M240 compressed DNG

[algrove]

Advertisement (gone after registration)

OK, I will start compressing, but if I loose anything in next 15 years you are all to blame, starting with RickLeica!

[Advertisement]

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

[lct]

Would be interesting to study the effect of compression on digital noise and banding though. Wonder if the good techies of this forum will still sing the laurels with certainty if Leica change the compression ratio in the next firmware update BTW. Why would they do that if there is no difference between compressed and uncompressed files mmm? Pure imagination from my part but i would not be the the last to laugh then you can trust me.

[mblaze]

Would be interesting to study the effect of compression on digital noise and banding though. Wonder if the good techies of this forum will still sing the laurels with certainty if Leica change the compression ratio in the next firmware update BTW. Why would they do that if there is no difference between compressed and uncompressed files mmm? Pure imagination from my part but i would not be the the last to laugh then you can trust me.

 

If it is lossless compression, there will be no difference whatsoever in any aspect of image quality. That's not merely some expert's opinion about lossless encoding, that's the *definition* of lossless encoding. If it is a lossy encoding, of course, then there may well be differences. But the the claim is that Leica M DNGs are compressed with a lossless encoding.

 

This is not to say there aren't other tradeoffs. With more computation or with more clever algorithms, for example, it may be possible to losslessly compress images more efficiently than the current scheme does. Lecia could, conceivably, in a future firmware release switch to another encoding scheme that achieves greater compression (probably at the expense of additional computation) but that would still be lossless.

 

I'm not sure why anyone would want to pretend they live in a world where extremely well understood mathematical principles (in this case, those of information theory) don't apply, but I suppose that's your prerogative if it makes you happy. Personally, I find that there are more than enough subtle effects in photography to keep things interesting without the need to make up stuff.

[lct]

He he when i was a teacher at university i used to tell my students that all scientists should be modest as new discoveries will prove that they were wrong soon or late. This made my students laugh a bit but not my colleagues to my (lack of) surprise i much say.

[mblaze]

He he when i was a teacher at university i used to tell my students that all scientists should be modest as new discoveries will prove that they were wrong soon or late. This made my students laugh a bit but not my colleagues to my (lack of) surprise i much say.

 

That is certainly true. However, definitions of technical terms are not "scientific discoveries".

[algrove]

If it is lossless compression, there will be no difference whatsoever in any aspect of image quality. That's not merely some expert's opinion about lossless encoding, that's the *definition* of lossless encoding. If it is a lossy encoding, of course, then there may well be differences. But the the claim is that Leica M DNGs are compressed with a lossless encoding.

 

This is not to say there aren't other tradeoffs. With more computation or with more clever algorithms, for example, it may be possible to losslessly compress images more efficiently than the current scheme does. Lecia could, conceivably, in a future firmware release switch to another encoding scheme that achieves greater compression (probably at the expense of additional computation) but that would still be lossless.

 

I'm not sure why anyone would want to pretend they live in a world where extremely well understood mathematical principles (in this case, those of information theory) don't apply, but I suppose that's your prerogative if it makes you happy. Personally, I find that there are more than enough subtle effects in photography to keep things interesting without the need to make up stuff.

 

Now you have me worried again. "More computation and more algorithms" Could this ever change the "definition" of lossless compression?

[Stef63]

Advertisement (gone after registration)

Now you have me worried again. "More computation and more algorithms" Could this ever change the "definition" of lossless compression?

 

Lossless compression by definition means a compression technique which guarantees you can decompress the compressed information back to its 100% original.

 

The compression technique is however not one single predefined way of getting from point a to point b. It's mostly a tradeoff between storagespace you gain and cpu power you need (to do the compression / decompression step in getting from a to b and back).

 

So the availability of higher computation power and better algorithms can indeed lead to more compressed files. But again if the compression is defined as lossless, one can always decompress back to the original and this by definition.

[mblaze]

Now you have me worried again. "More computation and more algorithms" Could this ever change the "definition" of lossless compression?

 

No, it won't change the definition of losslessness, but it might new mean that new files it produces will be incompatible with software for the previous scheme.

 

The property that makes a compression scheme lossless is that you put bits (your sensor output) into the encoder, producing data that you put into a corresponding decoder. If the output of the decoder is identical, bit for bit, to the input to the encoder, the scheme is lossless. However, you still have to make sure you have a decoder that corresponds to the encoder that was used. (In practical terms, that means that, lossless or not, there's no guarantee that future camera firmware will produce files your current image processing software can make sense of, or that all future image processing software will continue to support your current file formats. That only happens if someone cares enough to maintain the formats. But that has nothing to do with whether the encoding is lossless or lossy - orphaned file formats cause another kind of loss entirely.)

[Peter Branch]

I sense more differences between AIFF and MOV files than between TIF and 100% JPEG. Theory and practice as usual.

 

Tiff is by definition lossless because it is uncompressed. It uses a lot of memory.

 

Jpeg is inherently and by design not lossless, no matter if it is so called 100%. It uses a variable amount of memory but always much less than tiff, or any of the other lossless formats.

 

Every time you open a Jpeg file and then save it, rather than closing it, more data is lost.

[lct]

Thank you Peter but i learnt that in the kindergarden... or a bit later.

[01af]

TIFF is by definition lossless because it is uncompressed.

Except, of course, when it's compressed.

 

 

JPEG is inherently and by design not lossless ...

Except, of course, when it's uncompressed.

[pico]

Would be interesting to study the effect of compression on digital noise and banding though. Wonder if the good techies of this forum will still sing the laurels with certainty if Leica change the compression ratio in the next firmware update BTW. Why would they do that if there is no difference between compressed and uncompressed files mmm? Pure imagination from my part but i would not be the the last to laugh then you can trust me.

 

DNG is open source developed by Adobe. It is supposed to be an archival standard. If Leica's contractor is not following the DNG standard, then we should know right now so we can convert our files to a proper, standard DNG or leave Leica until they get it right.

 

Has Adobe's DNG processor had any trouble with Leica DNG files? If it has, then that suggests that Leica's contractor is at fault.

[pico]

Tiff is by definition lossless because it is uncompressed. It uses a lot of memory.

 

DNG is an extension of TIFF and as 01af suggested, TIFF can be compressed without loss (using the brilliant LZW algorithm.) I don't understand the concern regarding memory because memory is cheap. So is storage.

 

Jpeg is inherently and by design not lossless, no matter if it is so called 100%. It uses a variable amount of memory but always much less than tiff, or any of the other lossless formats.

 

Lossless JPEG compression is somewhat questionable because there is no guarantee that a perfect decompression is possible, but in practice I can find no fault.

 

Every time you open a Jpeg file and then save it, rather than closing it, more data is lost.

 

Not necessarily true.

[Peter Branch]

DNG is an extension of TIFF and as 01af suggested, TIFF can be compressed without loss (using the brilliant LZW algorithm.) I don't understand the concern regarding memory because memory is cheap. So is storage.

 

Indeed it can but then it is not a tiff file, it is a compressed tiff file. The compression can be lossless or not, depends on the algorithm. The issue with memory size has several causes. One is the time it takes to write a file. Another is the ability to send images via the web. This has got better than when jpegs were introduced but remains limited to between 2 and 10 MB. A compressed DNG from the M240 is still ~25 MB so can't in practice be sent by this means.

 

Lossless JPEG compression is somewhat questionable because there is no guarantee that a perfect decompression is possible, but in practice I can find no fault.

 

I assume that everyone would agree that any jpeg file will be smaller than the equivalent image in any of the non-compressed formats, one of which is tiff. That is the whole point of jpeg files. The question therefore is whether the data which has not been stored can be recovered. I believe the answer to be no. The fact that a high quality, i.e. large, jpeg is for many, but not all, practical purposes indistinguishable from, say, a tiff file is why jpeg is so universal. ( As an aside it is interesting to take a tiff file and convert it to jpeg, then to apply sharpening to the tiff image and convert that. A "sharpened file" often looks "better" but it contains less data as evidenced by the reduction in the size of the resulting jpeg.)

 

 

 

Not necessarily true.

 

I believe this has been demonstrated many times. If jpeg algorithms are repeatedly applied to an image then the data is altered at each iteration. Depending on the sensitivity of the subject and the viewing method after about 5 applications the effect can be quite noticeable.

[Peter Branch]

DNG is an extension of TIFF and as 01af suggested, TIFF can be compressed without loss (using the brilliant LZW algorithm.) I don't understand the concern regarding memory because memory is cheap. So is storage.

 

Indeed it can but then it is not a tiff file, it is a compressed tiff file. The compression can be lossless or not, depends on the algorithm. The issue with memory size has several causes. One is the time it takes to write a file. Another is the ability to send images via the web. This has got better than when jpegs were introduced but remains limited to between 2 and 10 MB. A compressed DNG from the M240 is still ~25 MB so can't in practice be sent by this means.

 

Lossless JPEG compression is somewhat questionable because there is no guarantee that a perfect decompression is possible, but in practice I can find no fault.

 

I assume that everyone would agree that any jpeg file will be smaller than the equivalent image in any of the non-compressed formats, one of which is tiff. That is the whole point of jpeg files. The question therefore is whether the data which has not been stored can be recovered. I believe the answer to be no. The fact that a high quality, i.e. large, jpeg is for many, but not all, practical purposes indistinguishable from, say, a tiff file is why jpeg is so universal. ( As an aside it is interesting to take a tiff file and convert it to jpeg, then to apply sharpening to the tiff image and convert that. A "sharpened file" often looks "better" but it contains less data as evidenced by the reduction in the size of the resulting jpeg.)

 

 

 

Not necessarily true.

 

I believe this has been demonstrated many times. If jpeg algorithms are repeatedly applied to an image then the data is altered at each iteration. Depending on the sensitivity of the subject and the viewing method after about 5 applications the effect can be quite noticeable.

[Peter Branch]

DNG is an extension of TIFF and as 01af suggested, TIFF can be compressed without loss (using the brilliant LZW algorithm.) I don't understand the concern regarding memory because memory is cheap. So is storage.

 

Indeed it can but then it is not a tiff file, it is a compressed tiff file. The compression can be lossless or not, depends on the algorithm. The issue with memory size has several causes. One is the time it takes to write a file. Another is the ability to send images via the web. This has got better than when jpegs were introduced but remains limited to between 2 and 10 MB. A compressed DNG from the M240 is still ~25 MB so can't in practice be sent by this means.

 

Lossless JPEG compression is somewhat questionable because there is no guarantee that a perfect decompression is possible, but in practice I can find no fault.

 

I assume that everyone would agree that any jpeg file will be smaller than the equivalent image in any of the non-compressed formats, one of which is tiff. That is the whole point of jpeg files. The question therefore is whether the data which has not been stored can be recovered. I believe the answer to be no. The fact that a high quality, i.e. large, jpeg is for many, but not all, practical purposes indistinguishable from, say, a tiff file is why jpeg is so universal. ( As an aside it is interesting to take a tiff file and convert it to jpeg, then to apply sharpening to the tiff image and convert that. A "sharpened file" often looks "better" but it contains less data as evidenced by the reduction in the size of the resulting jpeg.)

 

Not necessarily true.

 

I believe this has been demonstrated many times. If jpeg algorithms are repeatedly applied to an image then the data is altered at each iteration. Depending on the sensitivity of the subject and the viewing method after about 5 applications the effect can be quite noticeable.

[pico]

IThe issue with memory size has several causes. One is the time it takes to write a file. Another is the ability to send images via the web. This has got better than when jpegs were introduced but remains limited to between 2 and 10 MB.

 

Without looking I presume you are a USA resident. There very many countries that have far faster networks. I think the reason the USA is lagging so far behind is the lack of market completion.

 

 

 

 

 

Sent from my iPhone using Tapatalk - now Free

[mjh]

Tiff is by definition lossless because it is uncompressed. It uses a lot of memory.

The TIF format has lots of variants, some uncompressed, some compressed, either lossy (using JPEG compression, for example) or losslessly (using LZW, runlength encoding or several other methods). A file being in TIF format says nothing whatsoever about whether and how it is compressed.

 

(So does a file being in MOV format, btw.: MOV is merely a container format supporting a variety of codecs. MOV files can also contain uncompressed data.)

 

Jpeg is inherently and by design not lossless, no matter if it is so called 100%.

Strictly speaking it could also be lossless as there are two lossless JPEG variants. Cameras saving JPEG images don’t use these, as a rule, so this may be of little practical relevance, but still: JPEG is not lossy by design.

[Advertisement]

Hallo,

Look here: Leica M

[mjh]

That is certainly true. However, definitions of technical terms are not "scientific discoveries".

Also we are dealing with mathematics here, not physics. There is no such thing as a proof in physics, but there certainly is in mathematics. There isn’t much room for argument here and one should not mistake mere stubbornnness for laudable scepticism.

[pico]

[...]A file being in TIF format says nothing whatsoever about whether and how it is compressed.

 

So true.

 

Sometimes I wish the operating system would show certain tags.

Users of an OS that supports BSD general commands can use tiffutil -i input.file | grep Compression

but that's not what the general market wants to do.