Giant JPEGs from Capture One?

[Carl Seibert]

Advertisement (gone after registration)

Hi, 

I've had Capture One for only about a month. I just discovered that it is making over-large JPEGs from my M10R DNGs. Sometimes the exported 8-bit JPEG file is bigger than the 16-bit RAW it came from!

I tried processing the an image file through other software and in every case the resulting JPEGs were smaller than C1's, by 25 to 50%.  (All these images were output at the 100% quality setting, so variation between quality slider scales shouldn't be a factor.)

I also made a 16-bit TIFF and exported it to JPEG from an assortment of programs, including Capture One. The resulting files' sizes were spread over a narrow range, as would be expected. This suggests strongly to me that Capture One has an issue with Leica DNGs, or at least DNGs from an M10R.

Has anyone else seen this?

Could somebody who has both an M10R and Capture One try to replicate my experience?

Thanks.

-Carl

[Photoworks]

DNG is 54MB

JPG at 100% about 27MB

Darker images are larger than white images

[Robert Blanko]

DNG 46,3 MB

JPG 45,6 MB, JPG maximum quality, 100%

Edited February 13 by Robert Blanko
switched to English

[Carl Seibert]

Interesting. Photoworks seems to be getting better compression than Robert or me.

It would be really freakin' nice if these developers would just tell us what JPEG settings they use. That might explain the difference. It might be down to something simple like C1 uses no chroma subsampling whatsoever and On1 and Lightroom do. Knowing what's going on there would be the first step toward the next - and bigger - question, which is "would it be acceptable to set C1"s quality down a few points (which would cause file size to plummet)?" Would I end up with the same quality that On1 is giving me at 100%?

I realized after posting that comparing to the RAW size is a bit fraught. If Leica does a much better job with lossless RAW compression than Nikon does, that would explain a difference in the RAW-to-JPEG ratios from the two cameras. But, honestly, it looks like the two cameras are achieving about the same RAW compression. The Leica files are about 10% larger than the Nikon's, which comports with the difference in resolution between the two.

For grins, I just looked at JPEGs from M246 monochrom files. Interestingly, C1, On1, and Affinity's results all fell in about a 10% range. That suggests to me that maybe chroma subsampling is the significant variable here. Well, OK, if we can confirm that is what's going on, I might be will to trade bigger files for fewer JPEG artifacts.  

[FaustoReid]

10 hours ago, Robert Blanko said:

DNG 46,3 MB

JPG 45,6 MB, JPG maximum quality, 100%

Thank you so much. 

Edited February 14 by FaustoReid

[Chris W]

17 hours ago, Robert Blanko said:

DNG 46,3 MB

JPG 45,6 MB, JPG maximum quality, 100%

Yeah. Capture One has multiple JPEG outputs you can choose, all result in different mb sizes - high quality, Instagram ready, for web use, small size etc..

From memory, I'm not at my computer, C1 full quality JPEGs are about the same size as their 16 bit TIFF. It's just a different format (TIFF vs JPEG), it isn't meant to be compressed or data thrown away.

And you can go into all the export settings and change things, or create your own JPEG settings.

I have zero issue working with C1 and my M10.

Edited February 14 by Chris W

[Robert Blanko]

Advertisement (gone after registration)

Yup, you can set different default output qualities.

IMHO, the JPG size to some extend also depends on the post processing of the photo, for instance due to adjusting the illumination and thereby eliminating possibly information.

If the compressed information contained in the RAW file / DNG is preserved in the JPG as well as much as possible, it would be surprising that the JPG size would be significantly different from that of the DNG. At the end of the day, it would be just the same information, however in a different representation - and possibly different lossless compression.

[Chris W]

I just checked a few jpeg vs tiff in Capture.

Using the default full quality TIFF and JPEG outputs both my M10 files are exactly the same 136mb. A smaller JPEG output (optimised for Instagram) yields a 1mb file.

Same with my Hasselblad X1 DII. Full quality TIFF and JPEG output are exactly the same, 281MB.

 

[Carl Seibert]

Chris, that's really interesting. I noticed that a Zip-compressed TIFF wasn't a ton bigger that C1's 100% JPEG, about 65 MB vs fifty-something, If I remember correctly.

So, the trade-off wouldn't be too bad to output TIFFs. 

But I use JPEGs as a distribution format. I keep the RAWs and all their processing information as a reference if I need to make a higher-quality output. So I still need to figure out a good JPREG compromise.

[Carl Seibert]

Hi All,

I've managed to come up with a little more clarity here.

The goal, I see now, is to come up with a quality setting for Capture One's JPEG exporter that will yield quality and file size results that reasonably match those that I get from On1 RAW at 100%.

The whole comparison with RAW file size appears to be a dead end. All we're learning there is that Leica and Nikon are doing a fabulous job with their lossless compression. They are both achieving, on average, with my test files, about a 1:5.3 ratio. 1:4.6, if I correct for the cameras' shooting at 14 bit, not 16. Back when the earth was cooling and I was being taught this stuff, 1:2.5 was what we were told we could expect from lossless compression. Whatever they are doing is very efficient. The 40 year old JPEG standard ain't so efficient. 

Capture One was achieving only about a 1:2.5 or 3 ratio on 100% JPEGS; On1 was getting about 1:4-ish. 1:4 is what I tend to expect since back when time began.
 
There's no real mystery about JPEG encoding. You choose parameters to get the best compromise between file size and image quality, which is dependent on image content and intended use. You've only got the parameters in the standard to work with. 

I would have expected that "100%" would mean that all parameters were set at their max possible values. But it's clear that "100%" means something different to each of these developers. And that's OK, I guess really, since different images react differently to each parameter. What's "best", let alone what's the best compromise becomes subjective.

I made four test images by saving edited images as TIFFs, then exporting jPEGs based on the TIFFs from each program, so as to have the same edits represented in all the sample files. I exported from On1 at 100% and from Capture One at 100%, 99%, 98%, and 97%. I did 90% as well, just for grins.

As expected, file size diminished rapidly as I stepped Capture One down through the quality levels while image quality diminished much less rapidly. (If at all, some would argue.) By the way, it wasn't a given that a 1% change in the slider would make any difference at all. It doesn't in Adobe products. 

The C1 files seem to shrink smaller than the 100% On1 files at about the 98%-ish level.

The tricky part is trying to tease out the image quality results. The longer I pixel peep at these images, the worse it gets. I start to not see differences I (thought?) I saw a minute ago or see differences that might or might not even exist. Trying to do this entirely subjectively would take forever and might not even work. Despite the fact that "subjectively" is what matters. We're making pictures here, not math.

I have asked both companies for their parameters and I'll post them here if I get responses. I think that having some quantitative input will help. Otherwise I'll just have to guess. Actually I might just have to guess anyway.

I do still see a difference in how C1 is compressing files between the Leica and the Nikon. It seems to do way better with the Nikon files. On1 does better with the Nikon files, too, but to a lesser extent. Now, there are certainly uncontrolled variables here. My test files are all from the same assignment. Meaning that the Nikon is for a long lens (a zoom) and the Leica is being used with short primes. Wide angle pictures look different than telephoto ones. But the fact that both cameras are producing equivalent sized RAW files suggest to me that the image subject matter is reasonable equivalent. Or maybe not.

I can share my files if anybody wants to torture themselves/play with them. 

Thanks for all your help.

I'll post more when I know more.

-Carl

[Robert Blanko]

vor 23 Stunden schrieb Chris W:

I just checked a few jpeg vs tiff in Capture.

Using the default full quality TIFF and JPEG outputs both my M10 files are exactly the same 136mb. A smaller JPEG output (optimised for Instagram) yields a 1mb file.

Same with my Hasselblad X1 DII. Full quality TIFF and JPEG output are exactly the same, 281MB.

 

136MB appears to be incredibly huge. 🤔 I can‘t remember that I have ever seen a three digit file size for my M10R photos. Mostly around 40-50 MB.

[Carl Seibert]

This just in.... I don't think using the size of an uncompressed TIFF of a de-mosaiced image as the equivalent amount of data in a RAW capture was a good idea at all. If you just do pixels-times-bits and add in some overhead you get a less than 100MB of data instead of over 200MB. Checking Fuji files compressed vs uncompressed seems to confirm this.

So RAW files - even though they are usually TIFFs -are small more because of the data than terrific compression. 

Mea culpa.

It turns out to be irreverent anyway, but still it sucks to use a wrong assumption.

[Robert Blanko]

"I do still see a difference in how C1 is compressing files between the Leica and the Nikon. It seems to do way better with the Nikon files. On1 does better with the Nikon files, too, but to a lesser extent."

That's interesting. Maybe the Leica-files have more information. The dull Nikon photos simply cannot resolve the fine details that top-notch Leica lenses can resolve, so that high frequency components in the Nikon images do not exist which in turn saves some storage space. 🤔😁

[Chris W]

21 hours ago, Robert Blanko said:

136MB appears to be incredibly huge. 🤔 I can‘t remember that I have ever seen a three digit file size for my M10R photos. Mostly around 40-50 MB.

This is just reading the file size on the export page - exporting full size TIFF. My X1DII files are almost double in size, which I would expect from a 50mb medium format camera.

[Chris W]

On 2/15/2025 at 11:11 PM, Carl Seibert said:

 

The C1 files seem to shrink smaller than the 100% On1 files at about the 98%-ish level.

The tricky part is trying to tease out the image quality results.

It depends what your distribution is for.

If I'm sending someone a file just for reference, I'll send them a jpeg 'for web use'. If I'm sending it to my iPhone to post on Instagram, I'll export jpeg 'optimised for instagram'.

Those jpegs by the way look very good quality to me.

If I'm sending someone an image to use (on a poster, book, or album cover, I send them full quality (only).

[Carl Seibert]

21 hours ago, Robert Blanko said:

Maybe the Leica-files have more information. The dull Nikon photos simply cannot resolve the fine details that top-notch Leica lenses can resolve, so that high frequency components in the Nikon images do not exist which in turn saves some storage space. 🤔😁

Giggle. Yeah, I actually thought about that. I wondered if the Leica was producing more color information in its 14-bit capture than the Nikon was in its 14-bit capture. Naw, don't think so. The most high frequency-ful image I looked at was a Nikon one. From the Leica, I had an image at 800 ISO (so, noisier) and an image at 200 ISO. From the Nikon, two at 400 ISO. The long lens vs short lens thing might have some traction.

I dunno. It's a mystery.

[hansvons]

On 2/13/2025 at 3:38 AM, Carl Seibert said:

I've had Capture One for only about a month. I just discovered that it is making over-large JPEGs from my M10R DNGs. Sometimes the exported 8-bit JPEG file is bigger than the 16-bit RAW it came from!

JPEGmini Pro can be a plugin in C1 and solves all JPEG problems. I use it all the time when exporting JPEGs.

[Carl Seibert]

It's been a couple of weeks since I asked the two companies for their JPEG parameters. ON1 responded right away but didn't say anything useful. Capture One ghosted me. 

I've been using 99% in C1 and it seems OK. I might go to 98%.

 

[Chris W]

Did you ask on the C1 support forum?

[Advertisement]

Hallo,

Look here: Leica M

[Carl Seibert]

According to Capture One, all of their forums are no more.  Apparently, they backed off their original plan to delete them completely and the old messages are still online, read-only. But you can't post new questions. I do see posts available again in Google. I haven't attempted to post anything.

If you're thinking that's not the behavior of a company that's customer focused, you're not alone.