Talk:High dynamic range imaging/Archive 2

From Wikipedia, the free encyclopedia
Jump to: navigation, search
← Archive 1 Archive 2

Dynamic Range / Contrast Ratio of various sources

I'm not sure of the accuracy of the table "Dynamic Ranges of Common Devices." I've seen disagreeing numbers for several items (see e.g. http://www.richardreddy.com/HTML/topics.html), and I can't find any authoritative support for the numbers given. Does anyone have any thoughts about this? I can dig deeper for some more substantiated numbers, but I don't want to bother if anyone has the expertise handy. KLuwak 22:25, 24 October 2009 (UTC) —Preceding unsigned comment added by KLuwak (talkcontribs)

The statements in the table do not match what I was taught in college. Though I would have to go break out my textbooks to find the exact numbers, I think one book indicated typical performance for electronic sources as being 4 to 5 stops total range while film was usually 7 to 9 stops. This was dealing with motion picture film and video, but my understanding is the technology is largely the same in still photography. The sources from the table do not appear to be coming from a scholarly or technical source but an individual doing his own research and posting on his own website.--jqubed (Talk | Contributions) 21:30, 27 April 2010 (UTC)
Agree there's something wrong here. Contrast ratio for LCDs is about right if you believe manufacturer figures. Film and digital are the wrong way around - film has a better range. "Range" for digital is meaningless unless we state RAW versus processing into a format and colour space - and that distinction especially important in this article surely. Would Hedgecoe's books be considered a good citation - if so I'll go and look this stuff up.Infojunkie23 (talk) 10:43, 22 August 2010 (UTC)

Broken Link

The [4] reference in the Dynamic Range of Common Devices table is no longer valid. —Preceding unsigned comment added by 71.197.185.206 (talk) 15:38, 27 February 2010 (UTC)

I don't know if anybody changed it, but as of now it works. —Darxus (talk) 22:30, 18 December 2010 (UTC)

Software links

There is a separate software list article - so why is there a list here? Is there some criterion for inclusion into the shorter list here? —Preceding unsigned comment added by Infojunkie23 (talkcontribs) 10:34, 22 August 2010 (UTC)

Added more information on HDR video and the limitation of tonemapping techniques imposed by color film. —Preceding unsigned comment added by 130.76.96.17 (talk) 15:44, 1 October 2010 (UTC)

Yeah, the software section tends to accumulate spammy cruft. Ick. Also, that "software list article" is not what the description implied, it's an HDR subsection of an article on image stitching / panorama programs.

  • CinePaint - open source HDR image editing software, forked from GIMP in 1998[1]

- good

- rfd, looks like spam (I'm requesting deletion of this document from wikipedia)

- rfd, looks like spam

- Not actually software. Might be useful to link from elsewhere on this page, but deleted from this list.

  • Hugin - HDR merging/panorama stitching (Linux, MacOSX, Unix, Windows; GPL-2+)

- good

- rfd, looks like spam

- rfd, looks like spam, flagged no references a year ago

- good

- rfd, looks like spam

  • HDRpad - free software (Win32/64)

- not HDR just tone-mapping, no references, unlikely to pass notability —Darxus (talk) 23:22, 18 December 2010 (UTC)

Photomatix already went through the deletion request process: Wikipedia:Articles for deletion/Photomatix ProDarxus (talk) 23:53, 18 December 2010 (UTC)
Wikipedia:Articles for deletion/Dynamic Photo HDR
Wikipedia:Articles for deletion/HDR PhotoStudio
Wikipedia:Articles for deletion/EasyHDR Pro
Wikipedia:Articles for deletion/SilverFast
Wikipedia:Articles for deletion/HDRpad
Darxus (talk) 00:13, 19 December 2010 (UTC)
Better link to all 5: Wikipedia:Articles for deletion/Log/2010 December 19#HDRpadDarxus (talk) 01:01, 19 December 2010 (UTC)

Tomlee2010 reverted my removal of this software without explanation. Ckatz restored it. Davidacc reverted it without explanation again. I just restored it again, mentioning this discussion as I did with my initial change. 216.254.152.106 also added a link to a non-existant article. I believe all of this software, even if it doesn't deserve to have its articles deleted from wikipedia, does not belong in this HDRI article because it is only related to tone-mapping, and only uses HDR images as temporary files if at all, only outputting the tone-mapped LDR images. —Darxus (talk) 02:17, 24 December 2010 (UTC)

In august Tomlee2010 added an external link to hdrpad.com from Image fusion which was reverted "rm promotional link": http://en.wikipedia.org/w/index.php?title=Image_fusion&diff=prev&oldid=379362070Darxus (talk) 05:07, 24 December 2010 (UTC)

Darxus has removed from this list several times links to very wellknown HDR or HDR based software, such as Photomatix, easyHDR and serveral others. Darxus, do you want to hide the truth that those apps are good HDR software ? — Preceding unsigned comment added by Davidacc (talkcontribs) 12:26, 25 December 2010 (UTC)
Davidacc, I moved your comment to the bottom of this section where it belongs, and so people will find it. My changes have been in compliance with wikipedia policy, mostly discussing them here. Yours have not. I believe easyHDR and Photomatix, which you just re-added to the article inappropriately without discussion, are primarily tone-mapping software and do not belong in this article. Feel free to put them in tone mapping. If you have evidence to the contrary, provide it before making your changes. I'm removing them again. —Darxus (talk) 20:56, 25 December 2010 (UTC)
Davidacc's entire contribution history has been these two inappropriate reverts of removals of these links. —Darxus (talk) 21:01, 25 December 2010 (UTC)

Darxus, it is ironic that you removed Photomatix from the HDR software list. I was wondering if you have ever used Photomatix. You should make sure you have knowledge about the software that you want to remove. Your attempt of hidding the truth is destroying the value of Wikipedia as a source of information. — Preceding unsigned comment added by Davidacc (talkcontribs) 08:59, 27 December 2010 (UTC)

Darxus, why don't you remove ALL the software links? Or maybe you endorse software for Linux? Would be better if you found more time to get into the topic you're cleaning, or you know, better leave it to someone who is interested in it. To make it clear. Photomatix, easyHDR PRO and a couple other software products DO produce HDR radiance maps. If you think that the fact, that they also do the tone mapping, makes them inappropriate here, then why do you leave hugin? It's primary goal is stitching panoramas, not merging to HDR radiance map - you know the difference? It's huge, you know. The title is "High dynamic range imaging" - software titles you removed from the list were 1000x more appropriate here than any panorama stiching app. And, notability of Photomatix, easyHDR, Dynamic HDR... come on. Did you even bothered to do the research? I think I know the answer... you don't use the software, so you think nobody uses this - it's bull**t. Go play with your own toys. Bartek.okonek (talk) 13:01, 27 December 2010 (UTC)

Bartek.okonek, I see you have deleted the (whole) software section with the description "panorama software links removed - irrelevant to HDR imaging topic". While the description is completely inappropriate to your edit, and I think the deletion of the section was premature, I do think it might be the best choice due to the difficulty of keeping promotional articles out of that section. —Darxus (talk) 18:00, 28 December 2010 (UTC)
Darxus, ok the description should be: "removed irrelevant links". I agree, better to remove the whole "software" section so there are no promotional links at all. It was unfair to allow some software to be promoted here while not putting links to other software products that are on the "HDR market". Actually I think that the note about Photoshop CS2 makes no sense either. Photoshop was not the first and, at least CS2, was rather poor at HDR generation and tone mapping.--Bartek.okonek (talk) 14:07, 30 December 2010 (UTC)

Added more examples of camera's with build-in HDR functionality

So anyone can make HDR pictures with camera's that allow for changing EV settings. Auto exposure bracketing makes it a lot easier. But nowadays more and more camera's seem to have build-in also the combining of the pictures to have an immediate result. The Canon G12 and S95 were already given as an example. But there are more. I added them. It might be better for this article to just mention the fact that this is a feature that can be seen in an increasing amount of camera's. And then to start some list elsewhere. (Because it really adds to how usefull this article is) Yvolution (talk) 17:49, 13 October 2010 (UTC)

Do these cameras actually output an image with a high dynamic range, with increased depth per color? Or are they just outputting the low dynamic range result of running a high dynamic range image through local tone mapping? —Darxus (talk) 22:33, 18 December 2010 (UTC)

Methods

The link to "High Dynamic Range rendering" article is definitely wrong here. It's not a method of HDR imaging... it's something opposite I'd say. Actually speaking of "methods" themselves makes no sense even. There is only one scientific way to merge to HDR... those "methods" could be/are variants of it. Those variants can have special features like ghost removal, noise reduction etc. Wide Dynamic Range... OMG... what's this? It's like "almost HDR" or what? Irrelevant, notability is very questionable. Bartek.okonek (talk) 19:49, 27 December 2010 (UTC)

The only way I can make sense of this statement is if you are using a definition of "imaging" that basically equates to "photography". Which is not appropriate for this article. Particularly where you say "There is only one scientific way to merge to HDR...". "Merging" is not even inherent in HDR. I would define "imaging" as "creation or manipulation of images". It's possible the title of this article could use something more appropriate than "imaging". But I doubt it. —Darxus (talk) 18:35, 28 December 2010 (UTC)

If this article is about both photography and rendering, then what to do with this one: High dynamic range rendering? And now there's a question... what is "HDR rendering" at all? I can tell you - it's another idea of guys who're responsible for marketing... they have invented also LCD screens with dynamic range of 70000:1 (I am sitting in front of such screen right now).

Consider this... what's the problem to render an artificial scene with dynamic range limited only by data storage format... float, or double? I can even "draw/render" an artificial 3x3 pixel image of a distant star, here it goes:

[0, 0, 0; 0, 1e12, 0; 0, 0, 0]

What's the dynamic range of this "image"? There is no noise (apart from quantization noise) and the data type is array of 64-bit double.

In case of photography it seems to me that HDR is reserved for merging several differently exposed photos together to form a radiance map of the photographed scene, that has more dynamic range than a single photo. But... If I had a CMOS/CCD sensor with linear sensitivity and SNR at 90dB could I call the photos taken with it a HDR? Or maybe I can call an image a HDR if it has dynamic range of 70dB, but it is a merge of several 55dB shots?--Bartek.okonek (talk) 13:46, 30 December 2010 (UTC)

When a wikipedia article gets too big, one way of managing it is to break one of its larger sections off into a separate article. That is the relationship between HDRI and High dynamic range rendering. They are separate only because they are more manageable that way than as a single article.
What LCD do you have with 70000:1 contrast? The only display I heard of (which certainly doesn't mean there aren't others) with a contrast ratio significantly above 700:1 was the BrightSide DR37-P, at 200,000:1 (17.6 stops). Dolby purchased BrightSide in 2007, and as far as I can tell the technology was never put into production. It should probably be mentioned in this article.
HDR rendering predates LCD monitors probably by over a decade, and started out free. "The use of high dynamic range imaging (HDRI) in computer graphics was introduced by Greg Ward in 1985 with his open source Radiance rendering and lighting simulation...." - the HDR rendering article. HDR rendering is using computers to synthetically create images with a high dynamic range.
While extending image rendering to higher bit depth is, as you point out, perhaps not terribly complicated conceptually, I believe it remains uncommon. The dynamic range of your "image" is undefined, because you have not defined a color profile for it. It could be 10:1 or 2^100:1. Most images use sRGB, which would make it... something LDR. That article could benefit from defining the maximum contrast somewhere. You could certainly create a 2 bit per pixel image and declare that its contrast ratio is 10,000,000:1, but I would not call that HDR because it lacks any kind of useful precision between brightness levels. I would say the minimum bits per color for HDR would be 16 (so 48 bits per RGB pixel), based largely on my belief that this is the minimum color depth of any image file format anybody actually uses for HDR.
HDR has nothing to do with the number of photographs you merge together, except that that is the only common photographic method currently available. If you had a camera with a sensor that produced a dynamic range equivalent to merging several LDR images, yes, that would be an HDR photo. The definition of HDR, as it relates to images, is only related to the dynamic range of an image - the contrast ratio or number of stops from lightest to darkest (and perhaps retaining useful precision between brightness levels). —Darxus (talk) 19:34, 30 December 2010 (UTC)
A google search for 70000:1 LCD produces plenty of hits on monitors. Looks like mostly Samsung, some LG.... —Darxus (talk) 19:37, 30 December 2010 (UTC)
That 17.6 stops of the 200,000:1 BrightSide display is still only the dynamic range of about two merged photographs. Kind of shows how misleading using a linear scale for contrast is for something that behaves more logarithmically, and why stops are a log scale. —Darxus (talk) 19:47, 30 December 2010 (UTC)
How can you call a 2-bit image a HDR by arbitrarily declaring contrast ratio? If so, then any tone mapped HDR radiance map could be still called a HDR image, because you "know" what was the real dynamic range of the photographed scene. Dynamic range of the >photo< is limited by noise, including the quantization noise.
Color profile? Can't I have a b&w image? My "space" is [min,max]. --Bartek.okonek (talk) 15:22, 31 December 2010 (UTC)

Demonstrating bit depth variance

The book High Dynamic Range Imaging, the first reference on this page, compares the difference between common 8 bit per color channel images (like all jpegs) to HDR images (often 16 bits per channel) within the limited dynamic range of a printed image by comparing a 4 bit per channel image to an 8 bit per channel image, on page 5. I would like to reproduce that in this article. I attempted something similar in the past with phenomenal controversy. My current plan is to use a color palette with the full range of red, green, and blue values, but only 16 (4 bits) colors per channel (evenly spaced) instead of 256 (8 bits). I believe this would faithfully reproduce the example in the reference. I think the first image in crepuscular rays would make a good example. Any opinions? —Darxus (talk) 23:29, 30 December 2010 (UTC)

  1. ^ "CinePaint Frequently Asked Questions". Retrieved 2009-08-31.  [dead link]