Pentax K-3 Mark III vs. K-70: ISO Comparison

Hi all, in this video we put the Pentax K-3 III against the K-70 in an ISO comparison test.
Firmware - K-3 III: 1.01 K-70: 1.12 (both most recent as of posting).
We acknowledge that there is about a 2MP difference between the cameras. The difference is slight and we wanted to present results of 'what ya see is what ya get' for zoomed out and 100% zoomed images.
We made a video comparing the K-3 III and the K-70.
Results were pretty much expected - the K-3 III holds it's noise incredibly well.
And so does the K-70 really.
But the K-3 III is clearly better controlled especially at high ISOs. However, anything over 102400 (the top ISO for the K-70) the ISO on the K-3 III isn't really workable.
Still incredible results and we wanted to share them with you here :)
Curious thing, it looks like my K-70 is overexposing by about 1/3 stop. The exposures are pretty darn close so I hope the images are still good for testing and you find the comparison useful :)
Just jump to 5:21 to see a 'corrected' exposure comparison.
Send us stuff!
Stephen Foster
PO Box 99900 WU 227 122
RPO Meadowbrook
Edmonton AB T6T 0W4

[@KobieMC]

Great comparison! The K-3 III does seem to expose differently compared to previous Pentax Cameras from what I've experienced. Very good control of color as well when the iso starts really climbing.

[@scrptwic]

I found pictures are useable up to 12800 on the K70 at night hand held which I find impressive for a entry level camera with noise reduction applied

[@PentaxTips]

yep the K-70 is definitely no slouch :)

[@johnlucas836]

@@PentaxTips yes I agree with you! For a camera that is only a third of the cost of the K3 lll, that is a good performance!

[@miguelorellana4107]

Nice video! Now a KP vs. K3M3 🔥🔥🔥

[@PentaxTips]

sure wish I had a KP to test!

[@johnlucas836]

Great video! Very informative! Thanks guys.

[@RobHardin-eu5vl]

Some of this could be washed out Using the Lightroom app Or any other app to fix the problems. I don’t deny the problems are existing. I own a k-70 it’s all I could afford. And I’ve been doing pretty good with it. Maybe one day when I can afford it, I’ll get a k-1iii But until then, I’ll stick to what I got.

[@6panel300]

For the price the K70 takes some beating. I've started fixing my iso at 400 then make my adjustments between shutter and aperture. It's far simpler to get good results. Just like the old days with film when you governed but what speed film you had in the camera.

[@angelodsouza1669]

Thanks for sharing 🙏

[@sdhute]

I think the KP has the added noise reduction to even get raw files but the k70 would be more like the k1 which o think has cleaner files without auto noise reduction

[@Oxygenumpl]

I have the K-70 and it is still a good camera .Up to ISO 6400 there are no differences..., wait for Pentax K-72 with 32 MP sensor ;-)

[@johnlucas836]

Don't know about a K-72? But there is rumoured to be a K-90 in the pipeline! It's said to have 4k video and better AF?

[@hoosierdome8698]

Pentax said the K2miii had ISO comparable to the Ki, I'm like to see this, find it hard to believe.

[@PentaxTips]

what I find exciting to think about is what the K-1 II is going to be capable of with K-3 III tech!

[@KosmatiBohem]

It seems K70 blue color is more accurate than K3-3 ...

[@KosmatiBohem]

I gained this video by 15 dB to hear it ... so silent.

[@junieclar]

Very informative. Please try the K5iis vs K3iii. Would be an interesting match! Thanks

[@PentaxTips]

I have a K-5 but no K-5iis and I understand there as significant differences. :) thanks for watching

[@wkunique]

You forgot to mention that it is impossible to completely disable noise reduction in the Pentax K3iii, because even RAW is corrected by the software of this camera

[@Oxygenumpl]

How version of firmware used in K-70 and in K-3 III in this comparision?

[@PentaxTips]

Firmware - K-3 III: 1.01 K-70: 1.12 (both most recent as of posting). Thanks for the comment I've added this info to the description. Take care

[@KosmatiBohem]

K-70 is actually better image, more natural, but red color is gone sooner. While K3-3 show hefty noise reduction which bothers me more than "grain" like noise. Kompare K3-3 o KP and KP will win. K3-3 is marginally beter in pure ISO performance but ruined with aggressive sharpening and noise reduction. For now, KP still holds the best IQ in Pentax lineup for SooC jpgs.

[@KosmatiBohem]

@M Tech I was talking about NR in JPEG, not RAW. JPEG images from K3iii have strong sharpening and NR, to appear better looking at normal viewing magnification, than KP. Looking at 1:1, KP have the best, the smoothest and most natural SooC JEPG files. I guess, that depands on how particular user had SET each of camera and both cameras can be set similar. I can't judge internet based images without having all image manipulation to off on both cameras. But one user uploaded DNG+JPEG and I was able to develop much better and more detailed image from DNG than camera developed (in JPEG). Based on those examples up to 25.600 is USABLE. There is no "noise reduction" on raw and I do not know what you talk about. Since current CaptureOne have no clue about new K3-3, I developed images by hand. Extracting 16-bit raw file, than used external noise reduction software, than edited that in CaptureOne. I like the results. ISO6400 is perfectly good, ISO12800 is good and ISO25600 is still usable, depending on how much "reduction" will be on final image. I guess for Facebook (2 megaxicel files or 2048 pixel by longer edge), even more ISO will be perfectly fine. Sensors really came long way. But based on all DNGs found on internet, K3-3 is not that far better than KP, which is expected, since KP sensor is just one generation older and really good already. I like K3-3 and is on my short list.

[@KosmatiBohem]

@M Tech I was passionately wait that you byte into this "NR in RAW" thing. And you are right. There is "noise reduction" in every digital image from any camera ever produced. There is no other way to make computer image from electricity that sensor outputs to the processor. You have complete lack of knowledge about how stuff really work behind the screen. I'm engineer. Even ISO in digital cameras alone is just deliberate technical decision that is completely "fake" and have absolutely no co-relation to film at all. It's there only to guide what image we can expect from what settings, to help photographers that were used to exposure on film. See in video cameras there is no ISO at all, but they say "gain", which is nothing else than amplification of raw signal, either analog or digital or both. Different brands are "faking" the image differently. They use sensor ans put it in camera with a given IQ, than they learn more, tweak the processor and sell the same sensor in new model, now with ton more ISO steps and better image. From same sensor. How? More refined noise reduction and other trickery. So noise reduction process is ground for any image processor in any camera, or else you wound end up with just useless garbage. Now HOW REFINED it is, that is the most kept science of each brand. Any evolution of cameras and software developers is - wait for it: NOISE REDUCTION process. Where you have endless goal of balancing speed, accuracy and efficiency. Efficiency is how well you interpolate noise into unified colours, accuracy is how well you can recognise usable shapes and considering that with interpolation. Than there is limited performance of processor and decision how long are you willing to let processor do it's thing before it have to flush memory and star next frame. Why K1 is slower than D810, but have slightly better IQ and more accurate colours? You guessed it... because they dedicated a bit more time to process each frame, rather than speed (and processor is ancient in that camera). Why KP is so much better than K70 with same sensor? They added co-procesor to calculate noise reduction in parallel with other needed steps to make raw image. NO ONE EVER noticed ANY detail reduction. If anything there is MORE detail regarding K70 at any given "ISO" setting. The only "fame " came out with K1ii where people compared directly either JPEG images either images imported into Lightroom where Adobe just used SAME profile for both camera. I do not know if today they differ. But I pulled out original raw images into 16-tiff files and than compared that and there was no detail loss at all, up to pixel magnification. Just the opposite: slightly more detail since better (longer) interpolation of raw sensor data (electrical signal from sensor). The procedure is exactly the same as with any commercial software noise reduction, like Neat Image. Every new version pulls out more details than version before, details that it seemed before it does not exist. Last but not least, people mostly mix fine details with fine noise. I personally ADD noise in the end of editing process since people perceive totally clean images as "plastic", "digital", with "lack of detail". But when I add slight noise back, they accept it as natural, detail rich. Same goes with music in process of adding noise in digital CD recording as called dithering. So YES: Every digital image saved in "raw" format was processed and noise reduction is one of the process. BTW images in every raw files of any camera are saved in TIFF format. The proprietary is only container and with Fuji, the arrangement of individual colours in effort to make more organic feel. Hope this helps debunk "NR in RAW" thing. BTW how on earth you know details are lost if you have nothing to compare to.

[@KosmatiBohem]

@M Tech I can only say LOL on your technical "expertness" (based on "he said it"). While you live out of reading internet tales, I live out from designing and developing things users like you than use and "expertise" about how they know how it works. Regarding camera(s), I buy them and use them to capture photographs that I can admire and sell, and the end result is my only concern, not the way manufacturer get there. BTW: What camera do you own?

[@KosmatiBohem]

@M Tech what is your current camera, I asked? Did you ever used K3-3? If you can't answer to both questions than you already describing yourself well.

[@KosmatiBohem]

@M Tech So you are judging K3-3 based on K1 mk1 experience. K3-3 have 10% less details than K1 mk2, which is remarkable. It have roughly 10% more noise than K1ii. And you "see" details lost. Funny. I own MZ50, K-r, K7, K5ii, K3ii, KP and tested K1mk1 and several latest Pentax lenses, but own few of therm. My expectation of K3-3 are met at certain parts, a few bits more than expected and one sad is black AF points in VF - regarding image, there is some 15% improvement over KP, but you NEED to consider that is based on JPEG which is not comparable at all between any different camera models since no developing software have proper K3-3 tuned profiles. So What we see today is not what we will see in the future and difference between KP and K3-3 will widen. I participate as beta used for well know image developing software and there was around one year of constant work on Nikon D850 to perfect profiles and around one and a half year for new Fuji X-trans - the amount of time Pentax will not invest-in to bring same profile perfection to users that other brands do. What I want to point out is how much work was done on proper developing X-trans array so we now can "admire" the IQ. Huge difference from initial "generic" profiles to last tuned out. As I pointed out "noise reduction" under the hood is present as in raw as is in software and have nothing to do with "slider" you can than pull or not. And this was key for X-trans where I participated, to fine tune the signal vs noise. There is always trade of in the end. And end users traditionally always prefer "clean" image vs detailed grainy one. Last but not least, compare ISO100 FILM to digital camera. Modern cameras far exceeds film in sensitivity if we compare amount of noise/grain per given exposure. And also resolution! APS-C with 24 megapixels of real detail unheard details in FF film era for any given gear/film combination. It was half that what was considered razor sharp image. So new K3-3 have 2 spare megapixels to have room to "crank-up" image processing a bit more, vs K3ii/KP - what is the problem? I see this debate as counting the salt particles one toss in his soup.