M9 Import - Copy DNG from camera or Copy as DNG

[k-hawinkler]

Advertisement (gone after registration)

Having developed and written a lossless compression algorithm myself from scratch - in fact for images - I would explain at a high level use of compression as follows: To some degree with computer systems one can substitute a scarce resource with a more plentiful quantity. For example, if I have limited transmission bandwidth or storage capacity, I can use abundant computational power to exploit repetitiveness in the data and produce smaller data sets for transmission or storage. Using lossless compression algorithms then permits the reconstruction of the original data set for further processing. Typically in an unbalanced computer, camera, you name it system this approach can be more efficient than a straight forward approach.

 

Best, K-H.

[Advertisement]

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

[rfunnell]

Thanks for the help on that one.

Ross

[hoppyman]

I use Lightroom 3.3 but kept my Capture One partly up to date.

 

CaptureOne 4.8.3 has no difficulty with the DNG1.3 standard losslessly compressed files produced by LR3.3.

 

It is worth noting that the DNG standard is not entirely prescriptive, which means that companies implementing it are not obliged to comply with every part of the specification. That was never intended. So Capture One did not strictly have to update, but of course it was in their interest to do so and it didn't take them long to implement DNG1.2 . Remember of course that the standard is openly documented and free to implement. There's some Leica Camera/Phase One politics in the background but that is a different subject.

 

Incidentally, Leica Camera does not currently use every part of the specification, notably the new (1.3) OP Codes. Ask Sandy about opportunities there

 

Here's the spec for anyone interested in understanding more.

http://www.adobe.com/products/dng/pdfs/dng_spec.pdf

 

 

.....As you know, Capture one was delivered with the M8 models, and could read Leica DNGs but not Adobe-compressed DNGs. That's a problem that shouldn't have occurred, since both products were built under the same standard. There were changes to the standard after that time, as you say.........

[ho_co]

Great, Geoff! Helpful information!

 

That's what I figured should be the case, but I'm glad to have it confirmed practically.

 

 

So, Ross, ignore my caution: I don't know any reason not to "copy as DNG."

[01af]

At least one aspect of the size reduction would therefore be the elimination of the LUT, and replacing the index values in the main matrix with the actual values they had pointed to.

Huh!?

 

As the purpose of the LUT is to reduce the data, eliminating it would increase the file size.

[ho_co]

Hmm. You're over-simplifying, aren't you? IIRC, the LUT was introduced in the M8 to enable storage of 16-bit data with 8-bit values.

 

It seems to me that if you start with two matrices and eliminate one, you'll save space.

 

Taking the M8's compression as an example: As I understand it, there's a matrix or table of 256 values, and another matrix of pointers pointing to various ones of those 256 values.

 

Wouldn't you save space by just writing one of those 256 values into each location that currently holds a pointer, and eliminating the first matrix?

 

 

As I understand what K-H said in post #21 above, the compression starts when you can group identical sets of data for recall via a pointer. That's not how the M8's LUT works.

[sandymc]

Advertisement (gone after registration)

Hmm. You're over-simplifying, aren't you? IIRC, the LUT was introduced in the M8 to enable storage of 16-bit data with 8-bit values.

 

It seems to me that if you start with two matrices and eliminate one, you'll save space.

 

Taking the M8's compression as an example: As I understand it, there's a matrix or table of 256 values, and another matrix of pointers pointing to various ones of those 256 values.

 

Wouldn't you save space by just writing one of those 256 values into each location that currently holds a pointer, and eliminating the first matrix?

 

 

As I understand what K-H said in post #21 above, the compression starts when you can group identical sets of data for recall via a pointer. That's not how the M8's LUT works.

 

A limitation of the DNG format is that that the only way to represent a gamma encoding (which is what the M8/M9 compression scheme is), is to use a LUT. The choice is either to use a LUT, or use 16-bit data. The LUT gives a smaller file size. If Leica had built their own raw file format, different answer.

 

Sandy

[ho_co]

While you smart guys are here, and while we're on the subject of DNG’s, here’s a totally OT question:

 

DNG Recover Edges recovers extra pixels at the edges of a DNG—with M8 & M9, very few extra pixels; with D-Lux 4 more; more yet with Canon and Nikon.

 

Why are they there? Why didn’t the manufacturer use them? (Since they contain recoverable image data, I don't think they'd be black-level encoding pixels?)

 

The only thing I could come up with was that they might be there to match the recorded image to what the viewfinder sees, giving the manufacturer some extra sensor space to more around in.

 

Any ideas?

 

Thanks.

[hoppyman]

Page 42 of the DNG specification that I linked to may help.

[pico]

While you smart guys are here, and while we're on the subject of DNG’s, here’s a totally OT question:

 

DNG Recover Edges recovers extra pixels at the edges of a DNG—with M8 & M9, very few extra pixels; with D-Lux 4 more; more yet with Canon and Nikon.

 

Why are they there? Why didn’t the manufacturer use them? (Since they contain recoverable image data, I don't think they'd be black-level encoding pixels?)

 

The only thing I could come up with was that they might be there to match the recorded image to what the viewfinder sees, giving the manufacturer some extra sensor space to more around in.

 

Any ideas?

 

Thanks.

 

Perhaps to make the format dimensions exactly correct.

[ho_co]

Page 42 of the DNG specification that I linked to may help.

I'd made it to p 23 so far; working my way toward 42. Thanks, Geoff!

 

 

Since these are all files converted to DNG, maybe the question should be different:

 

Is anyone here who has a camera other than the M8/M9 that also shoots DNG?

 

If so, does DNG Recover Edges also recover only 2 rows/columns on each of the four sides for your camera?

 

Maybe the question goes to the design of each manufacturer's Raw file and why DNG Converter doesn't convert all the data.

 

 

 

 

Perhaps to make the format dimensions exactly correct.

 

Interesting thought that brings another question to mind: Why do some cameras produce JPGs of different pixel size from their Raw images?

 

But now we're getting way off topic.

[FredSF]

I'm not sure this pixel edge recovery business is of the utmost importance. As in the darkroom days, anything close enough to edges may or may not go away... I have yet to read this DNG report, tho'.