Luminous Landscape Forum
Equipment & Techniques => Digital Cameras & Shooting Techniques => Topic started by: Marc Verschueren on August 16, 2011, 07:22:50 am
-
I have a question about HDR using only one shot. Normally you would use bracketing on the camera to shoot say tree images which are then blended together afterwards on the computer. But could you do it with one shot if you f.i. use lightroom and shoot raw?
Could you make tree copies within lightroom of the picture; then adapt the f-stop manually for the first copy to say -1 and adapt the thirth copy to +1 and then use those tree images to blend together?
Would it be the same as tree bracketted images coming out of the camera?
-
I've heard people talk about single exposure HDR, and most of it is nonsense. Use of ND grad filters extends the dynamic range within a single exposure, but simply taking a straightforward exposure & extracting whatever detail one can, doesn't amount to any extension of the dynamic range, so isn't HDR. Blending over & under-exposed copies from the same exposure isn't HDR.
-
I think Bill is right. The whole basis of multiple exposure HDR is that the over- and underexposed images will contain image information that the "correctly" exposed image does not. For example, the underexposed image might contain highlight details that are blown out in the others, and the overexposed image will contain shadow detail that has gone to black in the other images. Your single image technique does not add any additional image information, and whatever information there is in that one image should be accessible using other techniques, such as fiddling with the curves.
-
Would it be the same as tree bracketted images coming out of the camera?
The previous posters are correct, the purpose of HDR shooting is to increase the quality of the information in an image file. By manipulating a single file, no real data is added to manipulate, only manipulated data ...
Cheers,
Bart
-
I have a question about HDR using only one shot. Normally you would use bracketing on the camera to shoot say tree images which are then blended together afterwards on the computer. But could you do it with one shot if you f.i. use lightroom and shoot raw?
You can, the only difference between bracketing or not bracketing is the amount of visible noise in the shadows (assuming you preserve the highlights in the single shot and in at least one of the shots in the bracketing). The point is that with a regular camera and a high contrast scene, noise in a single shot will usually be so high that shadow textures will become invisible.
But conceptually there is no need to bracket for HDR tone mapping. If noise is not a problem in your single shot, make as many copies of it at different exposures as you will, and feed them into your favourite monstruous mapping application.
Regards
-
Ok! Many thanks to you all for this valuable information. I understand it now: it is the lack of information in the original shot that's killing my idea of being able to do HDR with copies of this on shot.
Now that you've all explained it is clear to me!
Thanks again.
-
You can, the only difference between bracketing or not bracketing is the amount of visible noise in the shadows (assuming you preserve the highlights in the single shot and in at least one of the shots in the bracketing). The point is that with a regular camera and a high contrast scene, noise in a single shot will usually be so high that shadow textures will become invisible.
But conceptually there is no need to bracket for HDR tone mapping. If noise is not a problem in your single shot, make as many copies of it at different exposures as you will, and feed them into your favourite monstruous mapping application.
Regards
Conceptually there's no need to bracket to tonemap an image; that is true. But you're not getting HDR by doing so. Multi-processing a single file (irrespective of noise) isn't adding any brightness range. That's really the point of HDR (or other extended exposure methods) so technically to get HDR (or extended exposure) you do need to bracket.
If you're going to tonemap a single image file, just toss the single image into the tonemap cooker. Creating +/- brackets from the single file won't make a difference.
-
That's really the point of HDR (or other extended exposure methods) so technically to get HDR (or extended exposure) you do need to bracket.
That is your photographic-oriented opinion Bob. But HDR algorithms were born in the computer rendered graphics world, as a tool to display big contrast gaps in low dynamic range output devices through global contrast reduction plus local contrast enhancement methods. Obviously it is nonsense to talk about any bracket in computer rendered graphics since there is only one 'picture' to tone map and render. HDR formats used in digital photography are single image formats as well, they simply put up together the best parts (shadows, highlights and mid tones) of the bracketing series.
So bracketing is just a consequence of the limited dynamic range of digital cameras because of sensor noise. As long as sensor designs are less noisy, bracketing will become unnecesary. But HDR tone mapping techniques will have exactly the same validity they have today, we'll simply need a single capture (with all the advantages this means: quicker shooting, no alignment nor ghosting issues, storage and CPU savings,...).
Regards
-
I understand what the origin of HDR is/was. But this is a photography forum. The vast majority of the people here are photographers. The vast majority using HDR will be doing so for photography related reasons, not CG. I'd venture that very few are doing 3D CG rendering. So in that context, I think my comments are valid.
Lower noise should be a result of a properly bracketed HDR merge, sure. But if the scene DR fits within the sensor DR then there's no technical reason to use HDR tonemapping (again, speaking from a photography standpoint). As sensors are manufactured with even broader DR HDR, as a technical method of extending brightness range, will be obviated. That doesn't take away from an artistic reason to use HDR tonemapping tools. But to say the validity of using HDR tonemapping methods where not necessary if the sensor can capture the entire scene DR will be no different I don't think is accurate.
Your definition of 'single image' as it relates to bracketed photographic HDR is somewhat specious.
-
Hi,
I'm doing some studies on a Stouffer wedge with density range of 4.1 right now. My impression is that Lightroom cannot extract all dynamic info from a raw image without either:
- Using extreme amount of fill light
- Using the built in graduated filter
In my experience, doing two separate developments and combine in Photoshop using some decent masking technique is actually a very good idea. A description of one such a method is described here:
http://echophoto.dnsalias.net/ekr/index.php/photoarticles/46-fixing-sky-with-luminosity-mask
Best regards
Erik
Best regards
Erik
I've heard people talk about single exposure HDR, and most of it is nonsense. Use of ND grad filters extends the dynamic range within a single exposure, but simply taking a straightforward exposure & extracting whatever detail one can, doesn't amount to any extension of the dynamic range, so isn't HDR. Blending over & under-exposed copies from the same exposure isn't HDR.
-
I have used the OT technique with a few ETTR images. In some cases, the sky is very bright, but not blown. I've been unable to darken the sky and separate the tones in the sky, without unduly darkening the rest of the image, WITHIN Lightroom alone. In these cases, I develop the image twice, once with normal tones in the non-sky portion and letting the sky go where it is, and the other with the exposure lowered and contrast raised, as well as other Lightroom adjustments.
I prefer to use LR over PS for this type of work.
Alan
Edit: I would not call this HDR, however.
-
If you are having trouble with lighroom resolving the dynamic range of the image, is the sensor you are using capable of resolving the entire dynamic range of the stouffer wedge? There were some examples posted a day or two ago and from my perception either the highlights or the shadows were blocked up depending on the level of exposure. I could get more dynamic range in the image by reducing the contrast setting in lightroom. I often find that the contrast setting is too high for my liking and by reducing contrast the available dynamic range is more workable.
-
Edit: I would not call this HDR, however.
Hi Alan,
Correct, neither would I. It's tonemapping, and extracting 2 (or more) images from different parts of the same tonescale doesn't make it an HDR. But it does allow to make a nicer blend/tonemapping due to the lacking tools to do it otherwise or as efficiently.
Cheers,
Bart
-
Hi!
I would not argue with your statement. On the other hand HDR is used for two related processing:
1) Achieving extended dynamic range by combining several images
2) Mapping the extended range into something that can be shown on screen or print
Real HDR is by necessity a combination of both techniques. The discussion in the original topic is more like tone mapping.
Edit: I would not call this HDR, however.
-
If you're shooting a camera with good dynamic range like a Phase One or Leaf/Mamiya Digital back then you easily get the benefits of HDR without multiple images. Much of the time HDR only requires a few stops outside what is normally used, and digital backs have enough reserve dynamic range to do this in a single shot.
Capture One and HDR (http://www.captureintegration.com/2010/12/28/hdr-via-lcc/)
Doug Peterson (e-mail Me) (doug@captureintegration.com)
__________________
Head of Technical Services, Capture Integration
Phase One Partner of the Year
Leaf, Leica, Cambo, Arca Swiss, Canon, Apple, Profoto, Broncolor, Eizo & More
National: 877.217.9870 | Cell: 740.707.2183
Newsletter (http://"http://www.captureintegration.com/our-company/newsletters/") | RSS Feed (http://"http://www.captureintegration.com/2008/08/11/rss-feeds/")
Buy Capture One 6 at 10% off (http://"http://www.captureintegration.com/phase-one/buy-capture-one/")
Masters Series Workshop:
New England Landscape - Fall Color (Oct 5-8) (http://"http://www.captureintegration.com/2011/06/30/nelandscape/")
-
1) Achieving extended dynamic range by combining several images
2) Mapping the extended range into something that can be shown on screen or print
Real HDR is by necessity a combination of both techniques. The discussion in the original topic is more like tone mapping.
Actually I'd say not so, Erik. HDR is, by definition, High Dynamic Range. The 32 bit image before tonemapping is an HDR image. Once tonemapped and dropped into an 8 or 16 bit space it no longer is High Dynamic Range but rather LDR - Low Dynamic Range. Real HDR doesn't require the second step. Real HDR (and to keep GL happy I'll say real HDR in standard photography terms) is only the first step. The second step is required because the devices we view images on and the devices we print on can't replicate the HDR image. At least not yet.
-
If you can use a single image to produce a "HDR picuture" depends on the dynamic range of the scene. A printed image has a dynamic range of around 200:1. A jpeg image has a dynamic range of around 400:1. HDR software is not required to map the jpeg into the printed image. If you use HDR software in this case you are just adding local contrast to porduce an "artistic effect". The same effect can be achieved with the unsharp mask with the radius set to a very high number.
A DSLR can capture a dynamic range af around 8,000 to 1 in RAW. If you have a scene with a dynamic range close to but less than 8,000 to 1 you will be able to produce two images from the RAW image, one of which will show the highlights but have the shadows clipped and the other will have the shadows but will have the highlights blown. It will be impossibel to use the recovery and fill sliders to get a composit. If this is the case then the HDR software can be used to tone map the image.
If the dynamic range of the scene is greater than the dynamic range of the camera sensor then multiple images are required to capture the full scene.
Using multiple images may improve the quality of the tone mapped image when the scene falls within the dynamic range of the sensor.
-
There is HDR, and there is tone mapping, and they are related but separate things.
Tone mapping is what is needed to map the tone and color range from one device or space to another. Perhaps the more photographic term is "development" or "process." If your camera can capture 10 useful stops of dynamic range, but your printer/paper can only print 6, then you will need to tone map your captured image to reproduce nicely on the print. This concept is true regardless of how one captured the image (single capture or multi). Tone mapping can be done in any number of ways. The simplest is using a global curve, and more complex methods include local adjustments (e.g., dodging & burning).
HDR is a somewhat arbitrary term to describe the idea of "more DR than usual" but ultimately (as Guillermo points out) all it really means is less noise. There are multiple ways to get more DR (i.e., less noise). For example, you can either use a camera whose capture capabilities inherently offer less noise (e.g., Canon 1Ds III compared to Canon S90), or you can use multiple captures, or both.
-
I understand what the origin of HDR is/was. But this is a photography forum. The vast majority of the people here are photographers. The vast majority using HDR will be doing so for photography related reasons, not CG. I'd venture that very few are doing 3D CG rendering. So in that context, I think my comments are valid.
As others pointed, we can talk about HDR imaging in digital photography without the need for bracketing (a sensor with enough DR to capture the entire scene in a single shot, use of ND filters to reduce the effective DR hitting the sensor,...), so it's nonsense to say bracketing is a part of the definition of HDR. Anyway I think this is a rather semantic discusion since we all know what we mean.
However I would like to make a consideration that is seldom made: since capturing DR in a sensor is all about noise in the shadows, and visible noise in the shadows is reduced (because SNR increases statistically) when an image is downsized, the ability to capture a given DR by a camera strongly depends on the ouput size of our application.
Quick numbers example: Canon 5D2, let's assume it has an effective DR of 9 stops at full resolution (21Mpx). If in our application we just want to upload a 800x533 image to our website, the effective DR the camera is capable of capturing for that size is larger:
- Original resolution: 5616 x 3744 = 21Mpx
- SNR slope in the shadows for 5D2 (SNR curves): 6dB/EV (typical in read noise limited cameras), this means DR will increase by 1 stop for each 6dB in SNR improvement
- SNR improvement for rescaling 5616 x 3744 (21Mpx) to 800 x 533 (0,43Mpx): 20*log10[(21/0,43)^0,5] = 16,9 dB = 16,9/6 stops = 2,8 stops
- Effective DR of 5D2 at web sized images: 9 + 2,8 = 11,8 stops
If our real world scene has 11 stops of DR, the 5D2 will need bracketing to produce a sufficiently clean of noise image for a large print at full res. But if we just plan to upload the resulting HDR image to a website, one shot will be enough. Both images are HDR, they simply have different requirements.
The formula to calculate SNR improvement when downsizing is simple statistics. See Emil's article about noise to know more.
-
Marc, you can tone map a single image with Photomatix. It's not HDR, but it is different.
-
Sorry, GL but I don't agree. One of the technicalities that you're either not considering or purposely leaving out is that HDR images are 32 bit. A camera that can capture 100 stops of brightness but does so in 12, 14 or 16 bit isn't capturing HDR.
For methods that don't take the resulting image into a 32 bit space JP Caponigro's term XDR, eXtended Dynamic Range, is more appropriate.
I'd go further to say that if some form of blending method isn't needed, or if some form of GND filtering isn't needed and the sensor can capture the entire brightness range of the scene being photographed that we're out of the realm of either HDR or XDR and into a 'normal' DR. If the sensor can capture it then it's the 'normal' DR of the sensor. Nothing High or eXtended about it.
I'm not going to, and frankly can't, get into a 'numbers' debate. I'm a photographer not a physicist and while LL is the web home of the measurebators I don't happen to be one of them. From a purely visual standpoint I don't buy into your supposition but I'm not going to argue with the math because I can't. But as it is, I don't know where the '20' comes from in your equation nor do I know where the '6 stops' comes from. I also don't know why the log equation is being raised to the .5 exponent.
Eric's point about tonemapping is a good one. And it's one that I've made in the past and has been discussed if not in this thread then in another one on here recently. Tonemapping is really just a fancy word for editing.
-
Sorry, GL but I don't agree. One of the technicalities that you're either not considering or purposely leaving out is that HDR images are 32 bit. A camera that can capture 100 stops of brightness but does so in 12, 14 or 16 bit isn't capturing HDR.
32 bit HDR formats are not needed to do HDR. I think again you are putting a tecnique (multiexposure first, now using floating point formats for tone mapping) as a part of the definition of a concept which is much wider, and cannot be restricted to simplistic rules so as which tools you used. As Eric said: 'HDR is a somewhat arbitrary term to describe the idea of "more DR than usual"'.
Who says you have to multiexpose to achieve that? who says you have to use 32-bit floating point formats to achieve that? why not 64-bit integer?. I have demonstrated that multiexposure is not necessary to capture what is considered a HDR scene (example of the 5D2 and the small format output), and I have demonstrated 32-bit floating point formats are not needed to hold HDR data (just download superhdr.tif (http://www.guillermoluijk.com/download/superhdr.tif), an integer 16-bit TIFF containing a scene with more than 16 stops of DR ready to be tone mapped). I do my HDR's starting from a unique 16-bit TIFF, and tone map them using regular PS tools (basically masked curve layers).
(http://www.guillermoluijk.com/article/superhdr/progev.jpg)
Regarding the equations, they are basic equations but explaining them is out the scope of this thread. You will have to believe them in this case.
Regards
-
I understand it now: it is the lack of information in the original shot that's killing my idea of being able to do HDR with copies of this on shot.
Not necessarily.
To make best use of the given information in one shot, may not always be possible with the on-board tools of a given Raw converter,
and may require HDR-typical tone mapping techniques.
http://imagingpro.wordpress.com/2008/12/03/expanding-the-dynamic-range-of-a-single-raw-file/
Peter
--
-
Well, in the end I think we're talking about the same thing GL, just using different terms to do it. Yes, I'm restricting HDR to those formats that are considered High Dynamic Range and those are 32 bit file types such as .exr, .hdr, 32 bit .tiff and 32 bit .psd. Anything that doesn't go into 32 bit and using techniques like layers and masking I put into the category of, as I noted earlier, XDR. HDR isn't an arbitrary term. It has become so, like Ski-doo or Kleenex or Xerox and like those it's become the manner of describing all extended dynamic range methods but also like those others, it's not technically correct.
As far as whether your equation is 'basic' I guess that depends on the background and point of view. I can bootstrap a yield curve on the back of a cocktail napkin but I don't think many people would consider that as 'basic' math either. ;D
-
It seems to me that the term HDR has come to mean the melding of 2 or more differently exposed images olf the same scene. An individual image may contain a wide dynamic range, and we can deal with that using curves and other standard techniques. It's when you have 2 or more differently exposed images of the same scene that you need specialized software and techniques to create the final image. This is what we are, or should be, talking about.
-
This discussion is much like discussing the term "broadband". Does it mean >640kbps? Does it mean fixed pricing? Or does it just mean "whatever speed that 30% of the users cannot afford at any time"? I am sure there are government agencies spending lots of man-hours trying to define the semantics, but in the end, what matters to me is what speed/characteristics I have, not what it is called.
I believe that HDR, in this context, as others have said, only means "more dynamic range than usual". There are/will be camera sensors that claim to do single-shot "HDR" (due to high native DR), no matter if it is saved to 14,16 or 24 bit integers. Exposure stacking is no direct consequence of "HDR" but an indirect one: if you want to capture better DR than usual, you may have to use unusual methods (ie stacking).
In the end, what matters most to me is that the recording device is capable of faithfully recording scenes whose dynamic range cannot directly be appreciated on todays LDR display and paper. This means that the high recorded DR can be used to:
1. Do image processing on a "linear" recording of the scene (even if the final output is clipped to LDR)
2. Postpone choices about exposure to post-processing. The LDR window can be moved up and down according to taste.
3. Fancy tone-mapping algorithms can be used to try to squeeze the impression of HDR into the limited LDR display/paper
4. The scene is captured for posterity (and possible improved display/paper tech)
I believe that fancy tonemapping algos is trickling into regular single-shot raw developers, therefore it may make sense to use the word "tonemapping" for that kind of operations.
-h
-
I believe that fancy tonemapping algos is trickling into regular single-shot raw developers, therefore it may make sense to use the word "tonemapping" for that kind of operations.
Regarding this, I think the more DR cameras can capture in a single shot, users will be wanting to enjoy it right from their RAW developer because they will now the information is there (I know many users reporting to be happy for doing 99% of their photographic work without leaving LR), and therefore tone mapping capabilities (in the sense you mean) will necessarily improve in RAW converters with more advanced algorithms. My ACR CS2 doesn't have highlight and shadow recovery sliders for instance.
This can lead to RAW converters becoming direct competitors of today's specialized tone mapping software such as Terrormatix.
Regards
-
This discussion is much like discussing the term "broadband". Does it mean >640kbps? Does it mean fixed pricing? Or does it just mean "whatever speed that 30% of the users cannot afford at any time"? I am sure there are government agencies spending lots of man-hours trying to define the semantics, but in the end, what matters to me is what speed/characteristics I have, not what it is called.
I don't think your analogy is a good one. HDR is clearly and appropriately defined as the technique of obtaining a greater DR by combining 2 or more differently exposed images of the same scene. This requires specialized techniques and specialized software, and is usefully distinguished from applying traditional digital manipulation techniques to get the desired DR from a single image.
-
I don't think your analogy is a good one. HDR is clearly and appropriately defined as the technique of obtaining a greater DR by combining 2 or more differently exposed images of the same scene. This requires specialized techniques and specialized software, and is usefully distinguished from applying traditional digital manipulation techniques to get the desired DR from a single image.
We seem to be disagreeing on semantics, lets put that to rest.
-h
-
Regarding this, I think the more DR cameras can capture in a single shot, users will be wanting to enjoy it right from their RAW developer because they will now the information is there (I know many users reporting to be happy for doing 99% of their photographic work without leaving LR), and therefore tone mapping capabilities (in the sense you mean) will necessarily improve in RAW converters with more advanced algorithms. My ACR CS2 doesn't have highlight and shadow recovery sliders for instance.
This can lead to RAW converters becoming direct competitors of today's specialized tone mapping software such as Terrormatix.
Regards
I agree. raw files and "HDR" files are both essentially linear representations of a scene with more DR than paper/displays. The difference is only in degree, not in nature. It only makes sense that the art/science of semi-automatically mapping them to LDR is unified.
I too enjoy the luxury of working mostly within LR. My only gripe is locking myself into a proprietary, closed database. I would much prefer that the edits and organization was stored into an open format (Adobe would still be free to implement the gui and signal processing closed).
-h
-
Making different exposures of a single RAW file and combine them, that is tone-map the image in a non-linear way, is indeed possible and sometimes quite useful.
I sometimes do it, especially in landscape photography, but the result is more of dodge-and-burn and rarely a cartoon-like HDR look.
Some RAW developer software now has built in functions for dodge and burn and local adjustments so you don't need to combine several "exposures" in external software, but my favourite RAW software (rawtherapee) does not have it so I have to use this technique.
Concerning dynamic range, the RAW file of a modern camera indeed has more than can be represented in a print, so doing some HDR-like tone-mapping may be relevant. If you want t call it HDR or not when using only one shot I don't really care, but most will associate HDR to bracketed shots.
-
This is a HDR right out of the Sony A55 (downrezed in Paint). The original is quite spectacular for its realism. It does not look like one of those industrial grunge HDRs. I no longer feel I have to try to learn HDR software. If the shot requires HDR, my camera can handle it.
-
Check out John Neel @ www.pixiq.com. John (one of the contributors) to PIXIQ has a 3-part blog post
on SS (single-shot) HDR: http://www.pixiq.com/article/ss-hdr-my-way-part-1
John is very innovative and is one of the more prolific contributors to the Pixiq site.
cheers!
-
He probably could have got it down to a single post if he'd trashed all the talk about how he doesn't follow rules and does his own thing and doesn't do what the software engineers tell him to do and how he doesn't follow rules and that he does his own thing and that rules are made to be broken and well, who needs rules. ::)