knickerhawk

Lives in United States United States
Joined on Jul 22, 2004

Comments

Total: 20, showing: 1 – 20
On photo Symphony of the wings ! in the A big year - birds 2020 challenge (16 comments in total)

Obviously these are two birds composited from two separate images. Want clear proof? Look here: https://www.dpreview.com/forums/post/64063655

Link | Posted on Jun 21, 2020 at 18:04 UTC as 2nd comment
In reply to:

brn: The headline of this article suggests that blockchains would be used for image compression. The linked article does not.

Yeah, but it at least it got some of us to read the article to figure out just how in the world blockchains could be used to compress images! ;-)

Link | Posted on Feb 26, 2020 at 15:47 UTC
In reply to:

matsonfamily: What's the point, if it's not the raw file that's cryptographically signed for authenticity? JPEGS often are horribly overly bright or dark, to allow for more dynamic range, or to shoot without flash. Not everyone's in a studio, so it's the raw formats that need signatures. This just needs a press foundation to contact the camera manufacturers and ask them to put a hash value [of the camera hardware, date, and smaller image] into the raw image, IMO.

I applaud the attempt, but it should be driven by organization (the press?) with the the need for authenticity, towards the company producing the devices that would have the option to enforce authenticity (the cameras). I'm more of an IT guy than a photographer, so maybe I'm missing something here.

Inclusion of metadata about the raw conversion method (converter, version, settings, etc.) along with the hash of the JPEG image itself should suffice for authentication purposes. In fact, you could use what's known as a Merkle Tree hash to validate a whole bunch of images packaged together.

Link | Posted on Feb 26, 2020 at 15:45 UTC
In reply to:

Horshack: I read through their blockchain whitepaper. The only application of the technology I see being useful for images is mutable digital rights management, where rights of existing images can be added and removed on images in circulation. Everything else proposed in the whitepaper can be accomplished via existing technologies, including fingerprinting, public-key encryption, and digests/signatures.

You need something like IPFS or another peer-to-peer strategy for image storage. Image storage can be tied to a blockchain, but it's not the same thing. Blockchains also don't validate when the image was taken, only when the hash of the image was added to the blockchain. At most, a blockchain can confirm that an image existed before the timestamp of the validated block but how much before is impossible to determine from the blockchain. Blockchains can't tell you anything about the validity of the image contents (image itself or its metadata). And, at most, it establishes who first claimed possession of the source image. It doesn't resolve disputes about off-chain theft/appropriation of the image preceding the on-chain posting of the image hash.

Link | Posted on Feb 26, 2020 at 15:30 UTC
In reply to:

PhotoRotterdam: Blockchain technology is the most energy inefficient technology imaginable. Having the validity of the incredible amounts of jpegs being produced checked against a blockchain will probably smoke our planet within a month. It is simply impossible to implement at such a large scale.

I don’t know who comes up with these inconsiderate plans, but ‘blockchain’ isn’t the magic word anymore, like it was 5 years ago. We’re past the hype.

What quiquae says above is correct. The energy inefficient "proof of work" method of confirming a block is used by Bitcoin and (still) Ethereum and a number of other digital ledgers. However, there's also a growing list of alternative DLTs that don't require proof of work and Ethereum is planning to switch to an alternative method "soon". Moreover, even with a proof of work platform like Bitcoin there are various "level 2" and other strategies for working around the energy and time penalties inherent in proof of work.

And the idea that "millions of computers having to synchronise billions and billions of transactions" is also an oversimplified understanding of how blockchains work (including Bitcoin). It's a totally naive characterization of the impact that image ID validation transactions would have in the grand scheme of modern network computing. One person streaming one movie at typical residential internet connectivity is also "billions and billions of [network] transactions."

Link | Posted on Feb 26, 2020 at 15:14 UTC

See link below for a close comparison of the S1R HiRes vs. the Phase One 100mp with both converted in ACR. It's important to note that (for whatever unspecified reason), the DPR studio comparison is using a Phase One rendering that was done in Capture One, which makes it somewhat of an apples-vs-oranges comparison when it comes to looking at micro detail. Better to use the same converter as I did here:

https://www.dpreview.com/forums/post/62592160

Link | Posted on Apr 24, 2019 at 16:26 UTC as 12th comment | 2 replies

These are the guys who brought us the now defunct PhotoAcute (alas). The super resolution zoom appears to be very similar to what PhotoAcute was doing.

Link | Posted on Feb 4, 2019 at 20:39 UTC as 26th comment | 1 reply
In reply to:

Naveed Akhtar: solid testing ..
really happy to see a more practical sensor testing than your partners Dxo.

Here I can see the test results and decide on the choice on my own, instead of mere numbers and others judgements.

The comment seems to imply that there is no DR/shadow noise advantage to utilizing exposure bracketing if the camera is "ISO invariant." At best, ISO invariance allows you to protect highlights without WORSENING shadow noise, but there's always a DR/shadow noise advantage to be gained from exposure bracketing. This can be seen in the Exposure Latitude studio scene shots for the A7Riii, which is similarly "ISO invariant." Look here: https://www.dpreview.com/reviews/image-comparison/fullscreen?attr144_0=sony_a7riii&attr144_1=sony_a7riii&attr144_2=sony_a7riii&attr144_3=sony_a7riii&attr146_0=100_0&attr146_1=100_2&attr146_2=100_4&attr146_3=100_6&normalization=compare&widget=397&x=0.20542277993042046&y=0.46579417945980534

Link | Posted on Mar 21, 2018 at 16:23 UTC
In reply to:

Tony Northrup: Crop factor was never about getting a proper exposure/brightness. In our first crop factor video, the very first thing we do is demonstrate that the same settings yield the same exposure/brightness regardless of sensor size. We repeatedly reinforced this fact in later videos.

Yet, there's this vocal group that insists if a light meter provides the correct exposure for cameras with different sized sensors, that proves that crop factor is completely meaningless. It's a strawman fallacy.

After talking with many people, crop factor became an emotional issue for some who spent hard-earned money without considering crop factor. After dropping $3k on an MFT camera and a 12-35 f/2.8, it sucks to find out that you're often not getting the same results as a 24-70 f/2.8 on an FF camera.

Nowadays, most people take crop factor into their buying decisions and still buy MFT--but they know exactly what they'll get. We do ALL our video with MFT cameras (10 of them!) and often shoot stills, too.

"Crop factor" is the real straw man fallacy because it only accounts for half of what is required for equivalency and misses the real zone of compromise (basically size vs larger/narrower equivalent shooting envelope). As long as the DOF equivalency half is ignored, then the confusion will continue. You're just perpetuating the problem if you're only talking in terms of "crop factor."

Link | Posted on Aug 23, 2017 at 16:55 UTC

This was supposed to be some kind of revelation we didn't already know or couldn't easily confirm for ourselves by looking at DXOMark field maps? The problem is that the "rule" seems to apply fairly consistently across wide zooms but things seem to shift toward the middle focal lengths in telezooms. And the notion that one can dispense with testing any focal length other than the "base" one (widest in the case of wide zooms according to Roger), is pretty wacko if the photographer intends to rely on any focal length other than the base one. As Roger admitted in his blog posted, at the longest focal length in the zooms he tested there was not only more softening BUT ALSO MORE VARIATION! If I'm spending my hard earned dollars on a new lens, you can be darn sure that I'm going to check the focal length most likely to be subject to unit variation. Indeed, I'd say it's even MORE important to check that focal length rather than the base one!

Link | Posted on Mar 31, 2017 at 14:20 UTC as 19th comment

Kudos to Richard and the others who helped him put together this piece that does an excellent job of explaining a difficult topic. Would it be too much to ask for a follow-on article about the software dimension of the problem - i.e., differences in demosaicing algorithms, deconvolution and related postprocessing strategies for minimizing and repairing the unwanted effects of aliasing (and anti-aliasing)? That would be genuinely helpful.

Also, using Siemens star crops from your studio scene test shots for your examples makes sense, but there was a lost opportunity to explicate and illustrate the common "false detail" effect of aliasing by including some magnified crops from the etching of the Dutch family in the DPR studio scene.

Link | Posted on Mar 29, 2017 at 19:02 UTC as 34th comment | 1 reply
On article An introduction to our studio test scene (121 comments in total)
In reply to:

knickerhawk: I have a couple of comments/suggestions for the studio scene itself. Due to the restriction on comment length, I will post these separately below.

1. The "neutral" grays used throughout the scene (e.g. the gray DPR logo box that appears in the middle) do NOT match the middle grays in the GretagMacbeth chart or Kodak gray scale. This is confusing/misleading and frustrating. This means you can't reliably white balance off of the gray objects other than the GM patches. This is particularly problematic in the low light scene when there's a very obvious color shift due to the lighting used and you want to WB from one side or the other.

Last comment:

7. It would be helpful to post somewhere a high-res close up of the etching of the Dutch(?) family. The PhaseOne benchmark is helpful but doesn't clearly show the actual "real" etching lines and may, of course, also be rendering some false detail. A more definitive close-up of the scene will be helpful for better understanding what's being incorrectly rendered in the various test shots.

Link | Posted on Aug 9, 2016 at 14:43 UTC
On article An introduction to our studio test scene (121 comments in total)
In reply to:

knickerhawk: I have a couple of comments/suggestions for the studio scene itself. Due to the restriction on comment length, I will post these separately below.

1. The "neutral" grays used throughout the scene (e.g. the gray DPR logo box that appears in the middle) do NOT match the middle grays in the GretagMacbeth chart or Kodak gray scale. This is confusing/misleading and frustrating. This means you can't reliably white balance off of the gray objects other than the GM patches. This is particularly problematic in the low light scene when there's a very obvious color shift due to the lighting used and you want to WB from one side or the other.

Sixth comment:

6. Care should be taken when shooting to avoid drafts and air movement in the room during shooting. The fuzzy green feather stuff at the top right and bottom left tends to move as can clearly be seen in some of the Pentax pixel shift images.

Link | Posted on Aug 9, 2016 at 14:43 UTC
On article An introduction to our studio test scene (121 comments in total)
In reply to:

knickerhawk: I have a couple of comments/suggestions for the studio scene itself. Due to the restriction on comment length, I will post these separately below.

1. The "neutral" grays used throughout the scene (e.g. the gray DPR logo box that appears in the middle) do NOT match the middle grays in the GretagMacbeth chart or Kodak gray scale. This is confusing/misleading and frustrating. This means you can't reliably white balance off of the gray objects other than the GM patches. This is particularly problematic in the low light scene when there's a very obvious color shift due to the lighting used and you want to WB from one side or the other.

Fifth comment:

5. The scene now lacks any indication of depth. The flat scene makes shooting easier but it creates a bias in favor of non-equivalent aperture setting. I don't mean to turn this into an equivalence discussion, but scene reinforces misunderstandings that DPR staff have fought hard to correct with articles and forum postings. It's ironic, then, that the setup used in the official DPR camera test intentionally circumvents the implications of equivalence.

Link | Posted on Aug 9, 2016 at 14:42 UTC
On article An introduction to our studio test scene (121 comments in total)
In reply to:

knickerhawk: I have a couple of comments/suggestions for the studio scene itself. Due to the restriction on comment length, I will post these separately below.

1. The "neutral" grays used throughout the scene (e.g. the gray DPR logo box that appears in the middle) do NOT match the middle grays in the GretagMacbeth chart or Kodak gray scale. This is confusing/misleading and frustrating. This means you can't reliably white balance off of the gray objects other than the GM patches. This is particularly problematic in the low light scene when there's a very obvious color shift due to the lighting used and you want to WB from one side or the other.

Fourth comment:

4. The scene lacks useful real world objects and over uses printed objects such as the cards. You need both, of course, but more "real" objects like the paint brushes would be helpful.

Link | Posted on Aug 9, 2016 at 14:42 UTC
On article An introduction to our studio test scene (121 comments in total)
In reply to:

knickerhawk: I have a couple of comments/suggestions for the studio scene itself. Due to the restriction on comment length, I will post these separately below.

1. The "neutral" grays used throughout the scene (e.g. the gray DPR logo box that appears in the middle) do NOT match the middle grays in the GretagMacbeth chart or Kodak gray scale. This is confusing/misleading and frustrating. This means you can't reliably white balance off of the gray objects other than the GM patches. This is particularly problematic in the low light scene when there's a very obvious color shift due to the lighting used and you want to WB from one side or the other.

Next comment:

3. The scene lacks extreme highlight and shadow detail. Thus, efforts to see what happens when you pull up deep shadows and push down extreme highlights are not very informative. In particular, there is a real absence of good highlights to work with. The only parts of the scene that are near clipping - e.g., the bowls in the aluminum paint pans are very featureless.

Link | Posted on Aug 9, 2016 at 14:41 UTC
On article An introduction to our studio test scene (121 comments in total)
In reply to:

knickerhawk: I have a couple of comments/suggestions for the studio scene itself. Due to the restriction on comment length, I will post these separately below.

1. The "neutral" grays used throughout the scene (e.g. the gray DPR logo box that appears in the middle) do NOT match the middle grays in the GretagMacbeth chart or Kodak gray scale. This is confusing/misleading and frustrating. This means you can't reliably white balance off of the gray objects other than the GM patches. This is particularly problematic in the low light scene when there's a very obvious color shift due to the lighting used and you want to WB from one side or the other.

Next comment:

2. There is no good continuous tonal ramps in the scene. The Kodak gray scale is step-based and there's no good continuous color ramps in the scene. That's a shame.

Link | Posted on Aug 9, 2016 at 14:41 UTC
On article An introduction to our studio test scene (121 comments in total)

I have a couple of comments/suggestions for the studio scene itself. Due to the restriction on comment length, I will post these separately below.

1. The "neutral" grays used throughout the scene (e.g. the gray DPR logo box that appears in the middle) do NOT match the middle grays in the GretagMacbeth chart or Kodak gray scale. This is confusing/misleading and frustrating. This means you can't reliably white balance off of the gray objects other than the GM patches. This is particularly problematic in the low light scene when there's a very obvious color shift due to the lighting used and you want to WB from one side or the other.

Link | Posted on Aug 9, 2016 at 14:40 UTC as 22nd comment | 6 replies
On article An introduction to our studio test scene (121 comments in total)

The article explains that the aperture setting that represents some kind of "overall" sharpness and "across the frame" consistency is used. There is a correlation with the aperture setting used for brands like Nikon and Canon (usually f/5.6) with the sharpness Profile scores in DXOMark but not one for the aperture setting used for mFT studio scene shots (also f/5.6 generally). DXOMark shows f/4 to be better across the frame for the Oly 45mm and the Panny 42.5mm. My question, then, is how is the optimal aperture actually determined? What parts of the scene are being compared when determining sharpness? Is the optimal aperture setting being determined for each studio scene test conducted or was there a one-time determination of optimal aperture and that's being applied to every subsequent test that uses a particular lens? What could explain the discrepancy between DPR staff's preference for mFT images and DXOMark's measurements?

Link | Posted on Aug 9, 2016 at 14:39 UTC as 23rd comment
On article Pentax K-1 Pixel Shift Resolution: Updated Field Test (209 comments in total)

In the static scene comparison, it looks to me like the "Pixel Shift On" versions are reversed. The one just labeled "On" shows no motion artifacts for the moving cars and the one labeled "Pixel Shift ON with Motion Correction" definitely shows the ghosts of the moving cars. See here:

https://www.dpreview.com/reviews/image-comparison/fullscreen?attr253_0=pixel%20shift%20off&attr253_1=pixel%20shift%20on&attr253_2=pixel%20shift%20off&attr253_3=pixel%20shift%20on%20w%2F%20correction&normalization=full&widget=371&x=3.102545903818957&y=1.4959257870799816

Link | Posted on Jul 23, 2016 at 14:58 UTC as 15th comment
Total: 20, showing: 1 – 20