M 240 compressed/uncompressed and in-camera processing

[jdlaing]

Advertisement (gone after registration)

Pray tell, what is it then if not down sampling?

 

Image Downsampling

 

I believe the lossless compression reduces the "fat" in the file without reducing resolution.

[01af]

Will you now shoot compressed on your Ms?

Probably not.

Which is not a very clever decision.

 

 

Jeff Schewe suggests that downsampling reduces the original resolution ...

That's what it does indeed. After all, that's what the actual purpose of downsampling is.

 

 

... although he does not use the word lossless compression so far ...

Of course he doesn't, because downsampling and lossless compression are two totally different things. Unlike the former, the latter does not reduce the original resolution.

 

 

... and I intend to find out his views on this file type before contemplating a change.

His views on this file type are exactly the same as everybody else's: lossless compression is lossless. As simple as that.

 

 

From 48 MB to 24 MB must have some effect on images.

Yes, sure it has: the file will use up less memory space and hence, will write and copy faster.

 

 

Image Downsampling

Uh oh. This is a totally different thing again.

[Advertisement]

The Leica Q3 is available for pre-order at B&H Photo Video and Adorama

[jaapv]

I think Leica's decision to include uncompressed is a wise one showing insight in the psychology of their customer base.

[Jeff S]

All in the terminology....different discussion here if the terms were 'efficient' and 'inefficient.'

 

Jeff

[01af]

All in the terminology ... different discussion here if the terms were 'efficient' and 'inefficient.'

If there's any sense in this remark then I don't get it.

[Jeff S]

This was more a commentary on language than it was on the technical accuracy or merits of the concept. Simply put, I suspect people think compression is bad, just because of the word itself.

 

In effect, compression provides more efficient use of memory space, while uncompressed does not. But the term compressed seems to be a bad thing; after all, who wants to lose something, even if they don't know what's being lost? Conversely, if the term uncompressed were instead considered inefficient (less efficient use of memory), then I suspect most people would not think it was such a good thing.

 

Jeff

Edited September 18, 2013 by Jeff S

[stickan1]

Advertisement (gone after registration)

The amount of lossless compression that is possible depends on the information content of the image. If you make a shot without removing the lens cap you will have minimal information content. If you use compression the file size on a M240 is 3,7 Mb if you remove the compression the file size is 48 Mb. But the amount of information is exactly the same in both images.

 

All the algorithms for compression search for patterns in the data and if you have a lot of pixels that are exactly the same this info can be stored by a few numbers, basically the first and the last pixel that are on the same level + the values for these pixels. You do not have to store data for every pixel.

 

I you were able to find a subject matter that is totally chaotic, i.e. white noise, you would have the same size for the compressed and the uncompressed file.

 

If you want to verify check the file size, but do it on the card, because if you import them into Lightroom this program will also do a lossless compression. You may be able to turn this feature off, but I have mine on and was first surprised when I made the test.

 

The reason why Leica kept the lossless option is most likely that they know that they have many very conservative customers.

[01af]

... I suspect people think compression is bad, just because of the word itself.

 

In effect, compression provides more efficient use of memory space, while uncompressed does not.

Ah, now I see how you meant it—and I agree.

[pico]

It seems that people are applying their idea of compression as applied to general data and not necessarily what Leica is doing because Leica is rendering for human vision.

 

Begin with a question: Why does the Monochrom have no option to compress? Could it be because compression reduces brightness range which would be unwise in B&W, but make no significant difference in color.

[Thomas Chen]

I think Hr. Michael Hussmann is capable of explicating the difference between the color-depth- bits-compression and the image file compression such as JPEG compression or ZIP compression.

[sandymc]

It seems that people are applying their idea of compression as applied to general data and not necessarily what Leica is doing because Leica is rendering for human vision.

 

Begin with a question: Why does the Monochrom have no option to compress? Could it be because compression reduces brightness range which would be unwise in B&W, but make no significant difference in color.

 

I wrote about this back in 2012: ChromaSoft: The Leica M Monochrom's lack of DNG compression options

 

Note however that the blog post refers to the M8/M9's lossy compression, not lossless as discussed in this thread. There is no reason why the Monochrom could not have lossless compression other than for the same reason as the M8/M9 - probably too little CPU power.

 

Sandy

[algrove]

All the algorithms for compression search for patterns in the data and if you have a lot of pixels that are exactly the same this info can be stored by a few numbers, basically the first and the last pixel that are on the same level + the values for these pixels. You do not have to store data for every pixel.

 

)

 

I have had many explain this to me before. It makes perfect sense. But how come when I look at an image over and over in always different scenes, why is the compressed half the size of the uncompressed-always?

 

The scenes are very different and some have repeating looking pixels and some do not, but again the files sizes are quite uniform uncompressed 48-48.3, compressed about 24. Why do I never see a compressed file of 24.8 or 25.3 or more?

[MirekE]

I have had many explain this to me before. It makes perfect sense. But how come when I look at an image over and over in always different scenes, why is the compressed half the size of the uncompressed-always?

 

The scenes are very different and some have repeating looking pixels and some do not, but again the files sizes are quite uniform uncompressed 48-48.3, compressed about 24. Why do I never see a compressed file of 24.8 or 25.3 or more?

 

I have no idea how the compression is done, but there is 1/4 of redundant data in each file (2 greens) and most likely all values are written as 16 bit numbers, but they occupy only 14 bit or less if you underexpose. So if you are able to address that by your compression algorithm, you can get to some 65% of the original size or less before you actually start compressing the image itself. So, if my logic is correct, 50% sounds about right. But it should not be the same number every time. It should fluctuate based on content and possibly ISO.

[stickan1]

There is a variation, but it is less than what I expected. On M240 I have from 28 to 20 Mb, I checked my D800E files also and the range was 46 to 32 Mb. And as I wrote in my first post I made an exposure with the lens cap on and it was 3,7 Mb. And the variation depends on the scene as expected. Look at an images with large out of focus areas and then a typical landscape or a forest where everything is sharp.

 

Anyway I have made my choice to always use compressed DNG as I trust that the algorithms are ok and I will not loose any details in my images.

 

If you look at old files in Lightroom remember that normally LR also use a lossless algorithm for DNG files.

[jaapv]

I have had many explain this to me before. It makes perfect sense. But how come when I look at an image over and over in always different scenes, why is the compressed half the size of the uncompressed-always?

 

The scenes are very different and some have repeating looking pixels and some do not, but again the files sizes are quite uniform uncompressed 48-48.3, compressed about 24. Why do I never see a compressed file of 24.8 or 25.3 or more?

On the M8 and M9 you mean? Because the values are written as pointers to a LUT.

[flyalf]

Yes, sure it has: the file will use up less memory space and hence, will write and copy faster.

 

I totally agree on your main points, but am a bit unsure about this. Of course plain copy of bytes will be faster, but everything else will be slower since the file needs to be de-compressed for any action? Basically you trade less storage volume against more need of processing power, be it in camera or PC.

[jaapv]

It does not need to be decompressed for copy or transfer; only for actions that actually need to use the file.

[flyalf]

It does not need to be decompressed for copy or transfer; only for actions that actually need to use the file.

 

Exactly. Good to see that you understood what I meant by "everything else" .

[algrove]

If you look at old files in Lightroom remember that normally LR also use a lossless algorithm for DNG files.

 

Not exactly correct. You had the choice to use lossy compression for DNG files under old LR variations LR4 included.

[Advertisement]

Hallo,

Look here: Leica M