uhoh7
Offline
• • • •
Upload & Sell: Off
|
JonPB wrote:
Whoa there, my friend. Perhaps period, but not end of paragraph. From the same article:
"The value of the predicted value is subtracted from the current pixel, and that (hopefully small) difference is what is encoded."
In this way, the bits in the file only need to capture which prediction method was used and how the data varied from that model, which will usually be a lesser amount of data (i.e., the data has been compressed) then simply recording how each value varies from zero (aka uncompressed data).
I've never seen anywhere -- in photography, in audio, in computers -- where lossless compression is specified using the casual sense that Jim mentions. Everywhere credible, such as most product specs, "lossless" means lossless. True, lots of bloggers can't discern the difference; this doesn't mean that the specs lie. And it is important to understand how the compression occurs. For example, GIF is a lossless format, but it requires abstracting an image into an indexed data stream (and usually, as a consequence, losing valuable data) before being stored as a GIF; it is a lossy format, but the compression itself is lossless. Sony's lossy compression algorithm appears similar to JPEG's, where blocks of data can have distinct artifacts; this is visibly lossy, but is often a good compromise between data volume and data quality. Leica's compression in the M9 just seems to just bin 16-bit data into 8 bits; this is the least sophisticated, the least effective from a data compression perspective, and also -- to my eyes -- the least objectionable from a visual standpoint.
So, details matter. And you're absolutely right to call for evidence that what is claimed is actually provided. But let's not go too far and question the very nature of "lossless compression." That compression can be actually lossless and not just predictive estimation has been established beyond question. The real question is where and how such an indisputable mathematical model is employed in the intricate and oft unanticipated complexities of real-world data management.
Cheers,
Jon...Show more →
You are making my point.
https://en.wikipedia.org/wiki/Lossless_compression
Claiming Lossless is one thing, proving it another. Note the many benchmarks, and the continued controversy.
"Lossless data compression algorithms cannot guarantee compression for all input data sets. In other words, for any lossless data compression algorithm, there will be an input data set that does not get smaller when processed by the algorithm, and for any lossless data compression algorithm that makes at least one file smaller, there will be at least one file that it makes larger. This is easily proven with elementary mathematics..."
In fact as the article shows many methods are employed according to the form of original data.
What is the method used in Leica DNGs and what benchmarks have proved it?
What are the variables on the other end? IE your editor.
The idea that "lossless" is a magic term employed only when it's true is a sign you don't have a newer Volkswagen. 
However, it's totally possible there is no real world difference in the compressed DNGs M10 is making, ever, for editing, no matter the software, vs what would be the case with an uncompressed file. But I'm not going to assume that is the case without some evidence, as apparently almost everyone does.
One might assume that for 2600USD the Sony 70200/2.8 is the equal of Canikon. But maybe not. What I do know is the pliability of RAW files varies wildly in LR. What sliders do fundamentally changes with different cameras and formats. But maybe not in this case. I hope 
|
I think I just misinterpreted what he meant, which I think was that there is only one native sensitivity and everything at higher ISOs is pushed by the camera accordingly. Still wondering what the native ISO of the camera is...