[Kenn100D]

I noticed that the photo taken was at Pentax KP firmware 0.35. I'm sure it will be better if it was firmware 1.00 same as the sample phots in Japan website.

[Unregistered User 8]

I noticed that the photo taken was at Pentax KP firmware 0.35. I'm sure it will be better if it was firmware 1.00 same as the sample phots in Japan website.

Well spotted !!

[mee]

It would be more dramatic, I suspect, if they shot in the same spot with the same lens and the same settings using a K-3 II just to show the difference.

It looks good for crop to me, maybe half stop difference. But it isn't K-1 quality or even close... even at ISO 6400. Much of the detail is pulled from the buildings and structures in the shot.. they're all blotchy. Even zoomed out it is obvious.. lines aren't anywhere close to crisp, the pattern on the overpass is blotchy.. there is a lot of obvious noise in the sky. It appears to be a good crop sensor and electronics, but it isn't defying physical 'laws' here.

I'm more interested in seeing ISO 800 and ISO 1600 as that is where the major drop off has traditionally been for crop imo. If those are cleaner, then we have a winner.

[RobA_Oz]

For those of us (like me) who were impressed by being able to push-process Tri-X and HP5 to ISO3200, with grain the size of golf balls, the high ISO performance of current digital sensors is still impressive. The K-1 at 12,800 is highly usable, I find, but this looks even more promising for future bodies in all formats (even, dare I say it, the Q).

[clickclick]

Overall looks very promising and hopeful for Pentax development. Assuming the KP is part of a new line up, and not the new APS-C flagship, it hints towards even greater things in the works. I have to think this is good news for Pentax in the industry.

[FabioPirovano]

You can work at 12800 like be a 1600/3200.. pretty great.. too expansive camera however

[rawr]

but I couldn't find any of the new cameras, like K-70 that supposedly has a similar sensor.

The dpreview image comparison of the K-70 RAW's at high-ISO vs K-3/K-3II and D7200 does show very clear RAW's, with chroma noise in particular almost non-existant at high ISO.



Either the K-70/KP sensor is indeed somehow an improved low-noise/low heat [?] design, or the RAWs are also fed through the 'Accelerator Unit' a bit. DxO tests will be interesting.

[stevebrot]

I'm still not particularly impressed. One could get similar results using third party NR software, I think.

I don't know for sure, but example photos typically show base S/N without NR (actual data capture) as opposed to NR in which most data are retained and noise are replaced by non-data that resemble data. Translation? If the examples show base results, the detail capture at high ISO is incredibly good on the KP.


Steve

[stevebrot]

Not that impressing me is important but, here's a shot at ISO 800, pushed 4 stops (ISO 12800 equivalent):



I could barely see this little guy thru the OVF. DXO PRIME noise reduction.

So you are showing us ISO 800 noise reduced by DXO Prime at Web resolution with almost zero low value detail. Got it. BTW...ISO invariance is an illusion. No data in the shadows are still no data in the shadows regardless of how many stops boost is applied in PP and the amount of noise reduction applied.

Based on what I saw in the examples, the same subject taken with a KP with no NR in PP at 12800 will show better shadow detail and better detail capture overall. Not having a KP at hand, that statement is pure conjecture, however


Steve

[stevebrot]

I wonder if ISO figures are inflated, and if yes, by how much.

ISO as it applies to digital sensors is sort of a fuzzy concept. There is a collection of standards (five in total) that are not strictly comparable to each other and which are used by the various camera makers. It is enough to say that comparable sensor response and S/N ratio at a given flux for the same ISO is not something you are going to see between brands. Edit: Translation...yes, some makers ISO number are inflated/deflated (Nikon are you listening?).

Here is a brief discussion and enough to make most of us either very confused or fairly disgusted...

https://en.wikipedia.org/wiki/Film_speed#Digital_camera_ISO_speed_and_exposure_index


Steve

[luftfluss]

So you are showing us ISO 800 noise reduced by DXO Prime at Web resolution with almost zero low value detail. Got it. BTW...ISO invariance is an illusion. No data in the shadows are still no data in the shadows regardless of how many stops boost is applied in PP and the amount of noise reduction applied.

Based on what I saw in the examples, the same subject taken with a KP with no NR in PP at 12800 will show better shadow detail and better detail capture overall. Not having a KP at hand, that statement is pure conjecture, however


Steve

The NR in the KP shots is being done via "in-camera" PP thanks to the "accelerator unit". Pentax acknowledges this in their literature.

I didn't say anything about ISO invariance.

I'll see if I can find a scene to shoot at very high ISO that is somewhat comparable to the one we are discussing.

Here's a 100% crop of the pic I posted, no PP.

[interested_observer]

I shoot with my K5IIs and do use 51,200 for framing, leveling and focus check when I shoot astro landscapes. 50 miles away from the city - a couple of miles off the main highway at a trailhead parking lot out in the desert, no moon, and can't see your hand in front of your face dark in the middle of no where, you crank up the ISO to 51K shoot for 5 to 6 seconds. I get excellent results to check everything out and then you are ready to go - crank the ISO back down to 1600 and go shooting for keepers.

With the examples and in particular the K70 shots, it appears that the K70 and perhaps the KP are fairly clean up to 3200 or maybe 6400 - equivalent of 1600 on the K5? That would be wonderful! Now, I do wonder what the K3II replacement will bring?

For long duration exposures - I really don't see the D500 as much as a standard to shoot for. This long duration database is a good starting reference point - and the K1 and K5 are really good - better than the K3. So, a stop below the K3 is a good starting point....

• The Sensor Noise DB. – Brendan Davey Photography

Get me 1/3 or 1/2 of a stop under the K5 and I'll be a pretty happy camper - but I still want to see what the K3II replacement bring us. If the k70 betters the K3II and the KP betters the K70, I am expecting the new top of the line replacement to be the best of all - in terms of low noise, IQ and DR.

[Wild Mark]

6400 and below looks really good (high res images). With noise cleanup these images would be very acceptable.

The KP could be an excellent telephoto and macro camera.

[jpzk]

Impressive, to say the least!
That even beats what my K5 can do and I have pushed it often beyond ISO 3200, sometimes at 6400 and thought the results were more than acceptable.
It sure beats my K3 in that domain, that's for sure!
(always shooting RAW).

So one would expect the K3/K3II successor to be able to handle high ISO much better than the K3 (and hopefully better than the K5) AND have a better AF performance, right?
This KP seems to have what the next APS-C "flagship" should have in terms of ISO handling; now, we have to hope that they'll (Ricoh/Pentax) twin that with more AF "power".
(sorry about my not-so-great interpretation but that's the way I would like to see it).
Cheers!

[PiDicus Rex]

You guys n gals did notice the link at the bottom of each page?
The one that opens the full sized images?
With the exif data intact,...

If this Sensor & Processor came in a body that had decent Video-Out hardware and firmware, I could match the quality of output from a Red One or Red Epic with it at 6400 and 12800.

Do we know of any sensor tech breakthrough that would warrant a drastic increase in a sensor signal?

Yes.
The Sony sensor in the A7s.
Having shot stills and video with one, I would definitely say this is the next generation APSc development of that tech.

The noise in the 102400 photo would be fine in News footage (y'know, if we could get Video out of it via the missing HDMI,..)

The noise at 204800 is about the same as 3200ISO video has on K-01 thru K-3 with all the h264 compression.

---------- Post added 28-01-17 at 03:50 PM ----------

if they could wait a little and put this sensor tech in FF for K1, then likely we have a very usable iso 204800 in K1, an instant A7s killer.

Lolz,... cos the A7s is the 'Video focused' model in the A7 range.
The A7's also get sLog2 and sLog3 and 4K.

The KP looks to be more expensive locally then A7s body-only, but cheaper then the A7s2