[Class A]

I don't want to sound fanboyish, as that's not where I'm coming from, but it's not the sensor size that dictates the buffer depth etc.
Sony/Nikon and Canon have several FF cameras that can do fast shooting.

First, you are right, it isn't sensor size per se that dictates buffer depth, fps, etc. It is the number of megapixels. If you keep the pixel pitch constant, though, then larger sensors will have more megapixels (e.g., K-5 -> K-1), that's what I was referring to.

Second, the fast FF cameras by Canikony all have to correspondingly fast electronics to handle the bandwidth. I'm rather sure the K-1's electronics are not in the same league, meaning that Ricoh can lower production cost and charge less. This puts the K-1 out of contention regarding very fast-paced shooting but with 6.5 fps in APS-C mode, it is still usable, albeit with a crop factor of 1.5.

On sensor focus points is pretty important here, but I can't see that working in optical viewfinder mode.

Sadly, it seems that the light available for the metering chip is not sufficient to support something like eye-AF. Sony's SLT technology was interesting in that regard, but it always stole a significant amount of light from the main sensor.

I'm wary of of OPDAF because the split-sensels can cause issues like the striping that can be observed with some Sony cameras in certain backlight situations. It is not all roses when it comes to that technology and running the imaging sensor full-time also is suboptimal.

Before the respective technology improves, I think having an OVF and the option to use Live View (monitored on the back LCD or perhaps with a hybrid OVF/EVF) is a rather acceptable solution.

They already have focus peaking during video for example, it just turns off when you start recording?!

I presume that's due to the particular electronics (Pentax often use Fujitsu Milbaut processors) not supporting both recording and focus peaking at the same time.

Agreed, but honestly I think the definition of what a camera should be able to do is changing, and the 'stills only' score is a bit living in the past.

Not for those who are only interested in stills photography.

Perhaps those only wanting to take stills images and not video are living in the past, but personally I feel it is a valid choice.

I don't see the harm in having two separate stills & video evaluations on top of that dreaded (because almost never appropriate for a particular reader) combined score.

The general expectation is that a camera device should be able to do both, and agree with it or not, the lack of video capabilities is always going to hurt how any manufacturers product review in 2018.

I agree with you that lacking in video can potentially hurt sales.

I don't agree that we should accept that reviews slam a stills photography camera for not doing video well. I think good reviews distinguish between these applications and many reviews do that rather well. In my experience, DPReview are by far the worst offender in terms of having their overall conclusions heavily influenced by video concerns.

[BigMackCam]

Sadly, it seems that the light available for the metering chip is not sufficient to support something like eye-AF. Sony's SLT technology was interesting in that regard, but it always stole a significant amount of light from the main sensor.

Just on that point, my Hasselblad HV (Sony A99) uses SLT technology, as does the LA-EA4 auto-focus A-mount adapter for my A7 MkII. There's about one third of a stop of light lost due to the SLT mirror, which isn't as bad as people think. Obviously the impact is most significant at high ISO settings, but it's really not a big deal, and a more-than-acceptable trade-off. This is the route that Nikon is choosing for the F-mount adapter on its forthcoming mirrorless system, and with the performance of modern sensors, it's perfectly viable...

[Class A]

There's about one third of a stop of light lost due to the SLT mirror, which isn't as bad as people think.

Reports vary between 1/3 stop loss and 1/2 stop loss, which would mean that about 30%-37% of light is lost to the translucent mirror.

This isn't a disaster by any stretch of the imagination but not ideal for low light photography.

[BigMackCam]

Reports vary between 1/3 stop loss and 1/2 stop loss, which would mean that about 30%-37% of light is lost to the translucent mirror.

This isn't a disaster by any stretch of the imagination but not ideal for low light photography.

Subjectively, I'd say it's closer to one third of a stop...

I just took two similar photos on my A7II, both at ISO 6400, one with a manual lens on an M42 adapter (so no SLT in the chain), and the other with a Minolta AF lens with the LA-EA4 SLT-based adapter fitted. First 100% crop below is without the SLT, second is with. No noise reduction of any kind applied. You can see there's a difference (most noticeably at the top left corner), but it's remarkably minor...

Attached Images

   

[Class A]

I just took two similar photos on my A7II, ...

Thanks for sharing. The top right area looks quite a bit brighter in the second shot.

A test using two different lenses will always be limited by potential t-stop differences between the lenses, though.

DigitalRev just photographed a wall without a lens twice: With and without the translucent mirror in the optical path (it can be swung out of the way with an DSLT). One can then either try to adjust the shutter speed to obtain the same brightness (and work out the light loss through the shutter speed difference) or, probably better, use some post-processing to determine the boost required to equalise the two exposures. Pitfalls await using LR, or similar, but that should be the most accurate method.

I'm not sure whether all SLT designs use the same translucency ratio. An experiment involving an adapter may not be transferable to a camera and vice versa.

[beholder3]

No noise reduction of any kind applied.

I hope you did not use any Adobe product as raw converter. They do apply sharpening, exposure corrections and noise reduction based on camera model even if the GUI leaves the user with the impression nothing of the sort is active (sliders zero or neutral).

[BigMackCam]

DigitalRev just photographed a wall without a lens twice: With and without the translucent mirror in the optical path (it can be swung out of the way with an DSLT). One can then either try to adjust the shutter speed to obtain the same brightness (and work out the light loss through the shutter speed difference) or, probably better, use some post-processing to determine the boost required to equalise the two exposures. Pitfalls await using LR, or similar, but that should be the most accurate method.

In fact, I've just done such a test with the A7II and Helios-44-2... first, via a dumb M42 adapter, then with an M42-to-A-mount plus LA-EA4 SLT adapter. The difference is near enough half a stop, which isn't bad but is more than I thought...

I hope you did not use any Adobe product as raw converter. They do apply sharpening, exposure corrections and noise reduction based on camera model even if the GUI leaves the user with the impression nothing of the sort is active (sliders zero or neutral).

No, I'm running Darktable under Linux, and switching off all processing - no base curve, exposure adjustments, highlight reconstruction, noise reduction, sharpening... nada. It's one of many things I prefer with Darktable vs Adobe products...

[DennisP]

The reshot scene is up now...score updated to 80%. Still not a recommended camera by DPR, but that’s not shocking. Time to pixel peep again

[Class A]

The reshot scene is up now...score updated to 80%.

The update to the score is interesting as Rishi Sanyal had insisted before that a re-shoot will not affect their assessment because they had "intelligently" analysed the images (thus avoiding any lens influence).

I take my hat off to Carey Rose (and perhaps other DPReview staff) for explicitly acknowledging that they had indeed based part of their evaluation on the botched up ISO 12,800 file and their incorrect JPG settings. They didn't have to do that and I'm impressed that they are owning these mistakes.

I don't understand how the corrections only result in a 1% score increase, but overall I'm grateful for DPReview living up to their past mistakes by offering a heartfelt apology. I also happen to agree with an overall downgrade if RAW processing cannot be switched off. Whether other manufacturers should receive a similar penalty for doing the same thing at higher ISO levels is another discussion. I feel DPReview owes us an explanation as to why a threshold of ISO 640 deserves an explicit penalty while a threshold of ISO 1600 is tolerated across the board. I also feel that the formulation "near base ISO" is misleading, given that a loss of detail seems to set in at very high ISO settings only (based on their studio scene shots) and definitely not before ISO 640 (according to image analysis based on older firmware results).

I hope that DPReview is still in contact with Ricoh about allowing users to turn off the "accelerator" processing, if they feel it isn't helpful for some applications.

[BigMackCam]

The reshot scene is up now...score updated to 80%. Still not a recommended camera by DPR, but that’s not shocking. Time to pixel peep again

I've just had a look, and it's way, way better. The detail loss at mid range ISOs (800 - 3200) in RAW shots is, IMHO, utterly insignificant. For me, nothing has changed... for my requirements, the K-1II would be the slightly better choice over the K-1, because of lower noise...

[RGlasel]

The reshot scene is up now...score updated to 80%. Still not a recommended camera by DPR

I've just had a look, and it's way, way better.

If anyone needed proof that DPR isn't in the business of doing objective reviews, here you go. Not only is the image quality of the K-1 II equal to or better than all other FF DSLRs, it offers the best implementation of pixel shift in the industry along with the best in-camera noise reduction, at a very competitive price. The K-1's should not be recommended for video, but for a stills photographer, what other DSLR provides better value?


DPR is in the business of being the most popular digital photography review website. Period. That means primarily catering to prosumer wannabes, who like to daydream about equipment they don't own that will magically transform them into photographer all-stars, and secondarily to owners of photographic equipment who want reassurance that they made good buying choices. Showcasing equipment from a brand that no one aspires to, with less than 2% market share, does absolutely nothing for DPR's business.

[Class A]

The detail loss at mid range ISOs (800 - 3200) in RAW shots is, IMHO, utterly insignificant.

I found it hard to find portions of the image which supported the notion that there is detail loss.

The fact that most areas look as detailed as those of the competitors just with much less chroma noise seems to be immaterial to DPReview's overall decision.

The D850 shows substandard JPG processing at high ISO settings (at least in a portion of the published images), yet isn't criticised for "smearing detail" in JPGs. On the contrary, the JPG processing is said to be improved compared to the D810. I hope DPReview will improve on their current lack of consistent criticism.

[texandrews]

The update to the score is interesting as Rishi Sanyal had insisted before that a re-shoot will not affect their assessment because they had "intelligently" analysed the images (thus avoiding any lens influence).

I take my hat off to Carey Rose (and perhaps other DPReview staff) for explicitly acknowledging that they had indeed based part of their evaluation on the botched up ISO 12,800 file and their incorrect JPG settings. They didn't have to do that and I'm impressed that they are owning these mistakes.

I don't understand how the corrections only result in a 1% score increase, but overall I'm grateful for DPReview living up to their past mistakes by offering a heartfelt apology. I also happen to agree with an overall downgrade if RAW processing cannot be switched off. Whether other manufacturers should receive a similar penalty for doing the same thing at higher ISO levels is another discussion. I feel DPReview owes us an explanation as to why a threshold of ISO 640 deserves an explicit penalty while a threshold of ISO 1600 is tolerated across the board. I also feel that the formulation "near base ISO" is misleading, given that a loss of detail seems to set in at very high ISO settings only (based on their studio scene shots) and definitely not before ISO 640 (according to image analysis based on older firmware results).

I hope that DPReview is still in contact with Ricoh about allowing users to turn off the "accelerator" processing, if they feel it isn't helpful for some applications.

Agree completely.

---------- Post added 06-01-18 at 12:30 PM ----------

I've just had a look, and it's way, way better. The detail loss at mid range ISOs (800 - 3200) in RAW shots is, IMHO, utterly insignificant. For me, nothing has changed... for my requirements, the K-1II would be the slightly better choice over the K-1, because of lower noise...

Yes, for me as well.

---------- Post added 06-01-18 at 12:31 PM ----------

I found it hard to find portions of the image which supported the notion that there is detail loss.

The fact that most areas look as detailed as those of the competitors just with much less chroma noise seems to be immaterial to DPReview's overall decision.

The D850 shows substandard JPG processing at high ISO settings (at least in a portion of the published images), yet isn't criticised for "smearing detail" in JPGs. On the contrary, the JPG processing is said to be improved compared to the D810. I hope DPReview will improve on their current lack of consistent criticism.

Yeah, we've still got that problem.

[Rico]

The update to the score is interesting as Rishi Sanyal had insisted before that a re-shoot will not affect their assessment because they had "intelligently" analysed the images (thus avoiding any lens influence).

I take my hat off to Carey Rose (and perhaps other DPReview staff) for explicitly acknowledging that they had indeed based part of their evaluation on the botched up ISO 12,800 file and their incorrect JPG settings. They didn't have to do that and I'm impressed that they are owning these mistakes.

I don't understand how the corrections only result in a 1% score increase, but overall I'm grateful for DPReview living up to their past mistakes by offering a heartfelt apology. I also happen to agree with an overall downgrade if RAW processing cannot be switched off. Whether other manufacturers should receive a similar penalty for doing the same thing at higher ISO levels is another discussion. I feel DPReview owes us an explanation as to why a threshold of ISO 640 deserves an explicit penalty while a threshold of ISO 1600 is tolerated across the board. I also feel that the formulation "near base ISO" is misleading, given that a loss of detail seems to set in at very high ISO settings only (based on their studio scene shots) and definitely not before ISO 640 (according to image analysis based on older firmware results).

I hope that DPReview is still in contact with Ricoh about allowing users to turn off the "accelerator" processing, if they feel it isn't helpful for some applications.

Class A I am still not following the need to turn off the Accelerator Unit. If people think for some reason they don't need the Accelerator Unit because DPR is falsely claiming (which these new test images clearly show) it reduces high ISO detail and contrast and produces inferior image quality they can always buy the K-1 while supplies last.

[texandrews]

Class A I am still not following the need to turn off the Accelerator Unit. If people think for some reason they don't need the Accelerator Unit because DPR is falsely claiming (which these new test images clearly show) it reduces high ISO detail and contrast and produces inferior image quality they can always buy the K-1 while supplies last.

Well, as someone who is going to upgrade I do understand: it would be better. That said, I'm not sure it would be possible, but probably so.


Personally, I feel like if I want impeccable quality, then I'd be shooting at base ISO anyway, or close to it in a pinch. If I'm needing higher ISO's, which for me would mean smaller images/prints anyway, and that for me means up to 16x20 inches, then I'm pretty satisfied with what I have seen so far. And I can definitely see this streamlining my workflow---I use and will use my K1/K1mkII upgrade for fast turnaround stuff, and fussing with tricky NR is time consuming: I don't believe it can be universally dialed into a preset; I think it needs individual image tweaking. So, definitely a slowdown where I don't want one. Happy enough to have that done automatically for what I'm using this camera for.