Ad
Ad
Ad
Pages: « 1 ... 13 14 [15] 16 17 »   Bottom of Page
Print
Author Topic: Do you hate HDR too?  (Read 261368 times)
JohnKoerner
Guest
« Reply #280 on: September 16, 2011, 05:55:22 PM »
ReplyReply

It's a very nuanced discussion, isn't it Jack?  Grin

 Grin




The issue with using the final image as the determination of whether an image is HDR or not is that the final image isn't HDR either.  Take a camera like the K5 or D7000 that can capture around 12 stops of brightness.  Most monitors can't reproduce that nor can any printing media.  So the final image, the one that appears 'proper' on screen or is printable doesn't have all the DR that the camera was able to capture.  It's been edited or tonemapped down into something narrower that can be used.  Think of it like a Zone System image where the exposure was increased to generate desired highlight detail.  If the scene DR was, say, 8 stops and the shadow detail was lifted 2 stops, you're most likely going to have blown out highlights without adjusted development of the film.  But if the film undergoes contracted development, those highlights can be brought back into range.  That's one form of tonemapping.  The contracted development maps highlight tones from higher to lower zones.  Then, in printing, additional techniques would be used to render a pleasing print taking into account the fact that the paper couldn't reproduce all 8 stops of brightness.  That's another form of tonemapping.  But that final printed image doesn't contain all of the brightness range that was captured initially.  So if we use the final image as the determination, until technology changes no image would be considered HDR.

Hmmm, curious to what the point of it all is, then? Onscreen monitor appreciation?

Maybe that is further justification of focus-stacking versus HDR "tonemapping" is at least all of the work I put into focus-stacking can be seen in a print!

Jack



.
Logged
Eric Myrvaagnes
Sr. Member
****
Offline Offline

Posts: 7970



WWW
« Reply #281 on: September 16, 2011, 06:59:36 PM »
ReplyReply

I think there is one point about the proper definition of HDR that all posters here should be able to agree on. In the words of the eminent philosopher Lewis Carroll,
“When I use a word," Humpty Dumpty said in rather a scornful tone, "it means just what I choose it to mean - neither more nor less.”

Actually, I've learned quite a bit from this thread. And the discussion has remained nicely civil.

Eric
Logged

-Eric Myrvaagnes

http://myrvaagnes.com  Visit my website. New images each season.
jeremypayne
Guest
« Reply #282 on: September 17, 2011, 06:30:06 AM »
ReplyReply

focus-stacking versus HDR "tonemapping"
Jack

Where does "versus" come into the picture with these two techniques?  It isn't an "either/or" ... or a competition.

That's like saying "a ladder vs a dirt bike".

Both techniques have their place ... If by technique we mean "bracketing for DR/SNR" when we say HDR ... But are tackling completely different uses cases.
Logged
Guillermo Luijk
Sr. Member
****
Offline Offline

Posts: 1290



WWW
« Reply #283 on: September 17, 2011, 07:08:38 AM »
ReplyReply

Where does "versus" come into the picture with these two techniques?

Maybe 'versus' was not the most appropiate word. It's rather a conceptual comparision of techniques: Focus Stacking versus HDR?.
Logged

JohnKoerner
Guest
« Reply #284 on: September 17, 2011, 09:53:42 AM »
ReplyReply

Where does "versus" come into the picture with these two techniques?  It isn't an "either/or" ... or a competition.
That's like saying "a ladder vs a dirt bike".


I disagree and I am not sure what you're quibbling about here. No, it is not like saying a ladder versus a dirt bike. A ladder has no similarity to a dirt bike, it isn't even in the same ballpark, whereas there are plenty of similarities between focus stacking and HDR image making.

I was not trying to put focus stacking "in competition" with HDR, so I am not sure where you pulled that out of. Did you just wake up on the wrong side of the bed or something?

The "versus" clearly means to show the differences between the two seemingly similar techniques. It is a manner of speech: "Focus Stacking versus HDR."  I could also have said, "Focus Stacking as opposed to HDR," which is another manner of speech, and essentially means the same thing: namely to show the differences in two seemingly similar subjects.




Both techniques have their place ... If by technique we mean "bracketing for DR/SNR" when we say HDR ... But are tackling completely different uses cases.

I realize both techniques have their place. Speaking of quibbling over word use, ending a fragmented sentence with "uses cases" isn't exactly stellar English either, Jeremy.

Jack


.
Logged
jeremypayne
Guest
« Reply #285 on: September 17, 2011, 10:32:54 AM »
ReplyReply

BLAH BLAH

It was a typo, dude - get over yourself.  You'd argue with a doorknob ... and lose.
Logged
Peter_DL
Sr. Member
****
Offline Offline

Posts: 421


« Reply #286 on: September 17, 2011, 11:51:38 AM »
ReplyReply

Actually, I've learned quite a bit from this thread. And the discussion has remained nicely civil.

Well, Eric,
I’d have been happy to accept this as a final word,
but then, the initial question was already quite upfront:  Do you hate HDR too ?
And if I remember correctly, some agreement with such rejection was expressed as well.

Now, as we have learned HDR = bracketed exposures,
and so it seems that some people don’t like when others do exposure bracketing, if this is the definition for HDR  Huh

Peter

--
Logged
JohnKoerner
Guest
« Reply #287 on: September 17, 2011, 12:37:32 PM »
ReplyReply

It was a typo, dude - get over yourself.  You'd argue with a doorknob ... and lose.


Well, at least you got half of it right, Jeremy--and I think you've taken a large step toward recovery by recognizing that you're displaying the intellect of a door knob.

Perhaps I can be of further assistance in pointing out the fact you seem to have an pathological obsession with "competition": first you thought I was placing Focus Stacing in competition with HDR (wrongly so), and now you think you are in some sort of argumentative competition with me (again, wrongly so). In fact, I don't believe I originally addressed you at all until you started this bizarre off-topic digression of yours.

That said, I will leave you to your perceived "victory" sir, but the truth is the only thing you "won" here is creating a new low watermark for the word "petty."

Jack



.
Logged
Slobodan Blagojevic
Sr. Member
****
Offline Offline

Posts: 5890


When everybody thinks the same... nobody thinks.


WWW
« Reply #288 on: September 17, 2011, 01:13:17 PM »
ReplyReply

... And the discussion has remained nicely civil.

Looks like you jinxed it Wink
Logged

Slobodan

Flickr
500px
LKaven
Sr. Member
****
Offline Offline

Posts: 817


« Reply #289 on: September 17, 2011, 02:51:09 PM »
ReplyReply

Sometimes "versus" does sound like "as opposed to" making it an either/or proposition, but there is also a sense in which versus also means "as contrasted with."  The misunderstandings here are ... understandable.  Everything that came after that I think we can just dissipate as steam. 

In the end, the merits of "doing HDR," in whatever pragmatic semantics we have, come down to whether or not there is an artistic purpose served by it. 
Logged

Eric Myrvaagnes
Sr. Member
****
Offline Offline

Posts: 7970



WWW
« Reply #290 on: September 17, 2011, 04:48:58 PM »
ReplyReply

Looks like you jinxed it Wink
Yup. Sorry about that.

Maybe we need a new thread about "Do you hate 'versus' too?"   Grin
« Last Edit: September 17, 2011, 04:50:42 PM by Eric Myrvaagnes » Logged

-Eric Myrvaagnes

http://myrvaagnes.com  Visit my website. New images each season.
Monito
Jr. Member
**
Offline Offline

Posts: 96



WWW
« Reply #291 on: September 28, 2011, 05:47:51 PM »
ReplyReply

Simple example, 2 bits:
int: [0 1 2 3] (available codes)
float: [-1 -0.25 0.25 1] (available codes)

What is the _number_ of shades available in each? The answer is 4, 2^2 for both. What kind of data can be stored within them? Basically any data. But floating point is a lot easier to work with for many kinds of data and operations. The distribution of available numbers and the quantization error makes a lot of sense for many tasks.

No, it doesn't work like that.  Your example is just a wierdly shifted encoding.

A simple illustration would be 8 bits.   The integer version would be 0 ... 255.  The floating point version might be 4 bits main part (mantissa or significand) plus 4 bits exponent for a total of 8 bits.  But the exponent would vary from 0 to 15 (four bits).  The mantissa represents a fraction, with a hidden bit (always included in the math) in what would be the fifth bit (most significant bit).  That is added to the fraction, so the fraction ranges from 1.0 to nearly 2.0.  But if you put a 1 in the least bit of the exponent, then you multiply those values by two.  So:

0000  0000 = 1.0 x 2^0 = 1.0 x 1 = 1.0
0001  0000 = 1 1/16 x 2^0 = 1.0625
1111  0000 = 1 15/16 x 2^0 = 1.9375

0000  0001 = 1.0 x 2^1 = 1.0 x 2 = 2.0
0001  0001 = 1.0625 x 2 = 2.125
1111  0001 = 1.9375 x 2 = 3.875

0000  1000 = 1.0 x 2^8 = 1.0 x 128 = 128
0001  1000 = 1 1/16 x 128 = 1.0625 x 128 = 136
1111  1000 = 1 15/16 x 128 = 1.9375 x 128 = 248

0000  1111 = 1.0 x 2^15 = 1.0 x 32768 = 32768
0001  1111 = 1.0625 x 32768 = 34816
1111  1111 = 1.9375 x 32768 = 63488

So you can see that an 8 bit floating point number (in 4,4 form) has a much greater range than an 8 bit integer:   0 ... 63488 versus 0 ... 255.

That is why the floating point numbers used for 32 bit floating point colour have three 8 bit mantissas and an 8 bit exponent.  This is how they can store a huge dynamic range.

Yes, it does have to be compressed for viewing, which can lead to some posterization problems if not done well, but it does allow you to render or capture a scene with a high dynamic range and display it in a meaningful and sometimes beautiful way.
« Last Edit: September 28, 2011, 05:50:47 PM by Monito » Logged

MonitoPhoto (Landscape, Architecture, Portraits: Halifax, Nova Scotia)
Eric Myrvaagnes
Sr. Member
****
Offline Offline

Posts: 7970



WWW
« Reply #292 on: September 28, 2011, 11:16:00 PM »
ReplyReply

Sorry, Monito. Your illustration doesn't contradict Hjulenissen's assertion, which I understand to be addressing the number of data points expressible in equal numbers of bits in integer versus floating point. In your example, the range of values is greater for the floating point representation, but the number of distinct values representable is still 2 to the 8th power, or 256 in either case.

Eric
« Last Edit: September 28, 2011, 11:18:32 PM by Eric Myrvaagnes » Logged

-Eric Myrvaagnes

http://myrvaagnes.com  Visit my website. New images each season.
Monito
Jr. Member
**
Offline Offline

Posts: 96



WWW
« Reply #293 on: September 29, 2011, 03:55:13 AM »
ReplyReply

Ah, I see now what you and Hjulenissen are getting at, Eric.  Thanks.
Logged

MonitoPhoto (Landscape, Architecture, Portraits: Halifax, Nova Scotia)
Eric Myrvaagnes
Sr. Member
****
Offline Offline

Posts: 7970



WWW
« Reply #294 on: September 29, 2011, 09:29:29 AM »
ReplyReply

Ah, I see now what you and Hjulenissen are getting at, Eric.  Thanks.
I must admit that his post was a little terse for clear understanding.

Eric
Logged

-Eric Myrvaagnes

http://myrvaagnes.com  Visit my website. New images each season.
hjulenissen
Sr. Member
****
Offline Offline

Posts: 1679


« Reply #295 on: September 29, 2011, 04:00:38 PM »
ReplyReply

I must admit that his post was a little terse for clear understanding.

Eric
"terse" is not in my vocabulary.

Anyways, the discussion was a fundamental disagreement about the "number of colors" available in integer vs floating point numbers. My initial reply (that was not accepted) was a lot more compact, something ala: given a N-bit data element, you can encode up to N bits of information. This means up to N different "colors". How those bits are to be interpreted, how they will physically map to camera/display/printer hardware, and how preceptually relevant all of those colors are going to be is a question of how that information is decoded. If desired, a 32-bit floating point number can be "converted" to a 32-bit integer (not changing one bit) and back again without loss of colors: i.e. 32bit integers can hold the same information as 32bit floats.

A practical example is using integers to represent gamma-encoded values. In this case, the numerical precision is spent in the darker tones, just like floating point, and like images should be encoded.

-h
Logged
Eric Myrvaagnes
Sr. Member
****
Offline Offline

Posts: 7970



WWW
« Reply #296 on: September 29, 2011, 04:31:56 PM »
ReplyReply

Hi h,

It was totally clear to me the first time around, so I was a bit surprised that some others didn't seem to get it, a fact that I attributed to the myth that (digital) floating point numbers can represent vastly more different "numbers" than can integers of the same length.

Regards,

Eric
Logged

-Eric Myrvaagnes

http://myrvaagnes.com  Visit my website. New images each season.
Monito
Jr. Member
**
Offline Offline

Posts: 96



WWW
« Reply #297 on: September 29, 2011, 05:05:44 PM »
ReplyReply

Granted that a 32 bit floating point color number has as many colours as a 32 bit integer colour number.  It however has 256 times as many colours as a 24 bit colour integer (8 x 3 channels), plus more range.  Our images use 24 bits (in 32 bit containers).
Logged

MonitoPhoto (Landscape, Architecture, Portraits: Halifax, Nova Scotia)
LKaven
Sr. Member
****
Offline Offline

Posts: 817


« Reply #298 on: September 30, 2011, 01:10:07 AM »
ReplyReply

Granted that a 32 bit floating point color number has as many colours as a 32 bit integer colour number.  It however has 256 times as many colours as a 24 bit colour integer (8 x 3 channels), plus more range.  Our images use 24 bits (in 32 bit containers).
But we're discussing image data having 3 x 32 bit floats, or 32 bits per channel, 96 bits per pixel.  It represents both astronomical dynamic range and accommodates fractional results with sufficient precision.
Logged

hjulenissen
Sr. Member
****
Offline Offline

Posts: 1679


« Reply #299 on: September 30, 2011, 01:10:54 AM »
ReplyReply

Granted that a 32 bit floating point color number has as many colours as a 32 bit integer colour number.  It however has 256 times as many colours as a 24 bit colour integer (8 x 3 channels), plus more range.  Our images use 24 bits (in 32 bit containers).
I think that a thorough discussion of pros and cons of integers vs floating points is beyond the scope of this site and this thread. It introduce a lot of details that most of us simply does not have to think about. My point was simply that one should not make choices of format only based on it being float or int.

-h
Logged
Pages: « 1 ... 13 14 [15] 16 17 »   Top of Page
Print
Jump to:  

Ad
Ad
Ad