Forgot Password
Pentax Camera Forums Home
 

Reply
Show Printable Version Search this Thread
07-22-2019, 08:41 AM   #16
Site Supporter
Site Supporter




Join Date: Feb 2015
Photos: Gallery | Albums
Posts: 6,987
QuoteOriginally posted by photoptimist Quote
Demosaicing tries to fill in the missing data
the missing data will already be missing... Practically bayer interpolation works quite well because the chance is low that you photograph something that is 100% R (of the exact same color as the red filter), or 100% G or 100% of B , actual colors leak on RGB to a certain amount, so there is almost always some spacial information that can be extracted from each cell.

---------- Post added 22-07-19 at 17:45 ----------

Isn't the absence of optical anti-alias filter indicating that the sensor already out-resolve the lens (+diffraction)? I noticed, 16Mp apsc and 24Mp FF sensors are covered by an optical low pass filter, while 36+Mp FF sensors aren't having a low pass filter.

07-22-2019, 08:59 AM   #17
Senior Member




Join Date: Oct 2015
Posts: 101
Current DFA lenses definitely have more resolution capability than the K-1 sensor can capture; based on just the simple comparison of cropping on the K-1 versus the K-3 which has a denser pixel-pitch. If the higher resolution K-3 has any benefit, then it must be that the K-1 pixel pitch is insufficient to resolve all detail possible in the lens. That said, I still prefer the K-1 because I don't always have to crop...

There's also the 16x data -> better SNR -> better detail even without increasing resolution. I would be interested in a comparison of 4x stacked "normal" pixel shift shots compared to the newer 16x shot methods; it'd be a more fair comparison in terms of total light captured and differences then come from lens, sensor, and shift-methodology (still a lot of variables, but removes one at least).
07-22-2019, 09:03 AM   #18
Otis Memorial Pentaxian
Loyal Site Supporter
stevebrot's Avatar

Join Date: Mar 2007
Location: Vancouver (USA)
Photos: Gallery | Albums
Posts: 33,930
Of course, this whole discussion is a little strange in that it is highly unlikely that such might be added as a firmware upgrade (per OP) for existing K-1/K-1ii owners. Hardware upgrade, perhaps ($$$), but not firmware.


Steve
07-22-2019, 09:10 AM   #19
Site Supporter
Site Supporter




Join Date: Feb 2015
Photos: Gallery | Albums
Posts: 6,987
QuoteOriginally posted by fehknt Quote
Current DFA lenses definitely have more resolution capability than the K-1 sensor can capture
At f2.8 in the center of the image circle, yes you are right the lens will resolve more than the K1 sensor, but at f11 the center resolution of the lens will drop. With a K3 I could shoot with my lens at f8 without DoF blurr, for the same image with the K1 I need to stop down the lens to f11. That's also for that reason that the medium format camera at 50Mpixels don't have an anti-alias filter although the pixel density is even less than that of a K1 or K5. When you stop down at f11 on a K1, for the same image with a medium format camera you need to stop down the lens to f16, at f16 high frequency details are blurred by diffraction, in which case the sensor still resolve more than the lens stopped down.

07-22-2019, 09:52 AM   #20
Pentaxian
photoptimist's Avatar

Join Date: Jul 2016
Photos: Albums
Posts: 3,539
QuoteOriginally posted by biz-engineer Quote
the missing data will already be missing... Practically bayer interpolation works quite well because the chance is low that you photograph something that is 100% R (of the exact same color as the red filter), or 100% G or 100% of B , actual colors leak on RGB to a certain amount, so there is almost always some spacial information that can be extracted from each cell.
Yes, demosaicing does use the data in each color channel to fill-in the missing data on the other color channels. But that process makes some dubious assumptions that hue and saturation don't vary and that only luminance varies at pixel-resolution scales.

But with a "good" lens and some subjects, those assumptions fail which then creates faults in the image. For example, small red berries will disappear on a green leafy background. Stars take on unnatural or the wrong colors. Star trails and thin white lines show oscillating colors. Thin colored lines develop stair-step or dashed-line artifacts. Moire patterns appear in fabrics, feathers, and distant architectural subjects.

Part of the reason cameras have an anti-alias filter is to blur the under-sampled channels but then that really does convert a 36 Mpix sensor into a 9 Mpix one.

QuoteOriginally posted by biz-engineer Quote
Isn't the absence of optical anti-alias filter indicating that the sensor already out-resolve the lens (+diffraction)? I noticed, 16Mp apsc and 24Mp FF sensors are covered by an optical low pass filter, while 36+Mp FF sensors aren't having a low pass filter.
The absence of optical anti-alias filter is ploy to make the camera seem to take higher-resolution images with most lenses and subjects. Most images will look sharper without an optical anti-alias filter. But removing the filter creates a risk that artifacts in some images with some lenses.

Your intuition is right that lower-resolution sensors have more need for an anti-alias filter but artifacts can still occur on higher-resolution sensors with the best lenses and the highest-detail subjects.

P.S. One of the unsung features of later Pentax cameras is the anti-alias filter simulator which lets the photographer use or not use anti-aliasing depending on the lens and subject matter.
07-22-2019, 11:49 AM   #21
Site Supporter
Site Supporter




Join Date: Feb 2015
Photos: Gallery | Albums
Posts: 6,987
QuoteOriginally posted by photoptimist Quote
Yes, demosaicing does use the data in each color channel to fill-in the missing data on the other color channels.
Agreed. I understand what you wrote. Since I didn't study in the USA, I studied the "Theory of information" in Europe (Not sure it that's being taught the same way at universities in the US), but we distinguished data and information. Data can be void of information or loaded with information, but information requires data to be encoded. We distinguish data and information so that to deal with error correcting codes etc..

---------- Post added 22-07-19 at 20:53 ----------

QuoteOriginally posted by stevebrot Quote
Hardware upgrade, perhaps ($$$), but not firmware.
Sounds correct. 16 raw frames couldn't fit in the buffer of the K1. K1 can accumulate 11 or 12 raw images in its buffer.
07-22-2019, 11:56 AM   #22
Pentaxian




Join Date: Dec 2013
Posts: 549
QuoteOriginally posted by biz-engineer Quote
No sure, it depends... I've played with pixel shift of the K1 in order to find for myself when I get a benefit and when I don't get a benefit. For cases where I don't get a benefit it is more simple to take a single frame with electronic shutter.
I've tried pixel shift at various lens apertures with newspaper on the wall as a test chart, using the 100mm macro , focused with 8x mag. in LV.
At aperture f11, I get zero improvement of sharpness with PS vs a single shot ES.
At aperture f8, I get very little improvement of sharpness with PS vs single shot ES.
At aperture f5.6, I can see a bump in sharpness using PS compared to a single shot with ES.
At aperture f4, I can see super sharp pixels in the center but not at the edges when using PS , compared to ES baseline.
So , for example, I never use PS when shooting a landscape at f11 or f16 and ISO100, because.. for me the noise is already very low at ISO 100 and PS doesn't bring me more sharpness at f11 or f16 (lens diffraction).
However, for a macro flat subject, I set the crop mode on K1, get as close as possible, lens aperture at f5.6, and there I get wowed by the detail and color micro-contrast.

So given what I can see of improvements condition with a 4 x pixel shift on the K1, I don't think the 16 x half pixel shift on the Sony A7R4 will be having a benefit enough to bother storing 2Gb files and processing them.
Well... would be nice to have an option of choosing 4x full photosite movement or 16x half photosite movement.
Can k mount use 645 lenses with adapter? How much do those lenses resolve?
07-22-2019, 11:57 AM   #23
Otis Memorial Pentaxian
Loyal Site Supporter
stevebrot's Avatar

Join Date: Mar 2007
Location: Vancouver (USA)
Photos: Gallery | Albums
Posts: 33,930
QuoteOriginally posted by biz-engineer Quote
I studied the "Theory of information" in Europe (Not sure it that's being taught the same way at universities in the US), but we distinguished data and information.
It is taught the same way here, though error-corrected data are still data, though with more valid information.


Steve

07-22-2019, 12:14 PM   #24
Pentaxian
swanlefitte's Avatar

Join Date: May 2015
Location: New Orleans
Photos: Gallery
Posts: 2,407
QuoteOriginally posted by biz-engineer Quote
A center crop of the image benefits more from higher pixels density because lenses optical resolution is better in the center, but no advantage in bumping up pixel density at the edges of the image frame.
Just pointing out by center really means 80% and edges 20%. The standard mtf confuses this by focusing at the center. In crop mode it will cut off the edges on most modern lenses.
Lens Rentals | Blog
07-22-2019, 03:40 PM   #25
Site Supporter
Site Supporter




Join Date: Feb 2015
Photos: Gallery | Albums
Posts: 6,987
QuoteOriginally posted by photoptimist Quote
Part of the reason cameras have an anti-alias filter is to blur the under-sampled channels but then that really does convert a 36 Mpix sensor into a 9 Mpix one.
Yep, my previous post consider 9Mpixels (worst case) to sample 50lp/mm of the lens. Actual images compared at different aperture settings show that detail is already limited by diffraction, even with a 36Mpixel (9Mpixels) sensor.

---------- Post added 23-07-19 at 00:41 ----------

QuoteOriginally posted by swanlefitte Quote
Just pointing out by center really means 80% and edges 20%. The standard mtf confuses this by focusing at the center. In crop mode it will cut off the edges on most modern lenses.Lens Rentals | Blog
Oh for me the center, of the lens, is the apsc crop with the K1 camera.
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!
buffer, center, color, crop, data, dslr, edges, ff, film, filter, full frame, full-frame, information, k-1, k1, mode, pentax, pentax k-1, pixel, resolution, sensor, sensors, shift, stitching
Thread Tools Search this Thread
Search this Thread:

Advanced Search


Similar Threads
Thread Thread Starter Forum Replies Last Post
hoping that aslyfox General Talk 3 12-28-2018 05:55 AM
Pixel shift with pixel artefacts Dericali Pentax K-1 4 07-18-2018 12:57 PM
Lloyd Chambers Prefers Pentax Pixel Shift To Sony's Pixel Shift Fenwoodian Pentax K-1 2 07-16-2018 02:15 PM
K1/Samyang 24mm Tilt/Shift and Pixel Shift DDoram Pentax K-1 3 06-25-2016 02:05 PM
Was hoping to upgrade to K5 this year but... DaveHolmes Pentax DSLR Discussion 12 02-18-2012 07:23 PM



All times are GMT -7. The time now is 05:15 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