The case for Adobe Digital Negative Converter - page 2 - RAW Shooting and Processing - CHDK Forum

The case for Adobe Digital Negative Converter

  • 23 Replies
  • 12924 Views
Re: The case for Adobe Digital Negative Converter
« Reply #10 on: 21 / July / 2012, 19:54:56 »
Advertisements
Nothing has changed, and the discrepancies I noted earlier still remain.  ...  This particular issue, where DNG files from one model differ significantly from those from another model using the same DNG version, seems a bit more of a challenge to figure out.
Sorry .. I guess I'm a bit slow. Would you mind providing a more specific list of what you see different between the .dng files you get from one of your cameras and the .dng files from your other camera when you use the most recent version of CHDK on both cameras ?

Thanks
« Last Edit: 21 / July / 2012, 19:57:36 by waterwingz »
Ported :   A1200    SD940   G10    Powershot N    G16

Re: The case for Adobe Digital Negative Converter
« Reply #11 on: 21 / July / 2012, 20:08:03 »
Sure, here are some principal differences:

DNG files from an SX130 will display a 3MP preview image in Faststone, and open without difficulty in Silkypix.

DNG files from an S3 will display a 128x96 pixel preview image (with magenta cast) and not open in Silkypix.

Re: The case for Adobe Digital Negative Converter
« Reply #12 on: 21 / July / 2012, 20:21:51 »
DNG files from an SX130 will display a 3MP preview image in Faststone, and open without difficulty in Silkypix.
DNG files from an S3 will display a 128x96 pixel preview image (with magenta cast) and not open in Silkypix.
So the DNG implementation for the sx130 is good and the implementation for the S3 might be broken ?
Ported :   A1200    SD940   G10    Powershot N    G16

*

Offline philmoz

  • *****
  • 3450
    • Photos
Re: The case for Adobe Digital Negative Converter
« Reply #13 on: 21 / July / 2012, 20:49:02 »
Thank you gentlemen for your suggestions and observations.  I apologize for the delay in responding. 

I've uploaded in-camera DNG samples from an SX130 and S3:

Canon S3 - https://www.box.com/s/66fa69ca53f53f2f1538
Canon SX130 - https://www.box.com/s/786f909da63f6509427e

They are definitely not the same (variety of DNG).  I am using CHDK builds from March (SX130) and April (S3) and they do not offer an option to choose the version of DNG you wish to save.  Curiously, Adobe DNG Converter 7.1 offers three selectable options:  1.1, 1.3 and 1.4, but no 1.2!

The very popular Faststone Image Viewer (which I personally use) displays a 3MP preview for SX130 DNG files, and a 128x96 pixel preview (with magenta color cast) for S3 files.  I can't explain that, can you?  I have Faststone set to display embedded preview.

I also use Silkypix to process DNG files directly from the SX130.  Silkypix will not accept S3 DNG files until they have been converted by Adobe DNG Converter.  I have the converter compatibility set to Camera Raw 7.1 and later, so I would presume I am saving as DNG version 1.4.

Click the links for original in-camera DNG samples.  Attached here are a few screenshots to demonstrate just a few of the compatibility issues I've encountered.  Thank you for any additional suggestions, especially - how does one choose the DNG version in CHDK (as mentioned)?

I'm afraid the problem is with the Faststone & SilkyPix software not the DNG files.

Both of the files you uploaded pass the Adobe 'dng_validate' test program without any significant errors.
Both can be loaded without problems by Lightroom 4.1, Photoshop CS5, and RawTherapee 4.0.

Phil.
CHDK ports:
  sx30is (1.00c, 1.00h, 1.00l, 1.00n & 1.00p)
  g12 (1.00c, 1.00e, 1.00f & 1.00g)
  sx130is (1.01d & 1.01f)
  ixus310hs (1.00a & 1.01a)
  sx40hs (1.00d, 1.00g & 1.00i)
  g1x (1.00e, 1.00f & 1.00g)
  g5x (1.00c, 1.01a, 1.01b)
  g7x2 (1.01a, 1.01b, 1.10b)

Re: The case for Adobe Digital Negative Converter
« Reply #14 on: 21 / July / 2012, 20:52:10 »
Thanks for performing such a thorough test!  Would you not concede though that there must be some differences between these two DNG files, in that they behave differently with the same software?

*

Offline reyalp

  • ******
  • 14125
Re: The case for Adobe Digital Negative Converter
« Reply #15 on: 21 / July / 2012, 20:54:41 »
Thanks for posting the samples.
Some observations:
Both files open correctly in irfanview.
The sx130 image is from CHDK 1.0 build 1756
The s3 image is from CHDK 1.0 build 1825

The sx130 has the camera datetime garbled. This is a known issue which was fixed in 1806.
There were several other DNG changes in this period, including 1803 which involved the thumbnail code.

The sx130 also produces the following in dng_validate
*** Warning: Too little padding on left edge of CFA image (possible interpolation artifacts) ***
and similar for other edges.
it's possible that some image programs would be confused by this, but I would think this shouldn't be a fatal error.

Since the build you are using from sx130 had known issues with DNG, it would be nice to see a sample from a current version.

edit:
Regarding the edge warning, sx130 has
Code: [Select]
DefaultCropOrigin: H = 0.00 V = 0.00
DefaultCropSize: H = 4000.00 V = 3000.00
ActiveArea: T = 20 L = 36 B = 3020 R = 4036
which means that the default crop exactly matches the active area. I don't the the DNG spec prohibits this, but generally in CHDK, we make the default crop match the jpeg size, which is smaller than the active area.
« Last Edit: 21 / July / 2012, 21:15:08 by reyalp »
Don't forget what the H stands for.

*

Offline philmoz

  • *****
  • 3450
    • Photos
Re: The case for Adobe Digital Negative Converter
« Reply #16 on: 21 / July / 2012, 21:16:38 »
Thanks for performing such a thorough test!  Would you not concede though that there must be some differences between these two DNG files, in that they behave differently with the same software?

Apart from the RAW image size, the only significant difference, as far as I can tell, is the S3 is an old camera and has a 10 bit per pixel sensor. The SX130 has a 12 bit bit per pixel sensor. I suspect the software is not able to handle the 10 bpp RAW data.

Phil.
CHDK ports:
  sx30is (1.00c, 1.00h, 1.00l, 1.00n & 1.00p)
  g12 (1.00c, 1.00e, 1.00f & 1.00g)
  sx130is (1.01d & 1.01f)
  ixus310hs (1.00a & 1.01a)
  sx40hs (1.00d, 1.00g & 1.00i)
  g1x (1.00e, 1.00f & 1.00g)
  g5x (1.00c, 1.01a, 1.01b)
  g7x2 (1.01a, 1.01b, 1.10b)

Re: The case for Adobe Digital Negative Converter
« Reply #17 on: 21 / July / 2012, 21:18:53 »
Thank you very much for testing these files.  What I'm most concerned about are my S3 DNG files which won't load into Silkypix without first converting them with Adobe DNG Converter.

Here are samples from today's build of CHDK.  Would you mind checking them, especially the S3 sample?
Many thanks, and please excuse their terrible, out of focus content!

SX130 - from today's build - https://www.box.com/s/67010d18b12f8bd916d9

S3 - from today's build - https://www.box.com/s/abcfc875bdea9e921f5b

I post this for "reyalp", as he requested a sample from today's build.
« Last Edit: 21 / July / 2012, 21:30:00 by Mike Lee »

*

Offline reyalp

  • ******
  • 14125
Re: The case for Adobe Digital Negative Converter
« Reply #18 on: 21 / July / 2012, 22:06:55 »
I suspect the software is not able to handle the 10 bpp RAW data.
I'm inclined to agree with this, since it's the s3 that was problematic, and its the one that validates perfectly.

In the silkypix release notes http://www2.isl.co.jp/SILKYPIX/english/download/release_dsp5win.html
There are some limitations to develop the DNG. Please refer to "10.4.5.2 DNG Files Available with this Software" of the manual. I downloaded the 30 day trial of silkypix 5.0.18.0, but the manual isn't very clear and doesn't seem to mention bit depth at all.

This program fails to open DNG from my a540 (10 bit) and succeeds opening DNG from d10 (12 bit). It wouldn't be surprising if they don't support 10 bit, I doubt many (non-CHDK) cameras produce 10 bit raws.

edit:
Thanks for the additional samples. The sx130 still has the padding warning, I don't see anything else amiss. I hexed on of the sx130 images to make the active area full frame, and it looks like the active area could be made a couple pixels bigger (around 32,12,4056,3036) , but I guess this isn't the problem camera so maybe it doesn't matter.
« Last Edit: 21 / July / 2012, 22:16:29 by reyalp »
Don't forget what the H stands for.

Re: The case for Adobe Digital Negative Converter
« Reply #19 on: 21 / July / 2012, 22:41:15 »
No, I've never had any problems whatsoever with the SX130 files.  They handle and process perfectly.

I appreciate the extra effort you went to to try Silkypix, and your your detective work on the S3 files seems sound, but had we considered that once I run them through Adobe DNG Converter at its default settings (which I've confirmed generates v1.3 files) they do open just fine in Silkypix.  Adobe isn't converting any bit depth, just repackaging (and compressing) things.  Following this line of reasoning the 10bit vs. 12 bit rationale falls apart.  It must be something else.  What do you think about this?

 

Related Topics


SimplePortal © 2008-2014, SimplePortal