Forgot Password
Pentax Camera Forums Home
 

Reply
Show Printable Version Search this Thread
02-11-2015, 07:04 PM   #136
Senior Member




Join Date: Oct 2013
Photos: Gallery | Albums
Posts: 269
QuoteOriginally posted by Class A Quote
Eliminating demosaicing must always come first, AFAIC, and then one can essentially produce as many slightly shifted versions of RGB images as one wants. The achievable amount of resolution enhancement would only depend on the time one can afford to spend on the exposure and the precision of the SR mechanism (always assuming still scenes).


I don't know what Olympus is doing. Perhaps they just don't want to confuse regular customers with terms like "demosaicing" but do the right thing behind the scenes.
One thing I didn't see mentioned in Imaging Resource's review is what the 16mp RAW file produced from that process looks like (though maybe that's because of a current lack of compatible RAW converters?); I'd be curious just to see comparisons of that with normal 16mp RAWs from (say) the OM-D E-M1, to isolate the effect of full RGB data for each pixel (and thus no demosaicing needed).

02-12-2015, 06:28 AM - 1 Like   #137
Pentaxian
falconeye's Avatar

Join Date: Jan 2008
Location: Munich, Alps, Germany
Photos: Gallery
Posts: 6,863
QuoteOriginally posted by Class A Quote
Eliminating demosaicing must always come first, AFAIC, and then one can essentially produce as many slightly shifted versions of RGB images as one wants.
Probably yes (although a demosaicing algorithm for a subpixel raw could be developed too -- which made me talk about that mess ...).

However, eliminating demosaicing can be done in two ways:
1. demosaice all frames first, then deconvolve
2. combine 3-4 integer-shifted frames into RGB frames first, then deconvolve

Because the latter requires 12+ frames, I don't think that it is normally done. But (cf. below) Olympus may do so indeed.

QuoteOriginally posted by Class A Quote
I don't know what Olympus is doing. Perhaps they just don't want to confuse regular customers with terms like "demosaicing" but do the right thing behind the scenes.
Well, Olympus takes 8 frames and they say which ones:
R
0, 0
0.5, 0.5

G
0, 1
0.5, 1.5

G
1, 0
1.5, 0.5

B
1, 1
1.5, 1.5

(using my nomenclature, it looks a bit different in their animated gif but it is the same).

So, what they (must) do is this:

- Take two full RGB images (each composed from 4 frames with integer shifts, eliminating demosaicing, a total of 8 frames).
- The two RGB images are shifted (0.5, 0.5)px against each other.
- Overlay the 2x16MP RGB images into one sparse 64 MP grid.
- Deconvolve the sparse grid with the known 2x2 block blur kernel.
- Interpolate and scale the sparse grid into a full 40MP RGB grid.

That's a fairly clever procedure. And the fact that Imaging Resource found no traces of moiré in the 40MP images confirms it.

Class A, thanks for forcing me thinking this over. This will indeed yield much better images, not just resolutuon but also elimination of artefacts.

Resolution wise, it may be however that pxel shifting (rather than Olympus' approach using only two half-shifted RGB images) could yield better luminance resolution.
02-12-2015, 02:25 PM   #138
Pentaxian
Mistral75's Avatar

Join Date: Feb 2009
Location: Paris
Posts: 2,836
Original Poster
QuoteOriginally posted by blende8 Quote
Kimio Tanaka (product development) mentions it in an interview:
He says
(...)
- SR resolution enhancement will be 1,5 - 2x
data size goes up to 4x
only for non moving motives
02-12-2015, 02:34 PM   #139
Pentaxian
D1N0's Avatar

Join Date: May 2012
Location: ---
Photos: Gallery
Posts: 3,815
2x resolution is equivalent of 2x megapixel or is it 4x megapixel? (since filesize quadrupels).

02-12-2015, 02:35 PM - 1 Like   #140
Administrator
Site Webmaster
Adam's Avatar

Join Date: Sep 2006
Location: Arizona
Photos: Gallery | Albums
Posts: 41,365
During our CP+ interview, the reps acknowledged that they have experimented SR-based super-resolution, but they said that they will most likely not be using it for increasing resolution. Perhaps they'll be able to develop a way to increase detail without bumping up the pixel count.

The question is at 5m 45s:


Adam
PentaxForums.com Webmaster (Site Usage Guide | Site Help | My Photography)



PentaxForums.com's high server and development costs are user-supported. You can help cover those costs by donating. Or, buy your photo gear from our affiliates, Adorama, B&H Photo, or Topaz Labs, and get FREE Marketplace access - click here to see how! Trusted Pentax retailers:

02-12-2015, 02:46 PM   #141
Pentaxian
D1N0's Avatar

Join Date: May 2012
Location: ---
Photos: Gallery
Posts: 3,815
So that would mean down sampling the big file again
02-12-2015, 03:08 PM   #142
Pentaxian




Join Date: Sep 2010
Posts: 2,611
Or using the tech to create a Foveon sensor.
02-12-2015, 03:14 PM   #143
Pentaxian
JimmyDranox's Avatar

Join Date: Oct 2012
Location: Ploiesti, Romania
Photos: Gallery | Albums
Posts: 1,604
+++1 The most sincere replay I've ever had seen from a manufacturer. They say that they expected to a much positive response for the KS-1.

About the answer for the future... We hope for the most, and also do they...

Thanks Adam.

02-12-2015, 04:18 PM   #144
Pentaxian
falconeye's Avatar

Join Date: Jan 2008
Location: Munich, Alps, Germany
Photos: Gallery
Posts: 6,863
I guess Pentax intends to use sensor shift to emulate a Foven sensor (from remarks at German DFN forum by a Japanese member).

However, I can now see them play with the idea of superresolution more seriously, now that Olympus presented their approach.

Pentax approach is the first half (first 4 frames) of Olympus approach (8 frames). What is missing is the (0.5,0.5)px shifted Foveon RGB image (second 4 frames).

Problem with that (and why Pentax may see problems with getting extra resolution as mentioned in the interview) is that with only 2 RGB images shifted by a half pixel, only a sparse double resolution image can be made prior to deconvolution. I.e., it is mathematically non-trivial to treat. Moreover, it is numerically instable (two many unknowns) and computing intense.

If Pentax has higher resolution images in the first place (compared to Olympus), a lower frame rate, and less or equal horse power, then the Olympus approach may be out of reach for the Pentax FF camera, at least initially. Moreover, they may have less brain power to spend on this side line, now that they have to create a new product category ...
02-12-2015, 06:48 PM   #145
Junior Member




Join Date: Nov 2014
Posts: 38
So are we getting this new feature on K3? Or only newer models?
02-12-2015, 07:49 PM   #146
Pentaxian




Join Date: Dec 2013
Location: New Jersey
Posts: 884
QuoteOriginally posted by enyceckk Quote
So are we getting this new feature on K3? Or only newer models?
No new feature at all based on the new interview...
02-16-2015, 10:50 AM - 1 Like   #147
Pentaxian




Join Date: Feb 2015
Photos: Albums
Posts: 3,081
Actually, there is a way to boost IQ by doing stacking of multiple exposure such as stacking multiple frames of a video. In this way, it's equivalent to have iso 25 or lower. It's already possible to do this with the K-3 , multiexposure+liveview(=no mirror movement). Stacking 4 frames is equivalent to iso 25, i.e 6 dB improvement in IQ, equivalent to Nikon D800 at iso 50. However, with the K-3, there is still the mechanical movement and time lag of the shutter which create some vibration and lost exposure time. Unlike the HDR mode, multiple exposures does not have the auto-align function. I was thinking of using the video mode and stacking multiple frames , unfortunately, video is not 6k x 4k resolution.
02-16-2015, 10:58 AM   #148
Site Supporter




Join Date: Nov 2006
Posts: 190
QuoteOriginally posted by starjedi Quote
No new feature at all based on the new interview...
Kinda make you wonder why it's still a hot topic. Members either didn't get the memo or choose to ignore it.
02-16-2015, 05:07 PM   #149
Pentaxian
falconeye's Avatar

Join Date: Jan 2008
Location: Munich, Alps, Germany
Photos: Gallery
Posts: 6,863
According to the thread title, we can discuss super resolution independently from what is said in an interview. After all, it is a pending topic for any vendor with sensor shift technology.

In that respect, I found the Olympus interview (linked on DPR page) very revealing. Asked why now, not earlier, they responded that only now the sensor shift has become accurate enough to support subpixel shifts (1µm accuracy). And that this accuracy would be at the basis of the improved shake reduction efficiency too.

It may then be easy to conclude that superresolution and 4+ stops shake reduction efficiency are correlated. So, Pentax may first have to do some homework and improve the efficiency of their SR feature. It isn't yet on par with Olympus' achievements in this department.
02-16-2015, 05:12 PM   #150
JPT
Pentaxian




Join Date: Jan 2012
Location: Tokyo
Photos: Gallery
Posts: 1,202
Olympus was displaying a large print at CP+ to demonstrate their resolution technology, but it didn't look very impressive to me. There was plenty of resolution but it looked strangely flat, as if all the life had been processed out of it.

In reality, the output from most high end cameras can be printed very large to be chewed at normal distances. I don't see the need to resort to this kind of technique to increase resolution if it comes at the cost of reduced overall image quality. Some of the only prints that consistently stand out to me come from the Sigma Foveon cameras. I think I agree with Ricoh's approach here.
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!
architecture, camera, cameras, center, count, ff, file, future pentax, idea, image, images, k3, lenses, mode, models, olympus, pentax, pentax news, pentax rumors, pixel, post, resolution, resolution by pixel, sample, sensor, shift, shutter, technology
Thread Tools Search this Thread
Search this Thread:

Advanced Search


Similar Threads
Thread Thread Starter Forum Replies Last Post
For Sale - Sold: Pentax KX and ME Super film cameras - both in very good condition HenrikDK Sold Items 6 06-09-2014 05:04 AM
In-Mouth Cameras: The Future of Food Photography? RioRico Photographic Technique 2 10-15-2011 05:07 PM
Pixel pitch and lens resolution surfotog Pentax Medium Format 4 06-09-2011 12:01 PM
Future laid out by Pentax -- no full-frame rparmar Pentax News and Rumors 158 10-23-2008 06:35 PM



All times are GMT -7. The time now is 07:46 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