As photographers, one of the first things we learn is that we should always shoot in raw. This gives us the most flexibility and data to work with in editing, with the drawback being large file sizes that have gotten even larger over the years. What if you could have all the flexibility of a raw file at ten percent of the file size?
In this video, I show you the power of how efficient JPEG XL can be in a professional workflow. Keep in mind that just because it has "JPEG" in the name does not mean it is like the JPEG format we've used for over 30 years. This is much more advanced, supporting 32-bit color depth, HDR, wide color gamuts, and a myriad of other features. Adobe introduced the ability to utilize JPEG XL in their DNG files last year, and you've likely been using it without even knowing it!
Since 2023, anytime you merge an image to a panorama or an HDR, it is now utilizing JPEG XL—which you might have noticed because file sizes got noticeably smaller without any loss of data. This same pipeline can be used for converting your raw images to lossy DNG files that give you most of the flexibility you want in editing, with a huge file savings.
In the video, I show you examples by zooming in to the maximum 1,600% for extreme pixel peeping. I discuss the downsides of converting your raw images this way and encourage you to try this on your own by guiding you through how to do it yourself. You’ll likely not be able to see a quality difference through example images on YouTube or on the web, like here on Fstoppers. You really should try it for yourself, as the results are incredibly impressive!
I'm excited for the potential in JPEGXL, but I'm going to guess it's going to have the same fate as JPEG2000 just because it's called JPEG.
I think Nikon's High-Efficiency RAW is a good idea because out of cameras like my Z6III, the RAW file size is only about 10-12MB ea at full resolution with no quality loss, *and* it keeps that mosiac data you need for DXO.
That said, 85MB for that Fuji RAW file is gross and makes me happy I don't shoot Fuji because... c'mon, that file size is just dumb.
--- "but I'm going to guess it's going to have the same fate as JPEG2000 just because it's called JPEG."
To be fair, when exporting to JPEG XL via LR, it's saved in a DNG container and the option is "Use Lossy Compression". No one would see "JPEG" unless they dug in the exif.
Pretty nice. With the 5 minutes I spent testing it:
Pros:
1. Significantly smaller size.
2. Opens pretty quickly with Capture One, Evoto, Lightroom, and Photoshop. Because it's compressed, I was expecting a delay for it decompress similar to compressed TIFF and PSD files in Photoshop. No such delay.
Cons:
1. JPEG XL does not appear to be natively supported in Windows. Although, I can open the files via software I listed above, when viewing in file explorer there's no thumbnails. There's probably some workaround, but, I'm not going to do all that.
With that said, I think I'm too paranoid to convert fully. Having the raw files just gives me that extra sense of security. But, if collaborating where I need to send raw like images, I may try this route.
You have to be careful when you say "file sizes got noticeably smaller without any loss of data" because that simply is NOT true. The JPEG XL format Lightroom Classic uses for those HDR merges of your bracketed images, then saves them for you as DNG files IS LOSSY, pure and simple. I.e. data is removed and lost - so if you want to abide by the golden rule of photography and edit in a non-destructive / parametric / lossless (whatever you call it) manner, this is NOT it. Even if it still looks great to your naked eyes (i.e. better than other compresses/lossy methods you've used before), it is STILL lossy.
You can see for yourself by viewing the EXIF data of the DNG in terminal. Sadly EXIF (and every other one I've tried, even from obscure GitHub's) will NOT tell you what compression method JPEG XL used in that file (and JPEG XL can use both lossless AND lossy saving methods btw), BUT you can still see the specs of that JPEG XL as in pixel size, BitsPerSample, etc... then as proof this truly is lossy, save that merged non-DNG file yourself, in those same exact settings, as a JPEG XL, and you'll notice the lossless version is commonly 3+ times the file size LRC auto-saved. The ONLY way to get it down to the size of the LRC one is to used the compresses/lossy JPEG XL saving method, proving that it's a lossy image - data is stripped from it.
You can likewise prove that what LRC exports as your HDR merged image in DNG format isn't a true lossless image (i.e. same as Camera RAW which can also use DNG, which IS lossless) by bringing it into any of Topaz's (or similar) programs - and on export you'll see it won't let you save it as a DNG because it's detected a non-camera RAW file inside that DNG container (and their program is setup to ONLY save as DNG if it really is a lossless/RAW file). Don't hold me to this, but I also believe using Adobe's ancient DNG converter program will likewise give you errors saying it's not really a lossless RAW/DNG file and won't let you use that tool.
The ONLY way to get truly lossless (and I use that term lightly here, as the HDR merge process by definition STILL loses some data from your bracketed images in the merge process itself) is by using Photoshop's Merge to HDR Pro method - which is a pain because there is NO way to automate that as a batch, so it has to be done HDR bracket by HDR bracket one-by-one. Another proof that this way is more-proper is you'll see that image is 32-bit (which it expands that from the 12-16 bit RAW your camera takes, by enlarging the color gamut? or bit depth? or whatever that is called, making it the larger/greater 32-bit), AND bringing it into programs like Topaz there is NO error - that program thinks it's truly lossless RAW data as well.
Anyway, sorry for the wall of text. I Literally signed up just to post this comment since I've been digging DEEP into Lightroom Classic's HDR's saved as DNG (trying to get a proper workflow in place!). And since I'm failing at finding the perfect truly lossless method I've started posting on articles and reddit posts to figure this all out, as it's a bear!