Pentax K-1 Pixel Shift: An update

Hi, about the real value of pixel shift technology in a very high native IQ camera such as the K1: within its boundaries of use, pixel shift high res works very well. In any applicable field of photography where the right gear is medium format, the K1 in pixel shift high res can achieve comparable or better results with a smaller, lighter, significantly less expensive system (the 645Z is now $7,000 at B&H). For pro photographers in a number of fields, this means being able to expand the business with a lower gear investment (often a make-or-break condition). This means that the K1 is over-engineered for many pros (amateurs just need to want it and have the $ to buy it). With my fine art photography printed to a max 24x36" size on watercolor paper and canvas, I don't NEED this camera (but would jump through hoops to get it just because of its awesomeness). At last, there is no premium price attached to the relatively inexpensive K1 because of the presence of this technology.

way to turn those frowns upside down ;)

Thanks to the readers of Dpreview for pointing out the mistakes in the previous article.
As Simon himself wrote, ": “Mistakes happen - it's how we deal with them that matters".
It's a shame they weren't given credit for pointing out the errors in this article, though.

Maybe this has been covered elsewhere, but since the article has been pulled for now, this is what Pentax said about Pixel Shift on the K-1 at PhotoCON LA on May 21, in a nutshell:

"While in certain cases Pixel Shift may be able to do some motion correction for handheld shots, you'll want to use a tripod.

In parts of the scene where motion is detected between (any of) the 4 frames - such as with people walking - only one frame is used for the moving object. For that part of the scene the camera's standard resolution is used. For the rest of the frame the 4 shots are combined for the highest resolution.

[Implied in their presentation: "It's not always perfect, but it's impressively good!"]

"You may need to ensure this (motion detection) function is turned on in the camera, and when processing RAW files you need to use converters which are designed to support this function."

It seems most appropriate to test the gear with the bundled software as this is how the manufacturer has, in theory, optimized their gears for. Adobe can be tested for how well their RAWs can be processed by a man stream software but this is not a substitute for getting the best images possible from the cameras. Due to how Adobe implements the cloud subscription to their software, I am seriously considering switching to an alternative so whether Adobe is main stream or not does it matter to me.

Thanks for doing the right thing!

I missed the original article but may I suggest that the updated version not only includes the appropriate processing of the RAW files but also a) includes a still scene and b) avoids high f-stops in order to not limit resolution due to diffraction?

I don't mind a waterfall scene and other realistic landscape scenarios but surely Pixel Shift also has its usages in the studio for completely motionless product shots, for example.

Good deal, always appreciate when a company responds to its readership. Am truly in the Pentax. Like to upgrade my aging canon 40D but canon seems to have fallen behind the curve, since I don't upgrade often not happy about buying From a company whose sensors have fallen behind. I have some canon glass but not a ton, also have soft spot with Pentax. ME and ME Super were tone my first SLR,s.

I for one would not be to critical of the testers article. It's not so easy. Anyone on their high horse should apply for work at DP then you'll learn it's easy to be an arm chair quarterback, but in real life when you have to try and do it is a totally different story.

DP cost me nothing and is a great source so I'm always appreciative.

Dave

And in the meantime, Pentax probably suffered severe marketing damage.

Bravo! Very rare in journalism to admit a mistake.

Good on you DPR for acting with integrity in this matter and acting on readers comments.

Thanks, DPR. The mea-culpa is hugely appreciated.

recognizing our errors makes us greater, congratulations DPReview

While reading the article and seeing all that mess in the pictures I was wondering, why the hell are they using flawed Adobe Camera Raw instead of bundled software?

To Simon J & Barney B
Following 200+ comments in 24 hours from DPR members in response to ‘Waterfails: We test Pentax K-1's Pixel Shift’ by Chris Williams, and messages left in Site Feedback, you did the right thing and made a good decision to withdraw the article and “publish an updated analysis as soon as possible”. Anyone who uses Pentax Digital Utility will understand that this could take time – even if it is the best or only program that handles pixel shift raw files successfully – there’s a learning curve.
Barney; regular visitors to DPR who saw your reaction to the K-1 launch on video will already know of your genuine interest in the K-1. I also appreciated your comment Simon: “Mistakes happen - it's how we deal with them that matters. It's kind of ironic because everyone here loves the K1”. A visit to Chris’s website also illustrates that his landscape photography is of the highest order so he should be able to get the most out of this Pentax camera if using the right lens.

To Simon J and Barney B (continued)

Most comments from us readers identified the problems with subject, process, and write up. If some comments were reactionary, it was a response to the headline, conclusion, and Chris’s subsequent posts. It’s time for us all to draw breath and wipe the slate clean.

Speaking for myself, I visit DPR daily because I enjoy the content and trust the reviews as being impartial. I will continue to do this and look forward to DPR’s analysis of Pentax’s embryonic technology which, when used correctly, offer us such high resolution images.

I suggested this to someone else who (it seems) didn't run with it, so I'll punt it here too...

I have a suggestion for a review, if that's allowed? There's an increasing number of raw converters around now, and reviews of them are generally poor as reviewers just don't have the time to do in-depth testing (or an expertise with that product). My suggestion is to crowd-source the data.
Provide a set of six (-ish) images taken with four (-ish) cameras (at the same time) and ask people to provide a JPEG produced with a converter's default settings and the best image they can make from it. DPReview staff then pick the best "Tweaked" images (plus rule out any outliers in the "default" category) and make a review from the results.
Suggested scenes:
* Landscape
* Studio Portrait
* Architectural
* Bird/Animal
* Street
* Low light/High ISO
Suggested cameras:
* Canon 5Dsr (good off-sensor ADC cam)
* Nikon D810
* Sony A7rII
* Fuji X-Pro2 (some converters won't handle it)
What do you think?

Does the movement correction cancel out the benefits of pixel shift in the areas of the image that require correction?

ARRRGGHHH!! Make the RAW file available and let me run it through Capture One and DxO. I'll give you the run down on these files in 30 minutes.

Good start. Now would be the appropriate time to establish a policy to include use of bundled software in all reviews. Use third party (Adobe) as well if you wish to show what might be achieved, but learn from this that review should always reflect what the customer would receive.

Might be a great opportunity to look broadly at the wider array of Raw Converters in general, gaining a better understanding of products that might not rank highest on the sales chart but have surpassed the presumed leaders who have done nothing to better their product in years, or adequately support small players such as Fuji and Pentax.

Maybe you could test the pixel shift on the swirling water as you flush this bad article down the toilet.

First, thanks to DPReview for deciding to re-visit the original article.

I think a better slant to the article (that has been suggested elsewhere) is to choose subject matter at increasing levels of movement and show the weaknesses or tradeoffs of using pixel-shift. The user can then decide where their own personal limit is and whether pixel-shift is useful for them, if at all.

If DPR really wants to get involved they could mention recommended best practice post processing techniques or at worst lead users to an alternative source where users are using pixel-shift to their advantage.

Rather than trying to categorically say the feature is good or bad or a write-off or even fantastic, it can be written to be open ended. You know, the pluses, the minuses, best practices, alternative sources of information and un/successful use cases.

Continued in next post...

I'm pleased that you have decided to revisit this issue. Good on DPReview for making this decision.

It would be good to compare the new Silkypix images to the original ACR images. It would really show how different engines process the digital files.

"It's kind of ironic because everyone here loves the K1. SJ"/Simon Joinson

http://www.dpreview.com/forums/post/57857794

Good on you for doing the right thing. This is why I come to DPreview.

Looking forward to the rewritten article.
Looking forward to seeing a change in your looking to Pentax in general.
This camera has many unique features combined with a powerful body that makes it a unique camera and has no similar. OK it is not flawless but it is the nearest camera to perfect.

OK time to start the comments afresh also.

Kudos to Simon Joinson, editor-in-chief, and his team to take that article down.
See his response: http://www.dpreview.com/forums/post/57856750
It is good to see them listen to the readers and take them seriously.

And thanks to the community here who was very outspoken with factual criticism and with a plea for appropriate diligence in testing and journalism. I was impressed to see everybody joined in, not matter from what brand somebody's gear is coming.

Cheers

This is absolutely the correct way to handle the situation. Good on ya. Be sure to post some version of this announcement in a prominent place in the final article.

Wait. Dpreview not relying on Adobe anymore because it may not represent the best raw conversion for a particular camera??!!

Kudos! About time. and if you finally stop saying "Fuji xtrans lacks detail" because you always use ACR for it, and try the latest Iridient digital for it or even the very latest silky pix pro 7 for it which supports out of the box all Fuji film simulations directly (as well as a more trimmed version being what Fuji uses for its customers) you will get a double kudos bonus!

Bravo, DPR.

Sensible decision, albeit very late as this was a known issue for two weeks.

Fair enough.

Have a word with the editor on watch at the time as well...

"WaterFails" ... I mean... come on.... Really??!!

Awful, awful copy. I know it is tough providing so much content, but try and keep the QC up to standard.

Thank you for doing the right thing. And, hopefully, we'll see no more "WaterFAILS" articles - at least, until you make sure it really is the camera maker's failure ;-)

I hope you'll be able to "salvage" much of the hard work from the initial article (if not the conclusions).

Thank you for this. Its much appreciated.
You might want to also consider the F16 setting used. Although its common in landscapes, its really hitting the defraction limit on a standard 2.8 Zoom.

I feel that was an appropriate way to handle the situation, couldn't ask for more.
Looking forward to the updated review, and eventually the full review.
Thanks DPR.

I applaud you for your openness and honesty - It is why I feel I can trust a source like DPR over many others. Keep up the good work and I look forward to your sorting this out and getting us the facts.