M240 - Out of the camera JPEGs not too sharp??

[CheshireCat]

Advertisement (gone after registration)

Smaller in both. Original DNG is 69MB. Import to LR3, tweak image, export at 100% quality setting as JPEG, image is now only 51.6MB.

 

Even if you set the quality to 100%, LR is going to use a lossy JPEG transform.

And if information is lost, quality is actually less than 100%, whatever "quality" means, which is not clear at all.

 

You may also want to read this:

 

Jeffrey Friedl's Blog

[NZDavid]

Thanks CheshireCat, looks like an interesting link, I shall study it in detail. I know JPEGs are lossy but didn't think they should be that lossy. If I adjust and save in another program size remains the same.

 

I keep looking for the grin!

[CheshireCat]

I know JPEGs are lossy but didn't think they should be that lossy. If I adjust and save in another program size remains the same.

 

JPEG is really a complicated beast. The "quality setting" is just a way for applications to make it more user friendly. But it is not known how different applications adjust the several different JPEG compression parameters based on this setting.

 

Some of the comments in that link mention chroma subsampling.

Chroma subsampling means that at some lower quality settings, the color resolution is reduced to one fourth the one of the original image. E.g. a 1000x1000 image is compressed as a 1000x1000 b&w (luma) image + 500x500 color (chroma) data.

 

This is also interesting:

A Higher Quality Setting in Photoshop Sometimes Reduces JPEG Quality

[pico]

Smaller in both. Original DNG is 69MB. Import to LR3, tweak image, export at 100% quality setting as JPEG, image is now only 51.6MB. (Original is 48" x 72 " @ 72dpi, resized 41" x 61" @ 72dpi. Still plenty big enough but I'm curious. We print newspapers at 200dpi, magazines are 300dpi.

 

 

It is best to describe an image in pixel dimensions. Leave DPI, PPI, and so-forth out of it.

 

Exporting a DNG to JPEG drops the embedded JPEG which accounts for a bit of the smaller file size. Another caveat when looking at file size concerns the 'cluster factor' boundary of the storage device.

 

In all we have not progressed concerning the issue.

[CheshireCat]

Exporting a DNG to JPEG drops the embedded JPEG which accounts for a bit of the smaller file size.

 

Yes, but not only

 

A DNG contains raw sensor data. This is usually 16-bit linear RGGB Bayer matrix data compressed using a lossless algorithm, very different than the lossy JPEG we are talking about.

 

Due to the Bayer matrix, a 1000x1000 raw sensor image is composed of (sensels, not pixels):

- 250k red sensels

- 250k blue sensels

- 500k green sensels

A total of 1M sensels. And at 16 bits per sensel this makes 2 MBytes uncompressed.

Typical DNG lossless compression is 2:1, therefore the expected final image size is 1 MBytes.

 

The JPEG algorithm works on the final rendered image, after Bayer demosaic and gamma compression of the linear data. Let's see what happens.

 

First, the Bayer demosaic algorithm "invents" missing sensels by means of interpolation. In other words, each R, G or B sensel will become a full RGB pixel. Information here is "created", as the extra two values for each sensel are fake.

Even if fake, this triples the amount of data of the original Bayer raw sensor image, and we now have a 6 MBytes uncompressed linear image.

 

Then, the image is gamma encoded, transforming 16-bits linear data to 8-bit data. Information is obviously lost, as 8 bits are less than 16. This will severely limit your tone mapping options in the future. Regardless of the loss of information, we now have three times the number of color elements (2/3 of which is fake), and a 3 MBytes of uncompressed gamma-encoded image that can finally be displayed by your monitor.

 

Now, we run lossy JPEG compression on the gamma-encoded image. The final result depends on the actual "quality" setting, but -- like I said -- even if you set the JPEG quality to 100%, the JPEG is going to discard information from your image. The loss is bigger than you think, as the JPEG is working on the fake image that contains three times the number of color elements than what was actually captured by your sensor.

 

If the JPEG quality setting compresses at a rate of 3:1, now you have a 1 MB image, which is comparable to the 1 MB of the original DNG... but a lot of your data is gone forever

 

This is why you should only use JPEG for the following reasons:

- Compatibility with programs that can only open JPEG.

- Need for a much smaller file size (at the cost of quality loss). In this case, you will definitely set the quality to much less than 100%.

 

Hope this clarifies.

[algrove]

I bought my M240 to shoot RAW.

[farnz]

Advertisement (gone after registration)

I bought my M240 to shoot RAW.

It's getting a bit cold for that here now, Lou.

 

Pete.

[Rich Paul]

I shoot nothing but jpeg with my M240. They are incredibly sharp with extremely accurate color rendering. I also own other full frame digital cameras including a Sony RX1R. The jpegs out of the M are equal if not better.

 

I tried shooting in RAW but found I just don't enjoy sitting at my computer fooling around with the image.

 

Also, I saw no appreciable increase in sharpness over jpeg.

 

RAW is necessary if there's a tremendous difference in expose within the frame and one wants to achieve proper balance. But to tell you the truth, in the plus 2000 photos taken with the M, I had two I wish I had shot in RAW.

 

For those who ask, why in the world would you spend that much just to shoot jpeg, I answer with because it works. If it didn't, I'd shoot in RAW.

 

 

 

All;

 

I've been using Leica digital cameras on and off since the original M8. Back in the day, I recall that the out of the camera jpegs on my M8 or M9 were somewhat 'soft' and the general consensus was "if you wanted sharp images, shoot Raw". Nonetheless, for the purpose of viewing on the PC, they were ok.

 

So, now I have an M240, and have heard that the out of the camera jpegs are quite a bit better that the M8/M9, but to me, they are still a bit 'soft'. This occurs on both of my bodies with pretty much every lens that I have. I tweaked the sharpness from 'Standard' to Medium High' and it makes 'some' difference, but not a ton. When I had my Fuji X100, the out of the camera jpegs were fantastic. So much so, I pretty much never felt that I needed to shoot Raw. Yes, Fuji's are known for great out of the camera jpegs, and they are two entirely-different systems - agreed.

 

So, I have a question...Am I seeing what I'm supposed to be seeing? I mean, I've been told by my Leica dealer that Leica doesn't over-sharpen or over-saturate anything. They tend to stay neutral. Is that right? Do they leave it to the photographer-artist to correct this in post processing?

 

Any comments are appreciated and welcomed.

 

Thanks!

 

Brad

[jaapv]

An image is more than “sharpness”, whatever that may be...

[Berlinman]

With a simple preset You can process Your pictures in LR in 19 seconds if a standard is fine for You like the in-camera jpegs. But if You want to do more like sharpening for printing, BW-conversion etc You have more posibilities.

[CheshireCat]

With a simple preset You can process Your pictures in LR in 19 seconds if a standard is fine for You like the in-camera jpegs. But if You want to do more like sharpening for printing, BW-conversion etc You have more posibilities.

 

Indeed, Lightroom can export JPEG files from RAW in a few seconds.

 

And the RAW original sensor data is retained, containing a much richer image that we will be able to enjoy on future high dynamic range and wide color gamut displays.

 

Shooting JPEG is not really forward-looking.

[indergaard]

With the possibility to photograph with jpeg+raw without basically any side-effects in todays cameras, I don't see why anyone would photograph only in jpeg. That's like photographing with film, having the film developed in a regular cheap lab and throwing the film negatives away.

 

Just photograph in jpeg+raw, import both when you're done and use the jpeg's and archive the raw's in case you need them later. Disk space costs basically nothing today. You can get an 1-2TB external USB3 HDD for almost the same cost as 10 rolls of 120 medium format film (excluding development and scanning)! So I do not understand or accept the "storage space issue" that many people seem to complain about.

 

And if you want to shoot jpeg only there are far better and cheaper cameras out there that produce a lot better jpeg's... So save some money. Get a Fuji X100S or X100T with the wide and tele converters for about $1200-1500 in total, and produce some of the best out-of-camera jpeg's on the market. With the converters you'll run around with a great jpeg producing machine with a 28mm f/2, 35mm f/2 and 50mm f/2 equivalent lens for less than 10% of what a comparable Leica kit costs. And you don't have to worry about sensor dust either. Just saying...

[CheshireCat]

With the possibility to photograph with jpeg+raw without basically any side-effects in todays cameras [...]

 

Just keep in mind the buffer will flush to SD slower than with JPEG only, therefore less burst shots and more time to wait for the camera to "do its own things".

Not that the average Leica user shoots bursts like I do (or, better, like I wish I could)