|
Forum: Pentax K-1 & K-1 II
05-09-2018, 12:46 PM
|
| |
If you really believe that the Poissonian arrival times of photons are part of nature, then you should never correct the chroma noise of a high-ISO image from a later generation sensor. Those weird blotches of color really are in the low-light sample of the scene. They are NOT artifacts of the sensor as much as they are artifacts of the Poisson distribution in RGB.
|
|
Forum: Pentax K-1 & K-1 II
05-08-2018, 05:36 AM
|
| |
But what about the RAW convertor?
SteveBrot published a comparison of two versions of the same RAW file demosaiced with different convertors: (DPReview: K-1 IIs Noise Reduction costs Details - Page 2 - PentaxForums.com)
 
They show different artifact patterns and different amounts of edge softness.
How do we know that the loss of detail is being caused by the camera and not the RAW convertor?
|
|
Forum: Pentax K-1 & K-1 II
04-27-2018, 01:53 PM
|
| |
Actually, it looks more like moire to me. If you increase saturation to 100%, you can see an angled array of yellow blobs in some of the patches. But the array spacing and angle are quite different in the images from the two cameras.
Moire is exquisitely sensitive to angle, location, focal length, and focus. A mere 0.2° change in the angle of the sensor in the camera on the tripod can change a moire blob spacing of 100 pixels to 150 pixels. And a 0.2° change when swapping cameras on a tripod doesn't seem too unlikely.
|
|
Forum: Pentax K-1 & K-1 II
04-27-2018, 06:01 AM
|
| |
Yes, there are diagonal line artifacts and they certainly look like the effect gaweidert discusses.
Perhaps another problem with OP's image and these images come from not turning on the AA-simulator (or the simulated AA is not strong enough) while taking shots of high-detail images with a good lens. Without proper low-pass filtering, there are going to be unavoidable artifacts with a Bayer filter sensor and a high-detail image.
|
|
Forum: Pentax K-1 & K-1 II
04-26-2018, 10:14 AM
|
| |
Right, there no free lunch.
But that does not mean that Ricoh can't invest money in R&D to build non-destructive noise reduction processes in a custom chip and then sell you that lunch.
|
|
Forum: Pentax K-1 & K-1 II
04-25-2018, 07:08 AM
|
| |
But what about information-preserving or even information-enhancing NR? An NR algorithm that removes dark current, corrects for pixel-to-pixel sensitivity variations, or removes cross-talk both reduces noise and enhances information in the sense that the resulting RAW file is a more accurate representation of the light levels in the original scene.
Great points! You are right.
This is where we diverge because the digital chips on the camera certainly do have access to some sensor-internal data/properties from a number of sources. They potentially have in-factory or post-factory calibration files (such as can be created by the "pixel mapping" function) that have measured some aspect of that camera's specific sensor properties. That data is available to the in-camera accelerator and PRIME chips but not available to an post-procesing software.
What's interesting is that the "future methods" are a bit of a false hope. Every digital camera that I have owned has shown some evidence of sensor aging. Someone might create a future method for getting a more accurate, more pure RAW file but if that requires calibrating the camera to measure it's pixel-by-pixel artifacts, then that future method is of no use to old RAW files because one can't go back in time to get the calibration data.
------------
Of course, all of this is speculation because we don't actually know what noise is being reduced by the accelerator and whether the chip is destroying any information at all. That said, I do agree with you that we should worry about what the chip is doing and understand how it affects our images.
Comparing K-1 to K-1ii images is useful first step to detect some of the effects of accelerator chip but it is not a conclusive proof that the chip is damaging the image. Any difference between a K-1 and K-1ii image may be evidence of image damage by the K-1ii chip or image damage due to uncorrected noise in the K-1. (It may also be evidence of camera-to-camera sensor quality differences, too.) Even difference between K-1ii @ ISO 100 and K-1ii @ ISO 800 have interpretation difficulties. (It may also be evidence of simple shot-to-shot noise differences, too.)
The gold standard test is how closely do the K-1ii RAW files measure the true R, G, and B light levels of the scene. Measuring the exact number of electrons in each pixel is irrelevant -- electrons are not the image. The real number that matters and the true "pure" RAW file would tell us the photon flux at each pixel (which is actually different from the photon count!). There are clever statistical analyses that can spot anomalies in the RAW file to see what the chip might be doing. And there's always the possibility of cross-comparing an ISO 800 single-shot image from a K-1ii to a 3x3 Brenizer method ISO 100 pixel shift image.
Personally, I would suspect that a careful analysis of K-1 RAW and K-1ii RAW files will find: 1) the K-1ii files are more pure; 2) the two files both have artifacts but of somewhat different kinds.
|
|
Forum: Pentax K-1 & K-1 II
04-24-2018, 01:19 PM
|
| |
Interesting!
I notice that both images lack some "stars" found in the other image. But that simply highlights the problem that if the SNR gets too low, things disappear. It's also possible that some of the dim "stars" in both images are just random-chance clusters of the noise that look sufficiently like a line to fool the eye.
BTW, this kind of star-trail image would never suffer from the Sony star-eater problem which only affects unmoving single-pixel stars that the noise reducer confuses for hot pixels. It would, however, suffer from another of Sony's cooked RAW artifacts that affect very bright star trails and create a halo of blocky compression artifacts.
|
|
Forum: Pentax K-1 & K-1 II
04-24-2018, 09:55 AM
|
| |
But there's no such thing as "uncooked."
1. The lens cooks the photons of the scene by bending them hither and yon.
2. The sensor cooks the photons into electrons and then sprinkles noise and other artifacts on top.
3. The read-out and ADC circuitry cooks the electrons some more to make them into digital bits.
And then there's all the cooking by the RAW convertor which we'd like to ignore but does seem to have caused some of the problems discussed in this thread.
If anything, the accelerator is attempting to uncook the data to get something closer to the original, pure scene.
|
|
Forum: Pentax K-1 & K-1 II
04-20-2018, 08:56 AM
|
| |
It will be interesting to see if we can learn what kind of accelerator this is.
There are accelerators that are highly programmable. Math coprocessors, DSP (digital signal processing) chips, and GPUs (graphics processing unit) all have the properties of accelerating a broad class of computations (floating point, audio signal processing, graphics, respectively). Such chips can be used in many ways.
And there are accelerators that are extremely specialized and not very programmable at all. For example, there are specialized chips for doing the discrete cosine transform (DCT) that's required for jpeg and mpeg. The only thing they can do is take a block of 8x8 pixels and spit out the DCT coefficients.
It's easy to imagine versions of the accelerator chip of both kinds. More mathematical horsepower (like a DSP or GPU) enables more sophisticated software within the time limits of handling each frame. But an ultra-specialized chip might execute some especially interesting statistical corrections to the image data but have no other functionality than noise reduction that does not affect resolution.
|
|
Forum: Pentax K-1 & K-1 II
04-19-2018, 11:14 AM
|
| |
Yes, but it's a different kind of noise that what people often think about.
First, there's various types of noise intrinsic to the sensor's silicon and electronics that are added to the signal and multiplied by electronic or digital ISO gain. Even with the lens cap on, the image has noise.
Second, there's also the Poisson noise in the signal itself because the arrival of photons is not perfectly smooth. Instead, there's a random chance that more or fewer photons arrrived from some bit of the scene. Even if the sensor and electronics were perfectly noise-free, the image would still show "noise." That Poisson noise source is proportional to the square root of the light level.
For example, a pixel that should be getting a signal of exactly 1000 electrons will actually show a standard deviation of 30 electrons from shot to shot. Or, if a patch of pixels all see a uniform surface with 1000 e/pixel, there will be speckling from pixel to pixel of 30 electrons. That's a ±3% variation with some chances of ±6% or more. If you underexpose that scene by 2 stops so that each pixel should be accumulating only 250 electrons, then the Poisson noise will be about 16 electrons or about ±6% with some chances of ±12% or more.
(P.S. This physical phenomenon is one cause color noise in high-ISO shots. A gray card image that should have equal numbers of electrons in the R, G, and B pixels will show Poisson noise across the channels -- maybe the R and B pixels in one sport caught 3% more electrons but the G pixel caught 6% fewer to create a magenta blotch in the grey.)
|
|
Forum: Pentax K-1 & K-1 II
04-15-2018, 03:32 PM
|
| |
Yes, all these systems are so very very complicated.
And the "better" the software the more cleverness they will have packed into the demosaicer. But that cleverness probably has all sorts of embedded assumptions about optical sharpness, anti-aliasing filters, ISO-response, and the camera's electronics. The software may not be properly tuned for Pentax K-1ii and the new accelerator chip.
|
|
Forum: Pentax K-1 & K-1 II
04-15-2018, 02:40 PM
|
| |
So many hypotheses, so little time!
1. Some of the artifacts look like the result of using a very sharp lens on a Bayer sensor with no anti-aliasing.
2. How do the images look at ISO 100?
3. Is it possible that the "default" settings are wrong in that even the default is doing some clever sharpening-like process during demosaicing?
|
