Correct! Excellent, though why that should work is beyond me.
It does more than double the file size; 96.7MB from 42.3MB, but it’s still half the size of the uncompressed.
Cheers Dave
1 Like
Because the compatibility level limits the DNG compression choices to the “older” ones only (uncompressed or lossless JPEG).
So what happened was: uncompressed → lossy JPEG XL compression → decompress that lossy → recompress that w/ “old” lossless JPEG. There is some (hopefully unnoticeable) data loss compared to the previous direct uncompressed → lossless JPEG workflow so the file size can still end up smaller compared to the latter.
I just ran a 32MB Canon R7 file through DxO and generated compressed and uncompressed DNGs. The uncompressed DNG was 102MB and the high-efficiency (which DT cannot read) reduced to 16MB.
I ran both DxO generated DNGs through Adobe. The 102MB file reduced to 10MB and the compressed 16MB reduced to 9MB and I saw no discernible difference between the two. My take away is that it doesn’t matter much either way, but I will probably opt to generate the full sized DNG - which I would delete at a later point.
Not the same process as suggested above I believe - the outcome is dt unsupported lossy JPEG XL DNG 1.7 in both cases here, no? Just shows that Adobe is probably by default using slightly more aggressive JPEG XL compression parameters than DxO. (And in the last case you’re doing lossy compression twice, which is seldom a good idea, no matter how good the codec…)
Yes, you’re right - any compression will still require ACR 14.0. My question is if I want to use a compressed format then do I run the full sized DxO DNG or is there an advantage to further compressing the DxO high-efficiency file. The result seems to be that it doesn’t matter much.
Recompression lossy on top of lossy is a slippery slope. You can’t know for sure when it’ll catch you out.
1 Like
OMG … I rather stick to the original DXO DNG , instead of going this crappy route through another piece of software , creating another file and deleting the HF dng . Just a disaster IMHO .
But this is just me …
I keep hoping that these HF dng files will be readable by DT … sooner or later
Andreas
1 Like
Up until now I just copy my edits to the source raw after I’m done and delete the DNG. Drawn masks don’t line properly but I can always recreate another DNG if I decide to revisit the image sometime later.
I might change my workflow now that I realize compressed files are available, but I haven’t decided.
You should lossy compress the minimum times possible.
When compressing lossy with adobe DNG compressor the already lossy compressed Pureraw file, you are doing it 2 times.
With this you are risking the “photocopy effect”, meaning if you take a photocopy of a photocopy (and so on), each new resulting copy is worst than the one before.
My advice is, for the resulting DNG file to be open in darktable (at the moment): create the lossless (big) DNG in Pureraw, than lossy compress it with Adobe DNG Converter choosing a compatibility mode 14 or bellow (modes 15.3 and 16 use jpeg-xl in the DNG, a format darktable does not open).
2 Likes
High fidelity to save space makes no sense.
I am quoting from the DXO website: https://support.dxo.com/hc/en-us/articles/34336869360157-Is-the-new-High-Fidelity-DNG-compression-truly-lossless
Is the new High-Fidelity DNG compression truly lossless?
The new compression is perceptually lossless.
This means: There are minimal mathematical deviations at the pixel level. > However, these are not perceivable to the human eye, even under intensive image editing.The compression was specifically developed and validated for RAW image data. It reduces only information that, from a perceptual point, does not contribute to visible image quality.
The result: About four times smaller files with unchanged perceptual image quality.
Of course these claims shoot be verified. The same applies to the High Efficiency Nikon Raw files.
But if that is really the case… no visual difference and the same level of raw capabilities. Then I would very very much welcome a 1/4 of the storage space… Because that would mean
- ¼ of the storage space on my system
- ¼ of the storage space on my backup
- ¼ of the storage space on my cloud backup
- ¼ of the time uploading to my cloud backup
- less spending on camera cards
- ¼ of my spending on cloud backup
If The result: About four times smaller files with unchanged perceptual image quality. is really the case. I would very gladly use it.
1 Like
Why does it make no sense ?
Hello everyone,
Rounding the sizes, my Canon 7D outputs a 24MB file, the uncompressed DNG from DxO 6 is 64MB (49MB in DxO 5), and the compressed version is 16MB, unusable in Darktable.
With the latest Adobe Compressor, in 14.0 compatibility mode, with no lossy compression and a medium-sized JPEG preview, produces a 16MB DNG that is usable by Darktable 5.4.1. The only minor issue is that highlight reconstruction no longer works with LCH.
Until Darktable can read the compressed DxO file, I will only process the truly noisy RAW files with DxO and delete the uncompressed DNG after processing with Adobe DNG.
If it were lossless, you would call it lossless.
Since it isn’t…it’s a reduction in fidelity vs the original (and commonly-used) format.
Ergo “high-fidelity” is an inappropriate label.
Image fidelity , often referred to as the ability to discriminate between two images[1] or how closely the image represents the real source distribution.
So, given this definition high-fidelity is an appropriate label. It is a reduction in fidelity, but it is minimal. If there was no loss, you call it lossless.
Actually and that is what they are not claimed there is not loss of information, they claim is it visual not different.
Ergo “high-fidelity” is an appropriate label.
Lower than normal fidelity isn’t high fidelity to me. Shrug.
Are you honestly critisizing the name of a stupid DNG ?
This thing has to have a name … for me they could call it whatever they want !!
To me it is about saving time and disk space … and not about a name !
I do not even use the format , as at the moment it is not usable in DT … without going through another piece of software . So pretty useless for me .
Why spending thoughts and time about a name …
I rather spend time with the technical pros and cons 
I’m sure it’s not when you compare with a lossless compressed DNG without much editing, side by side.
It’s not magic, it’s a v1.7 spec DNG, using something like -jxl_distance 0.1 (highest quality).
The question is what what happens when you really begin to shove sliders around. Like, what happens when you try to to get a shadowed foreground back using tone equalizer?
My first gen Sony A7 only had lossy compressed ARWs and there was always something iffy about these files. Everybody said they were basically the same as lossless, but when doing more extreme corrections I perceived a very real lack of editing headroom. I can’t for sure say it had to do with lossy raw, but I have never ever experienced anything like it since I started shooting lossless or uncompressed raw back in 2001.
1 Like
Did you test the same scene with lossy and lossless?
It only had lossy raw. No, I did not try the same scene with two different cameras.
1 Like