Ad
Ad
Ad
Pages: « 1 ... 10 11 [12] 13 14 ... 17 »   Bottom of Page
Print
Author Topic: Do you hate HDR too?  (Read 255805 times)
hjulenissen
Sr. Member
****
Offline Offline

Posts: 1666


« Reply #220 on: September 11, 2011, 11:47:42 AM »
ReplyReply

H, don't disagree with the concept that the images started out in an integer space and will end up in an integer space.  It's what happens in between that's the issue and where I think I'm not completely following your line of thinking. 
32 bits is 32 bits. If a 32-bit floating point number could represent more than 32 bits worth of information, it would be an information-theoretical break-through.

-h
Logged
Guillermo Luijk
Sr. Member
****
Offline Offline

Posts: 1274



WWW
« Reply #221 on: September 11, 2011, 11:49:55 AM »
ReplyReply

What do I mean by such a broad scale?  Beyond what cameras can capture.

What 'cameras can capture' vary from camera to camera and from year to year, it's not that difficult a fact to grasp.


In both your scenario A and Scenario B there's no need for either HDR or XDR.  You're not gaining anything by using either HDR software, automated blending software or manual blending with layers.  It's a moot point.

You insist in being wrong Bob. Of course you are gaining something with the HDR software and with the blending, you are tonemapping the captured information, and that is the gain you have, display in the output format all the input information. Something which has always to be done someway since capture devices are linear and do not perform any form of local contrast arrangements.

So what you gain using HDR software to tonemap a {-2, 0, +2} bracket from a 8 stops camera over a 12 stops scene using HDR tonemapping software, is exacly the same as you will be gaining by tonemapping a single shot from a 12 stops camera over the same 12 stops scene using the same HDR tonemapping software.

You suggest you will have no problem to process HDR captures when cameras can capture the entire DR of any scene in a single shot, andt there will be nothing to gain. WRONG, you will have the same challenges as you have today, the need to tonemap all that captured information.

Your conception of HDR, the 32-bit floating point formats (BTW there are many 32-bit floating point formats, you regard them as a unique standard that take part in the definition of HDR), and need of bracketing is conceptually errated. Perhaps your lack of math knowledge prevents you from understanding that the real problem of HDR is not the capture of information in a high DR scene, but its later processing to adapt it to the output devices.

Regards
Logged

digitaldog
Sr. Member
****
Offline Offline

Posts: 8628



WWW
« Reply #222 on: September 11, 2011, 11:51:26 AM »
ReplyReply

Andrew, what are the advantages?  Depends.  Maybe there are none.  Horses for courses, as they say.  And if I misread your comments then I apologise.

I’m looking for clarity on the terminology. On one had, we hear Enfuse is not true HDR. Its exposure blending. We hear HDR is only true when 32 bit processing comes into play.

Quote
It's not true HDR; however.  The images don't enter the 32 bit space but are retained in the native bit depth.
Why is HDR only true when one uses 32 bits? Is that why Enfuse isn’t HDR and what is the benefit of 32 bit, “true” HDR over what Enfuse provides? Depends on what?
Logged

Andrew Rodney
Author “Color Management for Photographers”
http://digitaldog.net/
RFPhotography
Guest
« Reply #223 on: September 11, 2011, 12:57:12 PM »
ReplyReply

GL, if you continue to insist on using terms like 'wrong' then this discussion need go no further.

Yes, what cameras can capture varys from model to model and year to year as technology improves.  Where did I ever say that wasn't the case?  That's why there's no hard and fast number.  That's why it's not possible to say 11.4756947865740304575694 stops is the cutoff.  It absolutely depends on the camera being used.  If I'm using a K5 to shoot a scene that has 9 stops of brightness and someone else using, say, a Canon 10D, then I don't need to use anything other than a single shot and I don't need to do anything but edit (tonemap) a single shot but the Canon user will have to use some other methodology of capture if s/he wants to record the entire 8 stop range.  One person has no need for HDR/XDR, the other person does.  And let's be clear about something else as well.  Even if I take a single shot, process it multiple times and use those multiple layers to help with my editing (tonemapping), it's still not either HDR or XDR because nothing beyond what the camera captured is being created.  HDR/XDR are about capture and are methodologies to compensate for a capture medium that may be, in some situations, limited relative to what's being photographed.  Simple.  Maybe if you'd pull your head out of your algebra for a moment it would become a bit clearer.    Grin

I understand that there may be a need to edit (tonemap) the 8 or 11 stop scene into something narrower for reproduction on screen or in a print.  But you don't need to bracket and you don't need to use HDR or other blending methods to do that in the examples you gave.  The 'need' for bracketing is not, in photographic terms, erroneous.  You're missing the point of what I'm trying to get at.  If I have a camera that can capture 104,475 stops of light then there is no need for HDR or XDR or any other 'expanded' capture methodology.  What is needed then is to edit (tonemap) that 104,475 stops into something that can be used.  Tonemapping isn't an HDR-only term.  Tonemapping is simply a fancy word for editing.  I don't need to use the tonemapping operators of HDR software to 'tonemap' an HDR image.  I can do it quite well with the regular tools in Photoshop.  It's still tonemapping because tonemapping is just editing.

And just for the sake of clarity, people have been tonemapping since long before there was linear capture.  These concepts aren't new to digital photography.
Logged
hjulenissen
Sr. Member
****
Offline Offline

Posts: 1666


« Reply #224 on: September 11, 2011, 01:02:14 PM »
ReplyReply

Yes, what cameras can capture varys from model to model and year to year as technology improves.  Where did I ever say that wasn't the case?  That's why there's no hard and fast number.  That's why it's not possible to say 11.4756947865740304575694 stops is the cutoff.  It absolutely depends on the camera being used.  If I'm using a K5 to shoot a scene that has 9 stops of brightness and someone else using, say, a Canon 10D, then I don't need to use anything other than a single shot and I don't need to do anything but edit (tonemap) a single shot but the Canon user will have to use some other methodology of capture if s/he wants to record the entire 8 stop range.  One person has no need for HDR/XDR, the other person does.  And let's be clear about something else as well.  Even if I take a single shot, process it multiple times and use those multiple layers to help with my editing (tonemapping), it's still not either HDR or XDR because nothing beyond what the camera captured is being created.  HDR/XDR are about capture and are methodologies to compensate for a capture medium that may be, in some situations, limited relative to what's being photographed.  Simple.  Maybe if you'd pull your head out of your algebra for a moment it would become a bit clearer.    Grin
...
So what you are sayiing is that a bracketed sequence from a Canon 10D resulting in a 10-stops DR is HDR, while a single-shot image from a Pentax K5 having the same 10 stops of DR is not HDR?

-h
Logged
digitaldog
Sr. Member
****
Offline Offline

Posts: 8628



WWW
« Reply #225 on: September 11, 2011, 01:07:09 PM »
ReplyReply

So what you are sayiing is that a bracketed sequence from a Canon 10D resulting in a 10-stops DR is HDR, while a single-shot image from a Pentax K5 having the same 10 stops of DR is not HDR?

Seems pretty reasonable to me.
Logged

Andrew Rodney
Author “Color Management for Photographers”
http://digitaldog.net/
LKaven
Sr. Member
****
Offline Offline

Posts: 788


« Reply #226 on: September 11, 2011, 02:10:18 PM »
ReplyReply

32 bits is 32 bits. If a 32-bit floating point number could represent more than 32 bits worth of information, it would be an information-theoretical break-through.
In many ways, you are correct here.  But the allocation of bits is different between the two coding strategies in such a way as to even out the actual numerical resolution independent of scale.  You have a finer resolution on the "smaller" numbers using floating point.
Logged

hjulenissen
Sr. Member
****
Offline Offline

Posts: 1666


« Reply #227 on: September 11, 2011, 02:22:50 PM »
ReplyReply

In many ways, you are correct here.  But the allocation of bits is different between the two coding strategies in such a way as to even out the actual numerical resolution independent of scale.  You have a finer resolution on the "smaller" numbers using floating point.
Sure. But a gamma-encoded integer (such as most LDR image formats) will give you something similar. Its all a matter of how the bits are to be interpreted.

Bob did not seem to understand how integer representation of images could have as "many colors" as floating point representations.

-h
Logged
RFPhotography
Guest
« Reply #228 on: September 11, 2011, 05:42:19 PM »
ReplyReply

So what you are sayiing is that a bracketed sequence from a Canon 10D resulting in a 10-stops DR is HDR, while a single-shot image from a Pentax K5 having the same 10 stops of DR is not HDR?

-h

Yes.  That's exactly what I'm saying.  If it falls within the capture range of the sensor and no other means are needed to capture the entire brightness range of the scene then it becomes 'normal' dynamic range.  It's within the 'normal' range of what the sensor can capture.

And back to one of GL's points earlier:  No, I'm not using HDR as a proxy for all 32 bit image formats.  I'm using HDR as High Dynamic Range the method, distinct from the 32 bit image formats such as .hdr, .exr, .tiff, .psd, .bef, etc.  If you'd prefer, and to avoid future confusion, I can use HDRI rather than HDR. 
Logged
joofa
Sr. Member
****
Offline Offline

Posts: 486



« Reply #229 on: September 12, 2011, 03:20:40 PM »
ReplyReply

That's exactly what I'm saying.  If it falls within the capture range of the sensor and no other means are needed to capture the entire brightness range of the scene then it becomes 'normal' dynamic range.  It's within the 'normal' range of what the sensor can capture.

I think you are correct. This is what I understand as one way of describing HDR.

Sincerely,

Joofa
Logged

Joofa
http://www.djjoofa.com
Download Photoshop and After Effects plugins
LKaven
Sr. Member
****
Offline Offline

Posts: 788


« Reply #230 on: September 12, 2011, 04:55:38 PM »
ReplyReply

Trying to come up with either an analytical or 'dictionary' definition for HDR I think is somewhat fruitless.  It surely isn't an analytical term with an essential character.  As a 'natural kind' term, it is subject to the ins and outs of species definitions.  But among the commonplaces of HDR one should include the idea of /supersampling/ and the associated kinds of processes that this enables.  Normalized averaging of supersampled image data tends to produce representations that are most comfortably accommodated in floating point space, this partly due to the variable white/black points which dictate the need for increased precision, whether that increased precision is implemented in floating points quantities, or "wide" integers. 
Logged

Guillermo Luijk
Sr. Member
****
Offline Offline

Posts: 1274



WWW
« Reply #231 on: September 12, 2011, 06:16:19 PM »
ReplyReply

So what you are sayiing is that a bracketed sequence from a Canon 10D resulting in a 10-stops DR is HDR, while a single-shot image from a Pentax K5 having the same 10 stops of DR is not HDR?

Seems pretty reasonable to me.

Seems oversimplistic to me Andrew; the same as the need for floating point formats. To begin with, HDR is not a term born in the realm of photography, but of computer rendered graphics, where it makes no sense to talk about any bracketing. Secondly, the need for bracketing a scene with a digital camera is totally circumstancial; you may need to bracket a scene in the 10D and not need to bracket in the K5, it will remain being the same scene with the same DR in both cases. Why should the same scene be 'High Dynamic Range', or 'not High Dynamic Range' depending on the camera used? if it were, you would be admitting HDR will not exist anymore one day, when cameras will be able to capture a huge DR.

The point is that the difficult part of HDR is not capturing DR (this is easily done), but mapping it onto a LDR output device. No matter how high is the DR cameras can capture, we'll still have the hard task to map it onto a print or a monitor.

All this is basically a semantic discussion, but IMO it is wron... ooops unreasonable to define HDR according to the input device used. It's a broader concept, in fact much closer to the output devices (render) than to the input devices (capture).

Regards
Logged

digitaldog
Sr. Member
****
Offline Offline

Posts: 8628



WWW
« Reply #232 on: September 12, 2011, 06:27:06 PM »
ReplyReply

Seems oversimplistic to me Andrew; the same as the need for floating point formats. To begin with, HDR is not a term born in the realm of photography, but of computer rendered graphics, where it makes no sense to talk about any bracketing.

How else would a photographer produce HDR? The camera system in a single capture can’t. Otherwise the capture, no matter the range is non HDR (LDR?).

We therefore need multiple exposures. It should then be correct to call this mode HDR.

If we have camera systems that range from say 5 to 12 stops (the range isn’t important), how do we differentiae between HDR and just a system that has a wide capture range? I submit that the original capture system range is what it is. If its inadequate for the user, they bracket and produce a high dynamic range variation (HDR). Otherwise we have to say “this range is HDR, that range isn’t“. I think that is vastly more complicated.

Suppose some day we have a capture system that in a single capture, can provide the range we can see or more. We no longer need to bracket. We can retire the term HDR. If necessarily, we can say “this capture device has X range, that capture device has Y range”.

Quote
The point is that the difficult part of HDR is not capturing DR (this is easily done), but mapping it onto a LDR output device.
While true as well, we have to capture the data first. And we have to come up with some term to describe the capture. I feel that once someone decides they need to bracket and then map, they are implementing HDR. If they have a very wide range in a single capture, they have to probably tone map (something we have to do for most image captures no matter the range).

As for floating point, I still don’t see how that has anything to do (yet) with HDR despite asking. IOW, using the bracket criteria for the use of HDR term is simple, is there any reason to lump the processing (specifically the math) to continue to use or not use that term? As yet, no one has explained why a bracketed image that isn’t using floating point math is not HDR but taking that data into an application that does use floating point math is HDR.
« Last Edit: September 12, 2011, 06:29:07 PM by digitaldog » Logged

Andrew Rodney
Author “Color Management for Photographers”
http://digitaldog.net/
Guillermo Luijk
Sr. Member
****
Offline Offline

Posts: 1274



WWW
« Reply #233 on: September 12, 2011, 06:33:36 PM »
ReplyReply

How else would a photographer produce HDR? The camera system in a single capture can’t. Otherwise the capture, no matter the range is non HDR (LDR?).

This is a contradiction Andrew. You admitted that 'a bracketed sequence from a Canon 10D resulting in a 10-stops DR is HDR, while a single-shot image from a Pentax K5 having the same 10 stops of DR is not HDR'.

A photographer can produce HDR with a single shot, he just needs a HDR camera like the Pentax K5, or he can bracket a lower DR camera. HDR stands for 'High Dynamic Range', and that's it. It is not 'Higher than the camera Dynamic Range'. So everything considered 'high' (I think it was Eric Chan who defined 'high' as something clearly higher than the usual DR we are accustomed to deal with), no matter if we need a bracketing or a single shot to capture it, will enter the definition of HDR.
« Last Edit: September 12, 2011, 06:43:23 PM by Guillermo Luijk » Logged

Guillermo Luijk
Sr. Member
****
Offline Offline

Posts: 1274



WWW
« Reply #234 on: September 12, 2011, 06:37:28 PM »
ReplyReply

Suppose some day we have a capture system that in a single capture, can provide the range we can see or more. We no longer need to bracket. We can retire the term HDR. If necessarily, we can say “this capture device has X range, that capture device has Y range”.

This is exactly what I meant, we won't retire the term HDR because the DR will be as high as it is today (we just didn't need bracketing to capture it), and we will still need to use tonemapping techniques exactly the same as we do today.

HDR exists in computer rendering graphics where there is no bracketing involved, and HDR techniques will exist for long there. Why should it be different with photography HDR?. You would be putting a minor fact (the need for bracketing in some cameras in some scenes) into the definition of a much broader thing, the concept of HDR (High Dynamic Range).
« Last Edit: September 12, 2011, 06:42:16 PM by Guillermo Luijk » Logged

digitaldog
Sr. Member
****
Offline Offline

Posts: 8628



WWW
« Reply #235 on: September 12, 2011, 07:09:55 PM »
ReplyReply

This is a contradiction Andrew. You admitted that 'a bracketed sequence from a Canon 10D resulting in a 10-stops DR is HDR, while a single-shot image from a Pentax K5 having the same 10 stops of DR is not HDR'.

Its not a contraction because I’m using the term based on how the image is captured. The definition is, HDR is the process whereby one brackets.

Quote
A photographer can produce HDR with a single shot, he just needs a HDR camera like the Pentax K5....

My take on the semantics is, the photographer has a camera that can capture the range necessary without bracketing. So its not HDR, its a better capture device (better in terms of capturing a wider range).

Quote
HDR stands for 'High Dynamic Range', and that's it.
Yes it does. I submit however that one man’s HDR is another man’s LDR. So how do we define when to use the term? That’s the issue as I see it. A simple way is to define HDR when one has to bracket because his camera doesn’t produce the necessary range in a single capture. My definition is based on the fact that to produce the high dynamic range not possible with a single capture, a different method of capturing (bracketing) is necessary. This is simply my mindset in why I’d use the term HDR. I am open to changing my mind here. But the issue is, all the cameras are different. Who’s to say the K5 is HDR but the 5DMII isn’t? Someone has to set a value (10+ stops) is HDR? Where does this stop and start? There lies the problem. Who and how does one define a range as being HDR? Very complicated, open to too much debate.

By defining HDR as a process, one that requires bracketing, we don’t have to put arbitrarily set range values on anything. We the photographer decide that the range of a capture device and the range of the scene we wish to capture are such that bracketing is (or isn’t) necessary. Its not defined by a fixed value, YMMV. But if you find you can’t capture the scene with one capture, you must bracket, you are therefore based on this process, implementing HDR. Yes its simplistic because I don’t see the need to define the semantics in a more complicated fashion. That is why I wrote that what Bob wrote sounds reasonable to me.
Logged

Andrew Rodney
Author “Color Management for Photographers”
http://digitaldog.net/
digitaldog
Sr. Member
****
Offline Offline

Posts: 8628



WWW
« Reply #236 on: September 12, 2011, 07:25:45 PM »
ReplyReply

This is exactly what I meant, we won't retire the term HDR because the DR will be as high as it is today (we just didn't need bracketing to capture it), and we will still need to use tonemapping techniques exactly the same as we do today.

Certainly not until a larger group even agrees on when and how to use the term. If we could agree that HDR is a process, then when the process is no longer needed, we’ll stop using it. If we agree on what Tonemapping is, that its not exclusive to HDR (cause its not), then it will survive after the HDR term goes away. If we agree that HDR can be produced with or without floating point math, then we can agree what applications handle HDR.
Logged

Andrew Rodney
Author “Color Management for Photographers”
http://digitaldog.net/
Guillermo Luijk
Sr. Member
****
Offline Offline

Posts: 1274



WWW
« Reply #237 on: September 12, 2011, 07:38:32 PM »
ReplyReply

By defining HDR as a process, one that requires bracketing, we don’t have to put arbitrarily set range values on anything.

Basically what you are saying here is that you are defining HDR as a process where bracketing is needed, just because you cannot find a better definition for it.


If we agree that HDR can be produced with or without floating point math, then we can agree what applications handle HDR.

I can confirm you HDR can be done without floating point formats. I produced this 16-bit integer TIFF file encoding more than 16 effective stops of DR. Lift it by 12EV and you'll still have a lot of shadow detail with zero noise and no posterization. That file is ready for tonemapping using any tonemapping application or technique.

Regards
Logged

digitaldog
Sr. Member
****
Offline Offline

Posts: 8628



WWW
« Reply #238 on: September 12, 2011, 07:59:10 PM »
ReplyReply

Basically what you are saying here is that you are defining HDR as a process where bracketing is needed, just because you cannot find a better definition for it.
More for simplicity and because what group is going to define HDR otherwise? Like I said, X stops of range is HDR but Y isn’t? By using HDR as a term that defines a process, we are not stuck with fixed values as the technology improves.

I can confirm you HDR can be done without floating point formats. [/quote]
That’s what I thought. It makes the HDR definition easier yet.

This isn’t to say the proposal (originally brought up by Bob) will gain any ground. I think however its reasonable.
Logged

Andrew Rodney
Author “Color Management for Photographers”
http://digitaldog.net/
RFPhotography
Guest
« Reply #239 on: September 12, 2011, 08:02:53 PM »
ReplyReply

In the end, Andrew it's unlikely we'll ever get that level of agreement.  Some, like me, define HDR (the process) more narrowly.  And while I can accept that HDR isn't necessary to produce higher dynamic range scenes, I'm not as willing as others to use the term to describe all methods.  I've said it before and I'll say it again; I prefer Caponigro's XDR terminology to describe all processes of producing higher/extended dynamic range which include software like Enfuse/Tufuse, SNS HDR (which isn't technically an HDRI software app), HDR software apps and manual blending.  

I absolutely agree that tonemapping isn't an HDR-only term.  Again, I've said it before and I'll say it once more, tonemapping is just a fancy word for editing.  

GL, I've addressed the issue of the origins of HDR and why I feel that in photographic terms bracketing is necessary.  You won't accept that and that's fine.  Your continued statements that everyone else is wrong and you are right simply make further communication on the matter pointless and unpalatable.  But to close the point, the K5 and D7000 aren't the first what you're calling "HDR cameras" (and why is that the arbitrary cutoff point?  why not the cameras that could capture 10 or 11 stops since in both these cases too the output devices we use can't reproduce all those).  You've got to go back a lot further, even earlier than digital.  You've got to go back to film.  Because colour neg. film and b&w film could both capture more brigthness range than printing media at the time could reproduce.  Which is, in part, why I wrote an article earlier this summer for the Northlight Images site suggesting that the Zone System was analogue HDR and making comparisons between them (I tossed ETTR in there too, just for kicks and giggles).  Different 'process' but with a similar end goal.  And yes, it still was a process.  I'm done.
Logged
Pages: « 1 ... 10 11 [12] 13 14 ... 17 »   Top of Page
Print
Jump to:  

Ad
Ad
Ad