-
Welcome to the new forums! Please read this first. For known issues we are working to resolve, click here.
Re: High MP count? Bring it in
- Thread starter ET2
- Start date
Iliah Borg
Forum Pro
ejmartin
Veteran Member
The data in the last bit here is coming from the fact that Nikon clips blacks, which imprints data from higher bits into the last bit. This was explained long ago:
http://forums.dpreview.com/forums/read.asp?forum=1021&message=28196804
There is info in the last bit because it has been imprinted from data in higher bits, ie it is a copy of data in higher bits, not new independent data in the last bit.
--
emil
--
http://theory.uchicago.edu/~ejm/pix/20d/
ZoranC
Veteran Member
10 bits? Bah humbag. I hear some people still use Windows 3.1 on matching video cards. No more than 8 bits is ever needed.Here is, for entertainment, last bit of a 12-bit camera. Why do we have 12 bits in it anyway? Let's reduce it to just 10.You're just counting angels on the head of a pin... Iliah is saying there are 433 and you're saying that it's only 432 and 433rd one is not a real angel
Iliah Borg
Forum Pro
You have forgotten 640KB RAM, or was it just 64KB?10 bits? Bah humbag. I hear some people still use Windows 3.1 on matching video cards. No more than 8 bits is ever needed.Here is, for entertainment, last bit of a 12-bit camera. Why do we have 12 bits in it anyway? Let's reduce it to just 10.You're just counting angels on the head of a pin... Iliah is saying there are 433 and you're saying that it's only 432 and 433rd one is not a real angel
--
http://www.libraw.org/
Iliah Borg
Forum Pro
In the image that was overexposed (shot right after this one) the last bit contains no image data.
--
http://www.libraw.org/
Thanks Emil Martinec ... Your saved the day
And your article on your site
http://theory.uchicago.edu/~ejm/pix/20d/tests/noise/noise-p3.html
is the most informative on the topic.
Iliah Borg
Forum Pro
ROTFLMAO You really do not understand what my answer here http://forums.dpreview.com/forums/read.asp?forum=1042&message=39634256 means?Thanks Emil Martinec ... Your saved the day
--
http://www.libraw.org/
Not true: "The data in the last bit here is coming from the fact that Nikon clips blacks, which imprints data from higher bits into the last bit."
http://forums.dpreview.com/forums/read.asp?forum=1042&message=39634121
There is no benefit to 14-bit RAW, and it is especially true once you leave the base ISO and go higher (and by the way Thorn Hogan seemed to think 14-bit is more helpful at higher ISO ... LOL. The complete opposite. That just goes to show that he has no real understanding about the topic).
Iliah Borg
Forum Pro
Right, because it does not suite your agenda
http://forums.dpreview.com/forums/read.asp?forum=1042&message=39634256
--
http://www.libraw.org/
ZoranC
Veteran Member
64KB?! You are wasteful! Let's go all the way! Timex Sinclair ZX-80 baybeee! Whole 1 KB! OK, block characters but it proves we don't need no stinkin 14-bit RAW filesYou have forgotten 640KB RAM, or was it just 64KB?10 bits? Bah humbag. I hear some people still use Windows 3.1 on matching video cards. No more than 8 bits is ever needed.Here is, for entertainment, last bit of a 12-bit camera. Why do we have 12 bits in it anyway? Let's reduce it to just 10.You're just counting angels on the head of a pin... Iliah is saying there are 433 and you're saying that it's only 432 and 433rd one is not a real angel
Iliah Borg
Forum Pro
Oh you so right. Back to lilliputians we go.64KB?! You are wasteful! Let's go all the way! Timex Sinclair ZX-80 baybeee! Whole 1 KB! OK, block characters but it proves we don't need no stinkin 14-bit RAW filesYou have forgotten 640KB RAM, or was it just 64KB?10 bits? Bah humbag. I hear some people still use Windows 3.1 on matching video cards. No more than 8 bits is ever needed.Here is, for entertainment, last bit of a 12-bit camera. Why do we have 12 bits in it anyway? Let's reduce it to just 10.You're just counting angels on the head of a pin... Iliah is saying there are 433 and you're saying that it's only 432 and 433rd one is not a real angel
--
http://www.libraw.org/
ejmartin
Veteran Member
Yes, it means that data in the last bit is not correlated with image data in the higher bits, as it was in the image you posted. In neither case is there independent information in the last bit.ROTFLMAO You really do not understand what my answer here http://forums.dpreview.com/forums/read.asp?forum=1042&message=39634256 means?Thanks Emil Martinec ... Your saved the day
--
emil
--
http://theory.uchicago.edu/~ejm/pix/20d/
Iliah Borg
Forum Pro
Correct. It is independent image data, and it depends on exposure and scene dynamic range, as it should.Yes, it means that data in the last bit is not correlated with image data in the higher bitsROTFLMAO You really do not understand what my answer here http://forums.dpreview.com/forums/read.asp?forum=1042&message=39634256 means?Thanks Emil Martinec ... Your saved the day
--
http://www.libraw.org/
ejmartin
Veteran Member
Some Nikons (eg the D300) show improved shadow performance in 14-bit mode. This doesn't necessarily mean that it is due to an intrinsic advantage of the higher bit depth, rather 14-bit and 12-bit capture modes are done quite differently, as evidenced by the change in frame rate; the improvement is due to different data capture and processing, and not due to the bit depth of the data storage. This is easily verified by taking a 14-bit capture and truncating it to 12 bits, and comparing the result to the camera set to 12-bit capture.
--
emil
--
http://theory.uchicago.edu/~ejm/pix/20d/
ejmartin
Veteran Member
Again: In neither case is there independent information in the last bit, not already contained in higher bits.Correct. It is independent image data, and it depends on exposure and scene dynamic range, as it should.Yes, it means that data in the last bit is not correlated with image data in the higher bitsROTFLMAO You really do not understand what my answer here http://forums.dpreview.com/forums/read.asp?forum=1042&message=39634256 means?Thanks Emil Martinec ... Your saved the day
--
emil
--
http://theory.uchicago.edu/~ejm/pix/20d/
Iliah Borg
Forum Pro
Rhetorical questions: how many bits can be stripped from 12-bit capture? From 14-bit capture? That is, without the loss of image quality?
Non-rhetorical: do you know that adding dependent values to raw file results in regular artifacts?
--
http://www.libraw.org/
Another poster on Pentax SLR forum (Gordon) did exactly that with K-5, (truncated 14-bit into 12-bit) and found no difference in IQ
http://forums.dpreview.com/forums/read.asp?forum=1036&message=37015135
RussellInCincinnati
Senior Member
Julio: You can't give him heck for not posting an example (camera model is irelevant)
Camera model irrelevant? Professor Martinec's (you know, the physics professor Tony Beach so enjoys vaguely name-calling) and much of the discussion was about a Nikon D3. Of course there are some cameras at some ISOs that have way more than 12 bits of useful data. Recall that those guys (ET2, Tony Beach etc) were arguing about a Nikon D3 at ISO 400+.
Julio: and then give him heck for the example he posted all in the same post. At least use two differnt replies
In the context of the questions about examples from a Nikon D3 at ISO 400 or greater, it was odd for Iliah to post a picture that did not show EXIF data (or any documentation) as being from a Nikon D3 at ISO 400 or greater. Rather than "giving him heck", I would characterize my sentence on the matter as a rather bland note of Iliah's missing documentation.
Julio:Any image that shows data beyond 12-bits is bound to show subtle differences. If it were something obvious that just jumps out and says "boo!" there wouldn't be a discussion at all.
Funny, that's exactly how I feel. That it's so obvious that there's nothing of ordinary usefulness in more than 12 bits per pixel bit raw data on (for example) the D3 at ISO 400 or greater, that there really is no discussion at all.
Julio: Saying that a 10EV push is extreme and not in the realm of normal processing is redundant.
You do not share my view that if 14 bits of data on the ISO 400+ D3 had the slightest bit of real world importance worth arguing about, that we would be able to see the difference in some image that somebody would value?
Julio: You're just counting angels on the head of a pin... Iliah is saying there are 433 and you're saying that it's only 432 and 433rd one is not a real angel
And all this stemming from a thread about the obvious extra defects of more magnified files when viewed at 100% magnification!/
We agree again, it is indeed odd how many of us tend to mix up image "defects" that are easily detectable at an increasingly wild 1:1 magnification (corresponding to what, about a 4 or 5 foot wide monitor size nowadays?), with camera problems that some ordinary viewer of most images would ever care about.
Camera model irrelevant? Professor Martinec's (you know, the physics professor Tony Beach so enjoys vaguely name-calling) and much of the discussion was about a Nikon D3. Of course there are some cameras at some ISOs that have way more than 12 bits of useful data. Recall that those guys (ET2, Tony Beach etc) were arguing about a Nikon D3 at ISO 400+.
Julio: and then give him heck for the example he posted all in the same post. At least use two differnt replies
In the context of the questions about examples from a Nikon D3 at ISO 400 or greater, it was odd for Iliah to post a picture that did not show EXIF data (or any documentation) as being from a Nikon D3 at ISO 400 or greater. Rather than "giving him heck", I would characterize my sentence on the matter as a rather bland note of Iliah's missing documentation.
Julio:Any image that shows data beyond 12-bits is bound to show subtle differences. If it were something obvious that just jumps out and says "boo!" there wouldn't be a discussion at all.
Funny, that's exactly how I feel. That it's so obvious that there's nothing of ordinary usefulness in more than 12 bits per pixel bit raw data on (for example) the D3 at ISO 400 or greater, that there really is no discussion at all.
Julio: Saying that a 10EV push is extreme and not in the realm of normal processing is redundant.
You do not share my view that if 14 bits of data on the ISO 400+ D3 had the slightest bit of real world importance worth arguing about, that we would be able to see the difference in some image that somebody would value?
Julio: You're just counting angels on the head of a pin... Iliah is saying there are 433 and you're saying that it's only 432 and 433rd one is not a real angel
And all this stemming from a thread about the obvious extra defects of more magnified files when viewed at 100% magnification!/
We agree again, it is indeed odd how many of us tend to mix up image "defects" that are easily detectable at an increasingly wild 1:1 magnification (corresponding to what, about a 4 or 5 foot wide monitor size nowadays?), with camera problems that some ordinary viewer of most images would ever care about.
A non-rhetorical question: The newer Sony's 24 MP sensor has smaller pixels than the old 16MP sensor. Wouldn't that make 14-bit even more useless? In other words, is there a correlation between pixel size and bits required to capture data?Again: In neither case is there independent information in the last bit, not already contained in higher bits.Correct. It is independent image data, and it depends on exposure and scene dynamic range, as it should.Yes, it means that data in the last bit is not correlated with image data in the higher bitsROTFLMAO You really do not understand what my answer here http://forums.dpreview.com/forums/read.asp?forum=1042&message=39634256 means?Thanks Emil Martinec ... Your saved the day
Keyboard shortcuts
- f
- Forum