Forgot Password
Pentax Camera Forums Home
 

Reply
Show Printable Version Search this Thread
10-03-2022, 11:58 PM   #136
Pentaxian
Fogel70's Avatar

Join Date: Sep 2009
Location: Stockholm, Sweden
Photos: Albums
Posts: 3,062
QuoteOriginally posted by clackers Quote
Completely incorrect.

A RAW image is like a spreadsheet.

It is simply a list of the individual values of each pixel.

Couldn't be further wrong!
Smaller pixels are just a smaller sample of the resulting images. On an ideal sensor pixel size do not matter (current sensor are close to being ideal within the pixel size that are used). If the same total amount of photons captured by the sensor are divided up on smaller or larger pixels does not matter for resulting noise when they are put together in the image you view.
QuoteQuote:

That is also why you can never 'see' a RAW image, only a JPG or similar preview of it after processing.

All pictures are captured as pixels, and they are rendered on paper or on screen as pixels.
LOL, just getting further into obscurity here. How much importance have the data in a RAW file that you can not see? I think you may be the only exception that care more about it than the final image.
Larger pixels are a larger sample of the image you are viewing, so they will be magnified more when viewing the image and the noise will be more noticeable than for smaller pixels that are not magnified as much when view at the same size. So for images captured with same amount of photons, visible noise in images will not differ if using smaller or larger pixels.

But with a large difference in pixel size there will be a noticeable difference in resolution.

10-04-2022, 12:27 AM   #137
Site Supporter
Site Supporter
RobA_Oz's Avatar

Join Date: Nov 2008
Location: Tasmania, Australia
Photos: Gallery
Posts: 8,182
QuoteOriginally posted by repaap Quote
Bigger OVF not needed.
Agreed, but my eyes tell me a brighter one, especially for low-light shooting, would be appreciated.

The option for a brighter OVF was there for the K-3iii, but they chose bigger, which was probably sensible given the EVF competition. However, for the K-1iii, I think the reverse is the sensible choice.
10-04-2022, 02:27 AM   #138
Loyal Site Supporter
Loyal Site Supporter




Join Date: Mar 2009
Location: Gladys, Virginia
Photos: Gallery
Posts: 27,601
QuoteOriginally posted by clackers Quote
You just don't need that extra guesswork, Rondec, because SNR is a ratio. 18% on the DXOMark test means 18% of the pixels are noise, whatever their size, whatever the sensor they're on.

Extra pixels don't get you any more light or lower noise, they just help software noise reduction in postprocessing, by averaging.
Do you really think that if you take a iso 3200 shot on a K-5 and a K-1 and print them both at 18 by 24 inches the noise will look the same between the two images, simply because the DXO Mark screen tab says they are the same? My experience is that the K-5 has quite a bit more noise when shot at high iso and printed similar size to the K-1, even though the K-1 is basically a full frame sized version of the K-5 sensor.
10-04-2022, 03:58 AM - 1 Like   #139
Pentaxian




Join Date: Oct 2010
Photos: Gallery | Albums
Posts: 6,241
QuoteOriginally posted by RobA_Oz Quote
Agreed, but my eyes tell me a brighter one, especially for low-light shooting, would be appreciated.

The option for a brighter OVF was there for the K-3iii, but they chose bigger, which was probably sensible given the EVF competition. However, for the K-1iii, I think the reverse is the sensible choice.
Indeed, brighter OVF for K-1III. What is beautiful with K-3III is brightness but also low distortion, and they did it right for many reasons. Basically K-3III reminds me of FF camera, with smaller sensor. Even if it is crop. With right lenses, there is no problems and end result is quite close to FF…with smaller footprint

10-04-2022, 04:41 AM - 5 Likes   #140
Veteran Member
noelpolar's Avatar

Join Date: Oct 2013
Location: Goolwa, SA
Photos: Gallery
Posts: 4,310
mmm.... i just got a Lieca email..... that I clicked on and had a look at.... the M11 is just under $15,000aud ....and the new 35/1.4 was a bit over $9,000aud.... very sobering. I'm off to pat my FA31.
10-04-2022, 05:20 PM - 1 Like   #141
chd
Forum Member
chd's Avatar

Join Date: Dec 2018
Location: Tucson AZ
Posts: 54
QuoteOriginally posted by Fogel70 Quote
And the data he used in his charts is even older. Cameras like Canon 5D and 20D are not that useful to use today for a noise performance. The data he used in the presentation seem to have been from 2008 and earlier.
He is teaching the theory. Examples are just illustrations.

Physics hasn't changed since then. Math hasn't changed since then. So, what has changed that makes this correct in 2016 and not correct today?
10-04-2022, 06:23 PM   #142
Site Supporter
Site Supporter
RobA_Oz's Avatar

Join Date: Nov 2008
Location: Tasmania, Australia
Photos: Gallery
Posts: 8,182
QuoteOriginally posted by chd Quote
He is teaching the theory. Examples are just illustrations.

Physics hasn't changed since then. Math hasn't changed since then. So, what has changed that makes this correct in 2016 and not correct today?
From what I read, the difference is implementation, not physics, but the only real difference I can see is that the photosites on a BSI sensor are larger than on its predecessors, so the noise threshold will be higher for the same resolution sensor. I’m not sure what crosstalk does with the boundaries being closer, though. Have I missed anything?

10-04-2022, 08:09 PM   #143
Otis Memorial Pentaxian
Otis FanOtis FanOtis FanOtis FanOtis FanOtis Fan
Loyal Site Supporter
clackers's Avatar

Join Date: Jul 2013
Location: Melbourne
Photos: Albums
Posts: 16,394
QuoteOriginally posted by Rondec Quote
Do you really think that if you take a iso 3200 shot on a K-5 and a K-1 and print them both at 18 by 24 inches the noise will look the same between the two images
All you're saying is that more pixels can be printed at larger sizes without upscaling, Rondec.

That's not a statement at all about noise in a photo.

You point out yourself, that a K-5 image is really the central half of a K-1 image, it's exactly the same, being a crop.

There is no difference in resolution, there is no difference in noise or dynamic range.

If 18% of the K-1's RAW file is made up of noise, then if you turned the dial and put it into APS-C mode and take the same picture, that RAW file will also have 18% noise pixels, they'll be exactly the same.

Last edited by clackers; 10-04-2022 at 08:30 PM.
10-04-2022, 08:17 PM   #144
Otis Memorial Pentaxian
Otis FanOtis FanOtis FanOtis FanOtis FanOtis Fan
Loyal Site Supporter
clackers's Avatar

Join Date: Jul 2013
Location: Melbourne
Photos: Albums
Posts: 16,394
QuoteOriginally posted by Fogel70 Quote
Smaller pixels are just a smaller sample of the resulting images. On an ideal sensor pixel size do not matter (current sensor are close to being ideal within the pixel size that are used). If the same total amount of photons captured by the sensor are divided up on smaller or larger pixels does not matter for resulting noise when they are put together in the image you view.
I taught physics for a living, and Fogel, you would get a fail.

Pixel size is what it's about, it doesn't matter whether they're on an APS-C, full frame or medium format sensor.

You can learn in the link below from the distinguished Teledyne company about full well capacity. You put that value back into the formula I showed you in post 118 and you can figure out the noise expressed as the SN ratio.

Learn | Full Well Capacity and Pixel Saturation.
10-04-2022, 08:28 PM   #145
Otis Memorial Pentaxian
Otis FanOtis FanOtis FanOtis FanOtis FanOtis Fan
Loyal Site Supporter
clackers's Avatar

Join Date: Jul 2013
Location: Melbourne
Photos: Albums
Posts: 16,394
QuoteOriginally posted by RobA_Oz Quote
From what I read, the difference is implementation, not physics, but the only real difference I can see is that the photosites on a BSI sensor are larger than on its predecessors, so the noise threshold will be higher for the same resolution sensor. I’m not sure what crosstalk does with the boundaries being closer, though. Have I missed anything?
I think we can now understand BSI was overhyped, Rob.

If you look back at post 118, that 61Mp Sony is slightly worse than a ten year old K-5.

And here, is DPR's proof that Fuji went *backwards* going from the XT-2 to the BSI XT-3:

Fujifilm X-T3 Review: Digital Photography Review

As per our topic, you can tell I'd prefer the next K-1 have the 42Mp sensor, but I don't think I'm going to get my wish.

Last edited by clackers; 10-04-2022 at 08:33 PM.
10-04-2022, 09:05 PM   #146
Pentaxian
Fogel70's Avatar

Join Date: Sep 2009
Location: Stockholm, Sweden
Photos: Albums
Posts: 3,062
QuoteOriginally posted by clackers Quote
All you're saying is that more pixels can be printed at larger sizes without upscaling, Rondec.

That's not a statement at all about noise in a photo.
The K5 image will produce more visible noise in the image if captured at same condition with same exposure and ISO as K1. It is easy to test, just capture an image with your K1 at high ISO and then also make a APS-C crop if it . View them both on full screen and tell us which has more visible noise.
QuoteQuote:
You point out yourself, that a K-5 image is really the central half of a K-1 image, it's exactly the same, being a crop

There is no difference in resolution, there is no difference in noise or dynamic range.

If 18% of the K-1's RAW file is made up of noise, then if you turned the dial and put it into APS-C mode and take the same picture, that RAW file will also have 18% noise pixels, they'll be exactly the same.
They are not exactly the same when viewed at the same size, an APS-C image will be magnified more when viewed at the same size as an FF image, making the noise in the APS-C more visible.

---------- Post added 05-10-22 at 06:29 ----------

QuoteOriginally posted by clackers Quote
I taught physics for a living, and Fogel, you would get a fail.
I'm still amazed that you can not make the distinction between pixel noise and visible image noise. It is like you believe that noise can only be defined and measured in one way - in one pixel.
QuoteQuote:
Pixel size is what it's about, it doesn't matter whether they're on an APS-C, full frame or medium format sensor.

You can learn in the link below from the distinguished Teledyne company about full well capacity. You put that value back into the formula I showed you in post 118 and you can figure out the noise expressed as the SN ratio.

Learn | Full Well Capacity and Pixel Saturation.

Last edited by Fogel70; 10-04-2022 at 10:24 PM.
10-04-2022, 09:37 PM   #147
Site Supporter
Site Supporter
RobA_Oz's Avatar

Join Date: Nov 2008
Location: Tasmania, Australia
Photos: Gallery
Posts: 8,182
QuoteOriginally posted by clackers Quote
I think we can now understand BSI was overhyped, Rob.

If you look back at post 118, that 61Mp Sony is slightly worse than a ten year old K-5.

And here, is DPR's proof that Fuji went *backwards* going from the XT-2 to the BSI XT-3:

Fujifilm X-T3 Review: Digital Photography Review

As per our topic, you can tell I'd prefer the next K-1 have the 42Mp sensor, but I don't think I'm going to get my wish.
DPR (tentatively) puts the increase in noise down to faster read-out, in that article. If so, blazingly-fast read speeds have their downside, it seems. What’s your take on that explanation?

Anyway, what sensor do you think will be in the K-1 successor? I’m ambivalent about the personal need for anything above what I’ve got (a K-1 and a K-3iii) but I’m aware I’ve said that sort of thing in the past, too.
10-05-2022, 12:03 AM   #148
chd
Forum Member
chd's Avatar

Join Date: Dec 2018
Location: Tucson AZ
Posts: 54
QuoteOriginally posted by Fogel70 Quote
T
I'm still amazed that you can not make the distinction between pixel noise and visible image noise.
You've answered other more recent posts in this thread, but not mine, so I repeat...

Physics hasn't changed since then [the Stanford lecture I posted earlier]. Math hasn't changed since then. So, what has changed that makes this correct in 2016 and not correct today?


The question stands.

Has there been a breakthrough in physics or math that invalidates what's presented here? Has there been a breakthrough in the manufacture of sensors that allows them to exceed these basic physical constraints?
10-05-2022, 12:47 AM   #149
Pentaxian
Fogel70's Avatar

Join Date: Sep 2009
Location: Stockholm, Sweden
Photos: Albums
Posts: 3,062
QuoteOriginally posted by chd Quote
You've answered other more recent posts in this thread, but not mine, so I repeat...

Physics hasn't changed since then [the Stanford lecture I posted earlier]. Math hasn't changed since then. So, what has changed that makes this correct in 2016 and not correct today?


The question stands.

Has there been a breakthrough in physics or math that invalidates what's presented here? Has there been a breakthrough in the manufacture of sensors that allows them to exceed these basic physical constraints?
Sensor design has changed the behavior of noise, but also pixel noise is not the same thing as visible noise in images (different ways of measuring noise). On earlier sensor designs visible noise in images was dependent on both sensor size and pixels size. With latest tech visible image noise is almost only dependent on sensor size.

Which get quite apparent when you compare images.

Measuring noise per pixel vs visible image noise, are like the difference between measuring lenses in LP/mm vs LW/PH.
For pixel noise and LP/mm resolution you need to take sensor size into account to know what to expect from the image.

Last edited by Fogel70; 10-05-2022 at 01:24 AM.
10-05-2022, 02:37 AM   #150
Loyal Site Supporter
Loyal Site Supporter




Join Date: Mar 2009
Location: Gladys, Virginia
Photos: Gallery
Posts: 27,601
QuoteOriginally posted by clackers Quote
All you're saying is that more pixels can be printed at larger sizes without upscaling, Rondec.

That's not a statement at all about noise in a photo.

You point out yourself, that a K-5 image is really the central half of a K-1 image, it's exactly the same, being a crop.

There is no difference in resolution, there is no difference in noise or dynamic range.

If 18% of the K-1's RAW file is made up of noise, then if you turned the dial and put it into APS-C mode and take the same picture, that RAW file will also have 18% noise pixels, they'll be exactly the same.
No. All I am saying is that you need to pick a target size for viewing/printing and compare all sensors at that size. If you pick a large enough size then yes, some sensors will require upscaling to hit it. What doesn't really make sense to me is to compare all sensors at 100 percent.

As to cropping the K-1 down to APS-C size, that image will be the same as the K-5 image and it will have more noise evident when printed at our chosen target print size than an uncropped K-1 image.

To me, a perfect scenario would be to print all of the images at, say, an 8 by 10 at 300 dpi spec and then compare them all. My experience with the K-5 and the K-1 said that there was about 1 EV improvement in high iso performance when viewed at the same size. As I said before, I'm not sure why you should compare the K-5 at 10 by 16 and the K-1 on a 24 by 16 print. To me, you should either downscale the K-1 or upscale the K-5 to see how the sensors perform.
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!
24x36mm, 36mp, buffer, camera, cameras, cost, full-frame, hope, lenses, light, lot, mark, mp, pentax, pentax k-1 mark, people, pm, post, press, race, release, ricoh, sensor, sensors, sony sensors, thanks, version
Thread Tools Search this Thread
Search this Thread:

Advanced Search


Similar Threads
Thread Thread Starter Forum Replies Last Post
Master your Pentax K-3 Mark III with the Pentax K-3 III eBook Adam Pentax K-3 III 43 1 Day Ago 08:58 AM
Does the K1 Mark II still make sense to buy today? Is there an Mark III in the works? davidphoto Pentax Full Frame 161 07-18-2023 03:32 PM
K-70 upgrade - K-1 Mark II or K-3 mark III? Emirena Pentax DSLR Discussion 35 11-28-2021 09:40 PM
K3 mark III (not mark II !)& Sigma 50-500 (later version) - any experience or issues? jeallen01 Pentax K-3 III 7 08-14-2021 01:51 AM
K-3 Mark III and GR III topping one Japanese sales chart JPT Photographic Industry and Professionals 27 07-31-2021 07:22 AM



All times are GMT -7. The time now is 02:26 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