Seizing the opportunity from efficient lossless compression using Adobe’s DNG Converter with special settings that are easy to miss
Shooting in RAW provides you with the best way to record the information captured by your camera sensor, and provides you with the most latitude for editing (for example, recovering details in shadows and highlights). However, camera RAW files tend to be large, with major implications for storage. In this article, I show that there is an opportunity to substantially reduce the space occupied by original camera RAW files while preserving the image data exactly. That is – to use the latest version of Adobe’s DNG Converter with specific (non-default) settings.
I investigated this opportunity using my own photographic archive. I compared the original and optimised file sizes for 23,078 photographs, with the optimised DNGs occupying about half the original space. To put that into familiar terms, 10 TB of camera RAW files would become approximately 5.3 TB, if they had the same mix of cameras and recording settings as this collection. This suggests substantial benefits for many photographers.
Even if you have previously used Adobe’s DNG Converter, there is a second opportunity too. Most of my archive had already been converted to DNG using Adobe’s standard settings. Recompressing those existing DNGs with an optional ‘hidden’ lossless JPEG XL setting (discussed below) recovered a further 1.18 TB across my files. The photographs did not need to be re-edited.
This article explains the two experiments, the results and the setting that is surprisingly easy to miss.
What 10 TB of camera files could become
These are measured storage ratios from my photographs, scaled to a common starting point of 10 TB. The largest reduction in this comparison came from my Sony A7CR files. If my hard disk only contained Sony A7CR RAW files, 10 TB would become about 3.7 TB. At the other end of the scale, a hard drive containing 10 TB of Canon 100D RAW files would allow a 2 TB reduction – still a very useful saving. Across the collection, this is equivalent to reducing 10 TB of RAW files to 5.3 TB. All this without introducing lossy compression into the conversion, with absolutely no image quality degradation whatsoever.
My recommendation: Adobe DNG Converter, but with the right option
I recommend using the latest Adobe DNG Converter and explicitly selecting lossless JPEG XL compression. This is the approach tested here.
But beware. Simply running the converter with its default settings is not enough. I made that mistake myself. In my tests with Adobe DNG Converter 18.6 on macOS, the standard lossless conversion still used the older compression method for the RAW image, even when a recent compatibility setting was selected. A JPEG XL preview inside the file made it particularly easy to think the main image had also been optimised.
Adobe’s own command-line specification distinguishes the default lossless option from -losslessJXL, which is what you need to select. The practical instructions below use that option.
The important message is that a DNG can be lossless without using the more efficient lossless compression available today. Existing DNG users may therefore have as much reason to investigate their files as photographers still storing camera original RAW files.
What “lossless” means, and why newer compression can make the same image smaller
Lossless compression preserves the numerical image values exactly when they are decoded. It does not merely aim to look the same. There is no quality slider trading image information for a smaller file, and no reduction in resolution to obtain these results. So, you may be able to halve the space required to store your RAW files without losing image quality.
There is an important distinction between preserving image data and preserving the original file byte for byte. DNG is a container for RAW data and the information needed to interpret it. The compression method inside that container can evolve. Importantly, Adobe added JPEG XL lossless compression to DNG 1.7; the DNG 1.7.1 specification documents its use for both mosaic and linear RAW images, as well as previews.
JPEG XL stores the same image information more efficiently than the older lossless compression used in many DNGs. It makes better use of patterns in the data, allowing the original image values to be recovered exactly. For the technical explanation, see codec developer Jon Sneyers’ JPEG XL’s Modular Mode Explained.
Experiment 1: original camera RAWs to optimised DNGs
This comparison used 23,078 photographs from 13 camera models, with ARW, CR2 and RAF originals. The original file sizes had been recorded during an earlier duplicate scan, before those originals were removed. Those records were matched to the corresponding DNGs and their subsequently optimised versions.
This was therefore a comparison using recorded original sizes, rather than a new batch conversion of 23,078 still-present originals. The later pixel checks compared the existing DNGs with their recompressed versions. They do not constitute a fresh pixel comparison against the deleted camera originals.
My files had gone through two steps: camera RAW to a conventional lossless DNG (using lossless JPEG compression), then that DNG to a DNG using the latest lossless JPEG XL compression. A photographer starting with camera RAW originals today can, obviously, apply the newer compression directly; there is no need to make an intermediate older DNG first.
| Camera | Files | Original GB | Optimised GB | 10 TB becomes |
|---|---|---|---|---|
| Sony A7CR | 262 | 33.48 | 12.47 | 3.7 TB |
| Fujifilm X-T2 | 787 | 39.63 | 15.96 | 4.0 TB |
| Sony A7R III | 1,994 | 153.60 | 64.22 | 4.2 TB |
| Sony A7R IV | 1,710 | 210.51 | 89.64 | 4.3 TB |
| Sony A7 III | 3,548 | 169.09 | 72.38 | 4.3 TB |
| Fujifilm X-H1 | 321 | 16.15 | 6.92 | 4.3 TB |
| Fujifilm X100F | 9 | 0.45 | 0.24 | 5.4 TB |
| Sony A7R V | 6,932 | 564.98 | 322.28 | 5.7 TB |
| Sony A1 | 3,818 | 208.84 | 133.18 | 6.4 TB |
| Canon 5D Mark IV | 1,310 | 41.24 | 29.15 | 7.1 TB |
| Canon 5D Mark III | 993 | 26.84 | 20.62 | 7.7 TB |
| Canon 5D Mark II | 1,053 | 27.36 | 21.06 | 7.7 TB |
| Canon 100D | 341 | 7.81 | 6.22 | 8.0 TB |
| All measured files | 23,078 | 1,500.00 | 794.35 | 5.3 TB |
In total, 1,500 GB of camera RAW originals were reduced to 794 GB, saving about 706 GB. The familiar 10 TB illustration is calculated from the total bytes, not by averaging the camera percentages. These results describe my archive, with its subjects, exposure settings and camera modes; they are not a controlled ranking of camera brands.
Uncompressed RAW originals have more scope for reduction because compression is doing work that the camera did not do. Files already compressed in-camera can still benefit, but by a different amount. Scene detail and noise also affect what can be compressed.
Experiment 2: improving files already converted to DNG
The second experiment asks a different question: how much more space can I recover if I have already converted my camera files to lossless DNG? I recompressed 173,061 files spanning 2006–2025, covering two decades of photography. Overall, the total fell from 6.80 TB to 5.62 TB, recovering 1.18 TB.
| Year | Files | Before GB | After GB | Saved GB |
|---|---|---|---|---|
| 2006 | 2,034 | 17.50 | 14.45 | 3.05 |
| 2007 | 2,883 | 32.86 | 27.35 | 5.51 |
| 2008 | 3,800 | 46.05 | 38.82 | 7.23 |
| 2009 | 5,521 | 124.48 | 110.46 | 14.02 |
| 2010 | 5,632 | 123.57 | 109.66 | 13.92 |
| 2011 | 2,166 | 46.84 | 42.03 | 4.82 |
| 2012 | 11,758 | 260.32 | 230.85 | 29.47 |
| 2013 | 5,825 | 134.57 | 119.25 | 15.32 |
| 2014 | 5,354 | 120.75 | 107.97 | 12.78 |
| 2015 | 4,252 | 94.96 | 84.90 | 10.06 |
| 2016 | 4,749 | 109.99 | 97.25 | 12.74 |
| 2017 | 14,756 | 410.51 | 339.32 | 71.19 |
| 2018 | 7,825 | 211.87 | 184.82 | 27.05 |
| 2019 | 4,258 | 178.39 | 159.69 | 18.70 |
| 2020 | 1,524 | 80.11 | 71.29 | 8.82 |
| 2021 | 26,171 | 1,236.69 | 968.97 | 267.72 |
| 2022 | 17,810 | 945.06 | 775.11 | 169.95 |
| 2023 | 17,006 | 1,005.02 | 766.42 | 238.59 |
| 2024 | 22,744 | 1,323.84 | 1,103.09 | 220.75 |
| 2025 | 6,993 | 297.93 | 269.17 | 28.75 |
| Total | 173,061 | 6,801.32 | 5,620.88 | 1,180.44 |
Most of the space was recovered from later years because those folders contained many more bytes to begin with. A smaller percentage applied to a large folder can save more storage than a dramatic percentage applied to a few photographs.
How the additional saving varied by camera
The camera breakdown below covers 17,309 accepted files in the sampled camera results. To make the implications easier to understand, the last column again scales each result to 10 TB, this time starting with existing default-settings DNGs.
| Camera | Sample files | Before GB | After GB | 10 TB becomes |
|---|---|---|---|---|
| Sony A1 | 2,793 | 132.11 | 95.82 | 7.3 TB |
| Fujifilm X-T2 | 331 | 10.19 | 7.42 | 7.3 TB |
| Fujifilm X100F | 270 | 8.71 | 6.69 | 7.7 TB |
| Fujifilm X-H1 | 32 | 0.90 | 0.70 | 7.7 TB |
| Sony A7R V | 3,049 | 178.45 | 138.40 | 7.8 TB |
| Fujifilm X100V | 9 | 0.32 | 0.25 | 7.8 TB |
| Canon 350D | 132 | 0.93 | 0.77 | 8.2 TB |
| Canon 5D | 696 | 7.80 | 6.49 | 8.3 TB |
| Canon 5D Mark IV | 944 | 24.92 | 21.98 | 8.8 TB |
| Sony A7 III | 829 | 19.74 | 17.51 | 8.9 TB |
| Canon 50D | 49 | 0.83 | 0.74 | 8.9 TB |
| Canon 5D Mark III | 2,653 | 59.85 | 53.19 | 8.9 TB |
| Canon 5D Mark II | 1,914 | 42.39 | 37.69 | 8.9 TB |
| Sony A7R III | 774 | 30.79 | 27.55 | 8.9 TB |
| Canon 7D | 1 | 0.02 | 0.02 | 8.9 TB |
| Canon 450D | 4 | 0.06 | 0.05 | 9.0 TB |
| Sony A7R IV | 1,940 | 113.75 | 102.73 | 9.0 TB |
| Canon 100D | 90 | 1.74 | 1.58 | 9.1 TB |
| Sony A7CR | 799 | 46.59 | 42.61 | 9.1 TB |
The very small Canon 7D, Canon 450D and Fujifilm X100V samples should be read as examples, not dependable estimates for those cameras. The apparently different savings for the same camera across the tables are not contradictions: the starting formats and the photographs in each population differ.
How I checked that the image data was preserved
All 173,064 checked files preserved identical decoded image values. I compared the image data before and after conversion numerically, rather than relying on the photographs simply looking the same.
The checks also examined the compression used and selected metadata. They allowed 173,061 files to replace their previous versions; three were retained unchanged because of metadata differences or because the requested compression had not been used.
What about iPhone and Leica DNGs?
Our results showed that “already a DNG” does not necessarily mean “already efficiently compressed”. The older lossless-compressed ProRAW examples from my iPhone 12, 14 and 16 Pro Max became approximately half their previous whole-file size, with identical decoded pixels.
| Pilot group | Files | Before MB | After MB | 10 TB becomes |
|---|---|---|---|---|
| iPhone 12 Pro Max | 8 | 259.5 | 126.6 | 4.9 TB |
| iPhone 14 Pro Max | 8 | 576.0 | 269.1 | 4.7 TB |
| iPhone 16 Pro Max | 8 | 587.5 | 293.1 | 5.0 TB |
| Leica Q3 | 8 | 605.3 | 455.4 | 7.5 TB |
| iPhone X (app DNG) | 3 | 33.0 | 28.1 | 8.5 TB |
Apple ProRAW is also already computationally processed image data, rather than an untouched Bayer sensor mosaic. Lossless recompression preserves the data present in it; it does not reverse that earlier processing. Files already using lossless JPEG XL may offer little additional saving. These native-DNG cases were investigated separately and are not included in the 173,061-file library result.
How to try it on your own files
Download Adobe DNG Converter from Adobe. My experiment used version 18.6 on macOS. Choose examples from the cameras and recording modes you actually use, including edited files and any unusual DNGs. Create a separate output folder so that you can compare the results with their sources.
Remember to request lossless JPEG XL explicitly
On a Mac, open Terminal (in Applications → Utilities). The command below runs Adobe DNG Converter with lossless JPEG XL compression explicitly enabled. Replace the example input path and output folder with your own, create the output folder first, then paste the command into Terminal and press Return. This example converts one file.
"/Applications/Adobe DNG Converter.app/Contents/MacOS/Adobe DNG Converter" -losslessJXL -d "/Volumes/Photos/Optimised" "/Volumes/Photos/Originals/DSC1234.ARW"The important option is -losslessJXL. The -d option specifies the destination. The output will be a DNG. On Windows, Adobe documents the same options; the executable path must match your installation. The command shown here is the macOS route used for this experiment.
Compatibility is the main practical trade-off
An application supporting DNG in general does not necessarily support JPEG XL inside a DNG. Check the actual files in the versions you use.
The evidence below was reviewed in September 2026. It is deliberately specific about what the sources establish, rather than a blanket list of applications that supposedly work.
| Software or library | What the evidence establishes |
|---|---|
| Adobe Camera Raw / Lightroom | The tools used for this Adobe-based workflow. Use current versions and test your files; older DNG support alone is insufficient. |
| Iridient X-Transformer 4.0 | The November 2025 release notes explicitly add lossless and lossy JPEG XL DNG output. This is evidence of wider adoption, not a replication of my experiment. |
| LibRaw 0.22 | Its January 2026 release notes add DNG 1.7/JPEG XL via Adobe DNG SDK integration. An application must include the relevant build support. |
| darktable 5.6 | The June 2026 release notes list JPEG XL DNG under missing compression-mode support. |
| RawTherapee | The DNG 1.7 support request remains open. Do not infer support from the .dng extension. |
| DxO PhotoLab / PureRAW | The published PhotoLab input requirements and PureRAW input requirements restrict Adobe-converted RAW DNGs to files without compression. Check your current version rather than assuming acceptance. |
My results support this approach for my Adobe-based workflow. They do not establish compatibility for every editor, operating-system thumbnail generator or NAS photo application. That is the check to make before committing your own archive.
Where this fits in my photographic workflow
Camera RAW files take up a lot of storage space. Efficient lossless RAW storage gives me a sound foundation for both my existing photography and future HDR work. It substantially reduces the storage burden without using an image-quality compromise (for example, lossy compression) to achieve it.
For my archive, the evidence supports a clear recommendation: investigate lossless JPEG XL DNG before deciding that storage savings require discarding RAW image information through the use of lossy compression. The camera RAW originals delivered substantial file size savings, and even my existing lossless DNG collection contained another 1.18 TB that could be recovered. The key was choosing the newer compression explicitly and checking the results.
Scope and sources
The measurements come from my September 2026 conversion records, matched-original size records, verification logs, pilot results and camera summaries. Totals above were recalculated from byte counts where available. Tables show rounded sizes, so displayed rows may not sum exactly to displayed totals. Whole-file sizes include previews unless labelled RAW payload. The 10 TB figures are illustrations of the measured ratios, not additional experiments.
The main original-camera comparison covers Sony ARW, Canon CR2 and Fujifilm RAF. It does not establish savings for Canon CR3, Nikon NEF, Olympus/OM System, Panasonic or Pentax originals. The metadata checks were filtered, and the pixel verification relates to DNG recompression rather than a fresh comparison with deleted camera originals.
Technical explanations and application statements are linked to Adobe’s specification and command-line documentation, the codec developer’s explanation and the relevant software developers’ documentation. My separate lossy DNG investigation compares the additional storage savings and explains why I use lossy JPEG XL after editing while preserving RAW data losslessly. None of those additional savings have been folded into the lossless results presented here.
Technical detail: how the image data was checked
The verification decoded the main RAW image from each source and output DNG, restored sample ordering where interleaving was used, checked that the array dimensions matched, and compared the numerical values for exact equality. It also inspected the compression method of the main RAW image, independently of any embedded preview.
Metadata was checked separately using a filtered ExifTool comparison. Conversion history, storage layout and several image-directory groups were excluded, so this check was not intended to establish that every metadata field remained unchanged. Two files were held back for metadata issues and one because its output did not use JPEG XL. All three passed the image-value comparison.
Last updated: September 2026