method. No 16-bit TIFF, in any compression, was smaller than its RAW.

Compatibility is the other reason to think before compressing. The TIFF 6.0 specification moved LZW out of the baseline format into extensions "Due to licensing issues", and ZIP arrived later, through Adobe's Photoshop TIFF Technical Notes of March 22, 2002. In the spec's words, "A Baseline TIFF 6.0 reader is not required to support any extensions." If you don't know what software will open the file, uncompressed is the safe choice; for 16-bit files you control, use ZIP, not LZW.

Highlight recovery: the RAW kept detail the TIFF had clipped

On the most clipped file in RevelRaw's November 2026 test of 17 camera files, a Panasonic Lumix S5II frame, a −1 EV pull took 88.5% of the clipped pixels off pure white from the RAW and none from the TIFF. The reason: a TIFF's white point is set when it's made. Whatever reaches pure white then stays white, and pulling exposure down only turns it gray. A RAW file still holds the sensor's values above that point, as long as the sensor itself didn't clip.

Most of our files had little to recover: a median 0.0009% of photosites were clipped in the RAW, and 6 of 16 had none. Another six had at least 0.01%, up to 0.42% on a Panasonic Lumix S5II frame, the file whose TIFF clipped the most. Its 16-bit TIFF from the open-source converter had 5.54% of pixels clipped (the Mac's TIFF of the same frame had 2.52%), and 84.5% of those had no clipped photosite in the 3 × 3 patch around them: the sensor had recorded detail there, and the conversion set it to white. That file is the RAW's best case. Where the TIFF's white areas were mostly clipped on the sensor too, as on our three Sony a7 IV files, only 0.5% to 42% of the TIFF's white pixels had unclipped photosites around them, so the RAW had less to give back.

On that file we pulled exposure down by 1 and by 2 stops. From the TIFF, 0 of 1,334,832 clipped pixels regained any detail: every clipped value stayed on one flat value. From the RAW, a −1 EV pull matched to the same brightness took 1,181,212 of those pixels (88.5%) off pure white, spreading the formerly clipped values over 6,730 distinct levels; 84.5% of the clipped pixels sat over sensor data that hadn't clipped. That's an observation on one file, not a rate, but a TIFF has nothing above its white point to bring back. It keeps whatever its conversion kept, so pull highlights back in the RAW before you export. (The open-source converter's default render brightens until about 1% of pixels reach white, a rendering choice rather than lost data, so we switched that off for these numbers.)

Editors treat a TIFF differently too. Adobe Community Experts (not Adobe staff) answering a February 2024 thread about Camera Raw wrote that on a TIFF "Raw profiles can no longer be used" and that "Colors and tones will posterize much sooner in non-raw formats" (Conrad_C). RevelRaw's shadow test below bears that out for 8-bit TIFFs; a 16-bit TIFF came within about 10% of the RAW at the median.

8-bit vs 16-bit TIFF: what happens when you brighten the shadows

In RevelRaw's November 2026 test, a 16-bit TIFF kept nearly as many shadow tones as the RAW after a three-stop push at the median, and an 8-bit TIFF less than a quarter as many. A TIFF exported from a RAW file stores 8 or 16 bits per color channel; the Library of Congress notes "8-bit grayscale and 24-bit RGB color are common". That's 256 steps per channel at 8-bit against 65,536 at 16-bit. The 17 RAW files in the test were 12- and 14-bit.

To see what that means when you edit, we brightened the darkest 10% of each photo by three stops (+3 EV) and counted how many distinct green tones were left on a 256-step scale. Fewer tones means bigger jumps between neighbors, the steps you see as banding.

The median hides some spread: the RAW had 13% to 63% more levels than the 16-bit TIFF on 6 of the 14 files. If a photo might be edited again, export 16-bit; keep 8-bit for a finished file nobody will push.

What a TIFF drops: maker notes and the camera's white balance

No TIFF in RevelRaw's November 2026 test kept the camera's maker notes, the private block where each maker stores its own settings next to the standard EXIF any app reads. ExifTool counted 85 to 449 maker-note tags per RAW and none in any TIFF. All tags together fell from a median 465 per RAW to 94 in the Mac's TIFFs (17 files) and 50 in the open-source converter's (16 files).

No TIFF stored the camera's white-balance settings in a form you can re-adjust either. White-balance tags fell from a median 23 to 1 in the Mac's TIFFs, where the one left is only the Auto or Manual flag, and to 0 in the open-source converter's. The white balance is applied to the pixels, which is why setting it before export matters.

InformationMac's TIFF (17 files)Open-source converter's TIFF (16 files)
Camera make and model17 of 1716 of 16
Shutter speed, aperture, ISO17 of 1716 of 16
Lens17 of 170 of 16
Capture date and time17 of 170 of 16
Artist or copyright9 of the 9 files that had it7 of the 9 (lost on both Canon EOS R6 Mark II files)
Maker notesNoneNone
Camera's white-balance dataOnly the Auto or Manual flagNone
Color profileDisplay P3, 17 of 17sRGB, 15 of 16 (none on the Nikon Z50 II)
GPS locationNot tested: none of the originals had it

Recompressing can strip more. The open-source image tool we used to make the LZW and ZIP copies dropped all EXIF, but that's how that tool behaved, not something LZW or ZIP does. Check what your own exports kept with our free EXIF viewer.

When to keep RAW and when to export TIFF

Keep the RAW

Export a TIFF

Archiving: what the Library of Congress and the Smithsonian say

The Library of Congress's Recommended Formats Statement for 2025-2026 lists TIFF, JPEG2000, PNG and JPEG/JFIF as preferred formats for digital photographs. Its preferred characteristics include "Highest bit depth available, 16 bits per channel if available", "Uncompressed", "Unlayered" and an embedded color profile. It doesn't reject raw: "Digital Negative DNG (.dng)" and "Proprietary Camera Raw formats (.nef, .crw, .arw, .iiq)" are on its acceptable list, and its acceptable characteristics include "Lossless compression or lower compression ratios".

The Smithsonian's file-format guide for its digital collections (June 2019) is more specific about raw files. It rates NEF, a proprietary format, "High-risk long term", suggests converting NEF and CR2 files to DNG before they enter the archive, and names DNG its "Preferred RAW format". It describes TIFF as the most common format chosen for long-term preservation and, for NEF files, suggests generating TIFF for access alongside a converted DNG master. It accepts high-bit-depth TIFF masters in place of raw for some projects, but warns that with only a TIFF you may later "need to revisit original shot".

Neither tells you to throw the raw file away. For your own archive, that suggests keeping the RAW, or a DNG of it, as the master, with a 16-bit TIFF beside it as a finished copy.

DNG: the third option

DNG is a raw format, not a rendered one, so it sits on the RAW side of this comparison. We measured what a DNG conversion keeps and drops in our DNG vs RAW test of 16 camera files. Some converters can also write a "linear" DNG. In a March 2026 DxO forum test on one Fujifilm RAF, a 16-bit TIFF and a linear DNG could give "reasonably close results", but "The linear DNG still contains scene-referred data, while a 16-bit TIFF will usually contain output-referred data". In plainer terms, the DNG still holds the light the camera recorded, while the TIFF holds a finished picture. The author favored the linear DNG for archiving.

How to export a TIFF from a RAW file on a Mac

Preview opens RAW files from the cameras on Apple's macOS 27 camera list (current as of October 2026), and Apple's Preview guide covers the export:

  1. Open the RAW file in Preview.
  2. Choose File > Export.
  3. Choose TIFF from the Format menu. Preview offers HEIC, JPEG, JPEG-2000, OpenEXR, PDF, PNG and TIFF; Option-click the Format menu to see more.
  4. Pick where to save it, then click Save.

We didn't test Preview's export itself. We tested sips, an image tool built into every Mac that uses Apple's own RAW support. If you're comfortable in Terminal, sips -s format tiff photo.ARW --out photo.tif wrote a 16-bit, uncompressed TIFF with a Display P3 profile for all 17 of our files, in a median 0.50 seconds each.

Whatever you export with, choose 16-bit for anything that will be edited or printed, ZIP or no compression rather than LZW, and an embedded color profile. The Mac's converter embedded Display P3 in all 17 of its TIFFs and the open-source converter sRGB in 15 of 16; for print, WHCC lists "P3, Adobe RGB, and sRGB" as all fine.

RevelRaw, the Mac RAW editor we make, exports 8- and 16-bit TIFFs in sRGB, Display P3 or Adobe RGB, and has a Print export preset. We didn't run RevelRaw in this test. Editing is free, your first full-quality export is free, and Pro adds unlimited exports and batch export. It needs macOS 26 or later on an Apple M1 or later; the formats page lists the RAW formats it opens.

Common RAW vs TIFF claims, checked

The one measured pair we found agrees with RevelRaw's results. Lapseoftheshutter shows a "RAW file, 5472x3648px – 21.2MB" next to a "Processed TIFF file, no layers … 115MB", 5.4 times larger. It doesn't give the bit depth; the size matches a 16-bit uncompressed TIFF at that pixel count.

How we tested: 17 RAW files, two converters, two bit depths

Files. RevelRaw tested 17 CC0 RAW files from 11 cameras on November 12, 2026, converting each to TIFF two ways. We downloaded the public-domain files from the raw.pixls.us archive, the same set as our DNG vs RAW test; all 17 matched the archive's file fingerprints. They come from 11 cameras and six brands, listed in the size table, in CR3, CR2, ARW, NEF, RAF, ORF and RW2 at 12 and 14 bits.

Conversion. On an Apple M5 Pro Mac running macOS 27.0.1 we converted each file with sips (macOS 27.0.1), sips -s format tiff <raw> --out <tif>, at its default settings, and with version 0.22 of a widely used free, open-source RAW decoder, run from Terminal: 8- and 16-bit TIFFs with the camera's white balance, with and without auto-brightening. We recompressed those 8- and 16-bit TIFFs with an open-source image tool (LZW and ZIP at Deflate level 6, horizontal predictor on and off). The RAW side of the highlight pull and the shadow test used the same open-source decoder.

Measurements. Exact byte counts; a pixel-for-pixel check of every compressed file against its source; clipped photosites in each RAW against clipped pixels in each TIFF; a −1 and −2 EV pull from the TIFF and from the RAW on the most clipped file (the RAW pull used the decoder's own exposure setting with highlights clipped, not reconstructed, and the same tone curve as the TIFF, matched to its brightness); a +3 EV push on the darkest 10% of each photo, counting distinct green levels; and metadata read with ExifTool 13.55. Every number was measured on this Mac in this run; nothing was simulated.

Limits. The open-source converter couldn't read the Nikon Z50 II's High Efficiency NEF (it reported the data as corrupted but still wrote a TIFF that was 38% black), so that file is out of all its numbers (16 files); the Mac opened it, at 9.97 times the NEF. That converter keeps a black border on the Sony a7 V, which inflates its size, leaves its just-under-the-RAW ZIP result (0.993 times) uncertain and takes it out of the shadow test (15 files); it has no color data for the Nikon Z50 II, the other such case (0.995 times), so that TIFF has no color profile. Two builds of the converter rendered the OM-1 Mark II differently, so it's out of the RAW shadow comparison (14 files). The pull test covers one file, the most clipped in the set, and none of the files had GPS data. We didn't test Lightroom, Capture One, Photoshop, Preview's export or RevelRaw, and the Mac's results depend on the macOS version.

Cite this (the numbers are free to reuse under CC BY 4.0, with a link): RevelRaw, "RAW vs TIFF: 17 Camera Files Measured (2026)", November 12, 2026, https://revelraw.com/blog/raw-vs-tiff.html

FAQ

Is TIFF better than RAW?

Not as an original. In RevelRaw's November 2026 test of 17 camera files, 16-bit TIFFs were a median 4.8 times the RAW's size, kept no maker notes and stored no white balance you can re-adjust. On the most clipped file in the set, a −1 EV pull took 88.5% of the clipped pixels off pure white from the RAW and none from the TIFF. TIFF is better as a finished file to print, hand off or archive beside the RAW.

Does a TIFF keep the RAW data?

No. A TIFF stores finished red, green and blue pixels with the white balance already applied, not the sensor's one value per photosite. In RevelRaw's November 2026 test no TIFF kept the camera's maker notes or a white balance you can re-adjust, and highlights clipped at conversion stayed clipped. A 16-bit TIFF did keep shadow tones close to the RAW: a median 84 levels against the RAW's 87 after a three-stop push.

Can you convert a TIFF back to RAW?

No. A TIFF doesn't contain the sensor's per-photosite values, the highlight detail above its white point, the camera's white-balance settings or its maker notes, and nothing can rebuild those from finished pixels. Keep the original RAW, or a DNG of it, alongside any TIFF you export.

Why is a TIFF bigger than the RAW?

A RAW stores one 12- or 14-bit value per photosite, and cameras often compress it. An uncompressed TIFF stores three values per pixel at 8 or 16 bits each, so a 24-megapixel photo is 72 MB as an 8-bit TIFF and 144 MB as a 16-bit one, whatever the RAW weighed. In RevelRaw's November 2026 test, 16-bit TIFFs were a median 4.8 times their RAW.

Should I export an 8-bit or 16-bit TIFF?

16-bit for anything you might edit again or print. In RevelRaw's November 2026 test, brightening the darkest tenth of each photo by three stops left a median 84 green tones from a 16-bit TIFF, close to the RAW's 87, but only 19.5 from an 8-bit TIFF. An 8-bit TIFF is half the size, a median 2.44 times the RAW against 4.88, and fine for a finished photo.

Should I use LZW or ZIP compression for a TIFF?

ZIP, especially for 16-bit files. In RevelRaw's November 2026 test LZW made 16-bit TIFFs bigger than no compression in 15 of 16 files, while ZIP never did and saved a median 16%. Both are lossless: all 128 compressed files opened to exactly the same pixels as their source. On 8-bit TIFFs both shrink files when the horizontal predictor (a compression option for photos) is on. ZIP is slower to save.

Should I archive RAW or TIFF?

Keep the RAW, or a DNG of it, as the master, and add a 16-bit TIFF as a finished copy if space allows. The Library of Congress lists TIFF among its preferred formats for digital photographs and DNG and camera raw formats among the acceptable ones. The Smithsonian names DNG its "Preferred RAW format" and, for NEF files, suggests generating TIFF for access alongside a converted DNG master.

Does white balance still work on a TIFF?

Only as a correction on top of what's baked in. In RevelRaw's November 2026 test no TIFF stored the camera's white balance in a form you can re-adjust: it was applied to the pixels. Adobe Community Experts (not Adobe staff) wrote in February 2024 that on a TIFF, Camera Raw's white balance becomes "a relative -100 to +100 scale with the value at conversion as the new zero point, and this range is smaller". Set it before you export.

Why don't cameras shoot TIFF any more?

TIFFs are big and slow to write: Nikon's D2Xs manual says a TIFF (RGB) image took about 30 seconds to record to a 1 GB SanDisk Extreme III card, and in RevelRaw's November 2026 test even uncompressed 8-bit TIFFs, the kind the D2Xs wrote, ran a median 2.4 times the RAW. Some cameras still offer TIFF: Nikon lists TIFF (RGB) among the D850's file formats. Canon's EOS-1D X Mark III records JPEG, HEIF and RAW only.

Related reading: DNG vs RAW · Lossless photo formats · Color profiles for printing · How big is a RAW file?