Forgot Password
Pentax Camera Forums Home
 

Reply
Show Printable Version Search this Thread
02-07-2017, 03:40 PM   #61
Pentaxian




Join Date: May 2009
Location: Somewhere over the rainbow
Photos: Gallery
Posts: 1,001
QuoteOriginally posted by biz-engineer Quote
. You can even stack 200 x 14bits x 36Mpixels shots if you like; there is no buffer limit.
and how do you stack 200 images if by multiple images you mean to use multi exposures this is different than photo stacking, the major difference is that you have to expose the image for lets say using 5 exposures then you have to expose every single image at 1/5 of what is need for the final image. With photo stacking you expose for a single exposure and then combine them later.

---------- Post added 02-07-2017 at 04:48 PM ----------

QuoteOriginally posted by biz-engineer Quote
But wait, you gonna say "What if one of the stacked RAW was blurred and messed up the stack?". Well, on the K1 you can also check mark an option which is "Save process" , that means the K1 firmware does the real time stacking on a single raw, and also saves each RAW on the SD card. So, when you arrive home, you take the DNG already prepare by the camera for you, AND if you are not happy, you can ALSO do the complicated NIKON way.
Understood? :-)
But if one of those 4 exposures have been movement blur from camera movement then you loose the use of the pixel shift to increase the IQ. With stacking you can take 10 images and out of those images will see some that exhibit blurriness you can simple not use that image in the stake.

---------- Post added 02-07-2017 at 05:00 PM ----------

QuoteOriginally posted by normhead Quote
I'm not sure exactly what Ian is doing up there.
Its a simple blind taste test

If you have noticed throughout this I have not said which method of using photo stacking to pixel shift produces a better image. The above image is a direct IQ comparison using the same camera
If there is a benefit to using one over the other then it should becomes apparently clear. For the stack I used iso 100, 200, 400 and 800 for the stacking. This is less than idea for stacking as this is the bear minimum I would use and also I would use all 4 exposures at base iso for even a better cleaner data. Even using the suboptimal set of images for staking I could not say which produced the better image let alone say which was used. If it is as you are hinting at that photo stacking produces less IQ to pixel stacking then it should be easily witnessed and you could put your finger on the image that used pixel shifting.
IF you can point your finger to the pixel shifted capture , photo stacking image and the single exposure image.

Here is the image again

These are all taken using the same camera so it would produce a very good comparison of photo stacking to pixelshifting

---------- Post added 02-07-2017 at 05:19 PM ----------

QuoteOriginally posted by normhead Quote
Pentax PS image and Nikon D810 image at base ISO. And that would be consistent with my own experience. At Base ISO it's 50/50 which images is better, the PS image or the single image.
QuoteOriginally posted by normhead Quote
Not much to choose from. Different focal points have been used. The Nikon looks cleaner on the fabrics, on the top of the left most bottle the detail in the black is superior in the Pentax. Both images are better than the D800e image I prepared.


If a person go through all of the trouble to shot a scene with pentax pixel shift then the Nikon shooter that is fluent in photo stacking could do this


I will let you decide which is photo stacked D810 and which is pixel shifted

---------- Post added 02-07-2017 at 05:27 PM ----------

QuoteOriginally posted by normhead Quote
I have no idea what effect stacking would have on these images but out of the camera, the PS image absolutely wipes the floor with the D810 image at 25600 ISO, if you're interested in that kind of thing. Personally I tend to shoot at more favourable ISOs where it would seem to make less difference. But somewhere between base ISO and 25600 ISO Pixel shift starts to make a noticeable difference.
Point to me where this would be a useful technique in photography?
I ask you why would you go through all of the trouble of using pixel shift that needs 4 exposures that cannot have any movement during the process and just shooting the image at a exposure time 4 times ( plus the time needed between the pixel shifting exposures) longer and shooting at a lower iso ? I Know why because it looks better but in the real world there is no benefit to shooting hi iso and pixel shift.


Last edited by Ian Stuart Forsyth; 02-07-2017 at 05:17 PM.
02-07-2017, 04:27 PM   #62
Pentaxian




Join Date: Feb 2015
Photos: Gallery | Albums
Posts: 8,716
QuoteOriginally posted by Ian Stuart Forsyth Quote
But if one of those 4 exposures have been movement blur from camera movement then you loose the use of the pixel shift to increase the IQ. With stacking you can take 10 images and out of those images will see some that exhibit blurriness you can simple not use that image in the stake.
QuoteOriginally posted by Ian Stuart Forsyth Quote
and how do you stack 200 images if by multiple images you mean to use multi exposures this is different than photo stacking, the major difference is that you have to expose the image for lets say using 5 exposures then you have to expose every single image at 1/5 of what is need for the final image. With photo stacking you expose for a single exposure and then combine them later.
No, that's because you don't know how the K1 works. It's not pixel shift. For example , you take a shot full exposure (14bits), it goes into buffer, you take another shot full exposure (14bits) , it's added to the image from the buffer, divided by 2 and stored back in the buffer, you take another shot, it's added to the image of the buffer , divided by 2 and stored back in the buffer.... that a rolling average or multi-tap digital low pass filter, you can average noise indefinitely. You get a perfectly clean image output, you get the full 14 bits DR without any noise. the way it works is not documented in the manual. Go to Nikon and ask for the same thing.

Last edited by biz-engineer; 02-07-2017 at 04:34 PM.
02-07-2017, 04:45 PM - 1 Like   #63
Pentaxian




Join Date: May 2009
Location: Somewhere over the rainbow
Photos: Gallery
Posts: 1,001
QuoteOriginally posted by normhead Quote
SO I have to ask, is there one photo or series of photos out there that shows stacking can produce anything similar to the type of improvement Pixels shift can or is this just some kind of wild speculation?
Answer my blind tasted test and then you might find out as how they are taken with the same camera, one using photo stacking and the other pixel shift

---------- Post added 02-07-2017 at 05:47 PM ----------

QuoteOriginally posted by normhead Quote
At 3200 ISO which I actually use on the K-1 IR managed to miss focus on both K-1 images. But you can still see the Pentax noise advantage, so basically, $1000 cheaper, for in the end pretty much identical performance at useable ISOs and considerably better performance at ultra-high ISO. Not too shabby.


Again you could just shoot the camera without pixel shifting at an exposure that is 4 time as long and under the conditions of the IR scene they could be done handheld with no need of a tripod for pixel shifting

---------- Post added 02-07-2017 at 05:57 PM ----------

QuoteOriginally posted by mee Quote
I think a bigger issue is people (not you) coming into threads to defend Pentax, completely missing the point of the discussion, and then introducing their own zero sum, us vs them argument to try to change the course of the discussion.. when it was really friendly beforehand and the 'door breaching' was entirely unnecessary (esp when we're in the deemed Non-Pentax section).
You have to remember that the images that I have shown here are only using 4 stacked images at 10 you start to see the real fine detail and this is because of the sub pixel movement in the randomness found in using the photo stacking technique Other camera manufactures that use pixel shifting also have designed their cameras that will shift the sensor 1/2 pixel using up to 8 frames to give you truly an increase in resolution over the pixel count of the sensor

---------- Post added 02-07-2017 at 06:10 PM ----------

QuoteOriginally posted by biz-engineer Quote
No, that's because you don't know how the K1 works. It's not pixel shift. For example , you take a shot full exposure (14bits), it goes into buffer, you take another shot full exposure (14bits) , it's added to the image from the buffer, divided by 2 and stored back in the buffer, you take another shot, it's added to the image of the buffer , divided by 2 and stored back in the buffer.... that a rolling average or multi-tap digital low pass filter, you can average noise indefinitely. You get a perfectly clean image output, you get the full 14 bits DR without any noise. the way it works is not documented in the manual. Go to Nikon and ask for the same thing.
What feature is this, If it is mutli exposure that has been in all cameras back from the K20d and onward then you will have to select the camera to Auto EV adjust for the multi exposure if not then you will blow the image out

Edited

I see that they have added the feature of selecting how the images are stacked
.
This is far from what photo what photo stacking does,you need movement between the exposures to see an increase in resolution and to combine the images together you need to align the images back in align that will not be done in the camera. With out the movement you would not see any difference that is seen here
PhotoAcute Studio Example

and here
A Practical Guide to Creating Superresolution Photos with Photoshop

Last edited by Ian Stuart Forsyth; 02-07-2017 at 05:31 PM.
02-08-2017, 03:30 AM   #64
Pentaxian
D1N0's Avatar

Join Date: May 2012
Location: ---
Photos: Gallery
Posts: 6,421
Original Poster
Nikonrumors says it may have a 46mp sensor: New rumors: Nikon D820 with 46MP | Nikon Rumors

02-08-2017, 05:11 AM   #65
Pentaxian




Join Date: Dec 2011
Posts: 2,944
Will be interesting to see if they can catch up with Pentax.

While Sony's 42Mpx sensor really was a fail on
Noise/resolution: Studio shot comparison: Digital Photography Review
and dynamic range as well:
Image comparison: Digital Photography Review

Just too far behind, maybe with some luck they can draw even with a newer sensor?

Worst case situation is they only get the second grade stuff, like they had in the D5 with its abysmal 1-inch-sensor dynamic range:
Image comparison: Digital Photography Review

and mushy, soft high ISO:

Studio shot comparison: Digital Photography Review

I wouldn't bet on any image quality progress.
02-17-2017, 12:39 PM - 1 Like   #66
Pentaxian




Join Date: Feb 2015
Photos: Gallery | Albums
Posts: 8,716
QuoteOriginally posted by Ian Stuart Forsyth Quote
This is far from what photo what photo stacking does,you need movement between the exposures to see an increase in resolution and to combine the images together you need to align the images back in align that will not be done in the camera. With out the movement you would not see any difference that is seen here PhotoAcute Studio Example
I have a PhD in signal and image processing, especially in the field of automatic object recognition and artificial intelligence, nevertheless, I would not use it to back my claim. I claim that stacking does not increase resolution because, obviously, you can't align two frames to an accuracy better than the minimum size of quantization steps in the frames. The best accuracy you can get in a system is a best the accuracy of the best component of that system. You believe that multiple shots randomly shifted can provide higher resolution , but that's a lie, because in order to realign the frames, you need a reference, and the reference used is the image coded with finite spacial resolution (pixel step), so , the resolution of stacked image can never exceed the resolution of a single frame. However, once you stack multiple frame aligned to each other with the best accuracy of 1 pixel step, of could the S/N is high hence you can accentuate the high frequency components of the image that were sunk into noise before stacking. Then, everyone believe that the resolution was increased. In fact , the bandwidth remains the same, but the roll-off is different.

You could state it otherwise, for an image you could say that the multiframe image looks like having more contrast, but beyond the baseline resolution of the sensor, there is no signal, nothing, accept some moire artifacts.

Last edited by biz-engineer; 02-17-2017 at 05:32 PM.
02-17-2017, 04:22 PM   #67
Pentaxian
photoptimist's Avatar

Join Date: Jul 2016
Photos: Albums
Posts: 4,636
Stacking and pixel shift certainly do improve DR unless the lens or the air are fogged. Add four 14-bit numbers together and you get a 16 bit number or four times the signal. Although the noise adds up too, the standard deviation of the noise in the sum of the four images is only doubled which implies the DR goes up by 1-bit or 1 EV. Sure, true HDR can do even better but that does not mean stacking and pixel shift can't help bring out more detail in the shadows and more subtle gradations in tones.

Stacking may or may not improve resolution in the same way that pixel shift does because of the accidental micro-movements that occur in stacking. If the camera happens to shift an odd number of pixels left-right or up-down (or both), it's like doing pixel shift. Admittedly, this relies on chance so it's extremely unlikely that 4 randomly-shifted stacked images will offer anything as good as the resolution boost of the 4 carefully-shifted PS images, but the effect is there. If fact, it's statistically likely that if you stack enough images, the R, G, and B photosites will visit the "in-between" locations just like PS does.

02-18-2017, 10:09 PM - 1 Like   #68
Pentaxian




Join Date: May 2009
Location: Somewhere over the rainbow
Photos: Gallery
Posts: 1,001
QuoteOriginally posted by photoptimist Quote
Stacking and pixel shift certainly do improve DR unless the lens or the air are fogged. Add four 14-bit numbers together and you get a 16 bit number or four times the signal. Although the noise adds up too, the standard deviation of the noise in the sum of the four images is only doubled which implies the DR goes up by 1-bit or 1 EV. Sure, true HDR can do even better but that does not mean stacking and pixel shift can't help bring out more detail in the shadows and more subtle gradations in tones.

Stacking may or may not improve resolution in the same way that pixel shift does because of the accidental micro-movements that occur in stacking. If the camera happens to shift an odd number of pixels left-right or up-down (or both), it's like doing pixel shift. Admittedly, this relies on chance so it's extremely unlikely that 4 randomly-shifted stacked images will offer anything as good as the resolution boost of the 4 carefully-shifted PS images, but the effect is there. If fact, it's statistically likely that if you stack enough images, the R, G, and B photosites will visit the "in-between" locations just like PS does.
Its about having sub pixel movements this is why cameras that use pixel shift with half pixel shift you truly see a increase in resolution.

At about 4 stacks you start to see an increase in resolution, of course you could do more but I just wanted to test how which was better pixel shift 4 images or stacking 4 images

The answer



With only 4 stacks you can see an improvement over a single image and as for pixel shift I will let you decide whether or not there is a significant difference between the two but as far as I can see if I had not label which was which no one could pick the pixel shift and the photo stacking


These images are what you get form dcraw and RawDigger without any sharpening or subjective post processing applied. Jack Hogan extracted one single image from the PS file and then compared it to all 4 using pixel shift , this is what you see, while you see some increase I don’t think it’s too hard to believe that photo stacking has to go too far to see pixel shift like increases

---------- Post added 02-19-2017 at 12:00 AM ----------

QuoteOriginally posted by D1N0 Quote
Nikonrumors says it may have a 46mp sensor: New rumors: Nikon D820 with 46MP | Nikon Rumors
With the rumor of 46mp it would follow what they have done in the past, their pro level DX camera shared very similar pixel size and sensor properties as their high resolution FF bodies.
You had to look no further than the D300 and the D3X.
IF they where to use what is found in the D500 @ 20mp a FF sensor using the same tech would give you 45mp

Photographic Dynamic Range versus ISO Setting

any increases we see either using sonys 42mp or a larger D500 sensor would be welcome


Last edited by Ian Stuart Forsyth; 02-18-2017 at 11:00 PM.
02-19-2017, 03:25 AM   #69
Site Supporter
Site Supporter




Join Date: Mar 2009
Location: Gladys, Virginia
Photos: Gallery
Posts: 23,662
QuoteOriginally posted by Ian Stuart Forsyth Quote
Its about having sub pixel movements this is why cameras that use pixel shift with half pixel shift you truly see a increase in resolution.

At about 4 stacks you start to see an increase in resolution, of course you could do more but I just wanted to test how which was better pixel shift 4 images or stacking 4 images

The answer



With only 4 stacks you can see an improvement over a single image and as for pixel shift I will let you decide whether or not there is a significant difference between the two but as far as I can see if I had not label which was which no one could pick the pixel shift and the photo stacking


These images are what you get form dcraw and RawDigger without any sharpening or subjective post processing applied. Jack Hogan extracted one single image from the PS file and then compared it to all 4 using pixel shift , this is what you see, while you see some increase I donít think itís too hard to believe that photo stacking has to go too far to see pixel shift like increases

---------- Post added 02-19-2017 at 12:00 AM ----------



With the rumor of 46mp it would follow what they have done in the past, their pro level DX camera shared very similar pixel size and sensor properties as their high resolution FF bodies.
You had to look no further than the D300 and the D3X.
IF they where to use what is found in the D500 @ 20mp a FF sensor using the same tech would give you 45mp

Photographic Dynamic Range versus ISO Setting

any increases we see either using sonys 42mp or a larger D500 sensor would be welcome
Pentax chose not to have an increase in resolution with their pixel shift option. What you get is just a sharper, cleaner RAW file with more accurate color information. If you want bigger files than 36 megapixels, there are plenty of ways to do that and I imagine panoramas are actually a better way to do that then image stacking.
02-19-2017, 06:16 AM   #70
Pentaxian
D1N0's Avatar

Join Date: May 2012
Location: ---
Photos: Gallery
Posts: 6,421
Original Poster
Tony Northrup has some great advice for Nikon. Most of it could also be applied to Pentax.
02-19-2017, 07:26 AM   #71
Loyal Site Supporter
Loyal Site Supporter
monochrome's Avatar

Join Date: Mar 2008
Location: Working From Home
Photos: Gallery | Albums
Posts: 26,276
QuoteOriginally posted by D1N0 Quote
Tony Northrup has some great advice for Nikon. Most of it could also be applied to Pentax.
Is Nikon FAILING?! "Extraordinary Losses" - YouTube
  1. Finally someone said it is Nikon Precision that's the big problem
  2. Yes, Nikon is downsizing
  3. No, Nikon isn't going under
  4. Stacking - Pentax is showing this in KP
  5. Retro Cool - Pentax is (trying to) showing this in KP
  6. Apps and screens - would we pay for them?
  7. I think we're seeing the germs of Ricoh's strategy in KP
02-20-2017, 09:52 PM - 1 Like   #72
retired sw engineer
Loyal Site Supporter
reh321's Avatar

Join Date: Nov 2014
Location: South Bend, IN, USA
Photos: Gallery | Albums
Posts: 19,619
QuoteOriginally posted by D1N0 Quote
Tony Northrup has some great advice for Nikon. Most of it could also be applied to Pentax.
Is Nikon FAILING?! "Extraordinary Losses" - YouTube
Some of his advice to Nikon would be irrelevant if applied to Pentax. He thinks they should develop IBIS; in his opinion, IBIS is clearly superior to in-lens stabilization, and then he says you can do other things with the hardware, such as leveling the picture. Since Pentax already has IBIS and horizon-leveling, I saw this as his saying that Pentax is doing a lot of things correctly.
02-20-2017, 10:44 PM   #73
Pentaxian




Join Date: Jan 2015
Location: mid nth coast,nsw
Photos: Gallery
Posts: 5,608
QuoteOriginally posted by reh321 Quote
Pentax is doing a lot of things correctly.
And some things SLOWLY!
02-20-2017, 11:41 PM   #74
Pentaxian




Join Date: Feb 2015
Photos: Gallery | Albums
Posts: 8,716
QuoteOriginally posted by Rondec Quote
Pentax chose not to have an increase in resolution with their pixel shift option. What you get is just a sharper, cleaner RAW file with more accurate color information. If you want bigger files than 36 megapixels, there are plenty of ways to do that and I imagine panoramas are actually a better way to do that then image stacking.
Oh yes, panorama works way better than stacking or PS. Well know as gigapixel images.

---------- Post added 21-02-17 at 07:59 ----------

QuoteOriginally posted by D1N0 Quote
Tony Northrup has some great advice for Nikon. Most of it could also be applied to Pentax.
Outsiders can be spot on. The problem of turning around companies is that it is much more difficult to turn a successful company around. In some ways it is easier to turn a bad performing company around. Companies quoted in stock markets need to focus on financial figures, and when they are formerly successful in declining market, if they do not anticipate the decline they get caught up into a financial loss catch up. Restructuring must happen before the market declined. Nokia was like that, impossible to turn around in a short time, they never could get out of the downward spiral.

---------- Post added 21-02-17 at 08:03 ----------

Other thing is, Tony was given a K1 for evaluation and he is now saying sensor stab is great.

---------- Post added 21-02-17 at 08:19 ----------

Regarding the last comments about hiring the right people, that's right and wrong. The constraints on a camera are different from a mobile phone. Camera processors are designed to manage camera hardware and real time image processing, for instance real time image processing (performed in hardware) to cope with the flow of data in burst mode and AF tracking of large lenses. OTOH, smart phone don't have to drive specific camera hardware, the processing platforms of smartphones are geared towards connectivity, browsing, the main actuation is display, essentially. On a DSLR, that's a challenge to put together a software platform performing like smartphone and also processing for the hardware controls of the camera, added to that difficulty is the small volumes of camera sales as opposed to volumes of smart phone.

---------- Post added 21-02-17 at 08:29 ----------

Regarding camera leveling, yes. When I was shooting with the K1 beside a D810 , both using gimbals, the Nikon guy was handing me a spirit level to level the gimbal so that the camera and lens where rotating around the center of gravity. I switched on the level display of the K1, and said no thanks. He looked at the K1, astonished. Turn the K1 is two opposite direction and you calibrate the level of the gimbal, without needing any other accessory.

Last edited by biz-engineer; 02-21-2017 at 12:32 AM.
02-21-2017, 04:23 AM   #75
Pentaxian
D1N0's Avatar

Join Date: May 2012
Location: ---
Photos: Gallery
Posts: 6,421
Original Poster
QuoteOriginally posted by reh321 Quote
Some of his advice to Nikon would be irrelevant if applied to Pentax.
That's why I said most of it. Like decent software development and connectivity, developing a touch interface. and try to constantly improve them also on existing products. Ricoh is trying to do some of the stuff Tony is mentioning but it is very hard for them because they do not have a dynamic software development team. Maybe Ricoh development should get more deeply entrenched in the camera division.
Reply

Bookmarks
  • Submit Thread to Facebook Facebook
  • Submit Thread to Twitter Twitter
  • Submit Thread to Digg Digg
Tags - Make this thread easier to find by adding keywords to it!
af, consumption, couple, d810, ibis, image, images, increase, lenses, nikon, nikon d810, noise, option, pentax, performance, photoacute, pixel, pm, post, power, price, production, production d850, rumors, sensor, shift, system, time, updates, version
Thread Tools Search this Thread
Search this Thread:

Advanced Search


Similar Threads
Thread Thread Starter Forum Replies Last Post
Is the Sigma 500mm f4.5 Pentax fit going out of production? MetteHHH Pentax SLR Lens Discussion 13 06-24-2018 02:02 AM
Pentax K1 or Nikon D810 amarsh1958 Pentax Full Frame 14 12-10-2016 01:16 PM
Nikon D850 Bunch Non-Pentax Cameras: Canon, Nikon, etc. 7 01-11-2016 02:25 PM
Is Pentax DA 35mm f/2.4 AL out of production? uday029 Pentax SLR Lens Discussion 3 06-18-2015 01:43 PM
FA Limiteds (apparently) ARE out of production monochrome Pentax SLR Lens Discussion 70 04-25-2015 05:29 PM



All times are GMT -7. The time now is 01:28 AM. | See also: NikonForums.com, CanonForums.com part of our network of photo forums!
  • Red (Default)
  • Green
  • Gray
  • Dark
  • Dark Yellow
  • Dark Blue
  • Old Red
  • Old Green
  • Old Gray
  • Dial-Up Style
Hello! It's great to see you back on the forum! Have you considered joining the community?
register
Creating a FREE ACCOUNT takes under a minute, removes ads, and lets you post! [Dismiss]
Top