Wikipedia:Featured picture candidates/Sunset hdr

Source: Wikipedia, the free encyclopedia.

Sunset HDR

Comparison of normal exposure and a tone mapped HDR image
HDR
sunset
Normal Exposure

A beautiful image which illustrates HDR page well.

  • Support Self Nom. --Fir0002 22:43, 4 July 2006 (UTC)[reply]
  • Comment - I think this picture would illustrate HDR better if it had the picture with standard exposure aswell as this picture side by side (like your flood pic). Froggydarb 23:26, 4 July 2006 (UTC)[reply]
  • Oppose - Weak oppose - Neutral. I find most HDR images unnatural looking when converted for display, as this one is. (Technically speaking, that means this is not an HDR image, but a 'normal' image derived from HDR imaging techniques. A true HDR image cannot be displayed as such on a computer monitor.) So while I recognize the time and effort that went into making this image, I don't think it does a good job of demonstrating HDR imaging in the article. This is not meant as a knock against your work, Fir0002; it's a limitation of what can be captured in HDR images vs. what can be displayed. No image displayed on a computer display can accurately represent what HDR is supposed to do. -- moondigger 00:57, 5 July 2006 (UTC)[reply]
    • So your opposing because your monitor can't handle the image, that doesn't make any sense. Pegasus1138Talk | Contribs | Email ---- 04:25, 5 July 2006 (UTC)[reply]
      • No, I'm opposing because NO monitor can display a true HDR image, and this is not a true HDR image. It is normal image that is composed of three exposures tone-mapped into a dynamic range that can be handled by computer monitors. Such images do not demonstrate HDR; they demonstrate tone-mapping or dynamic range compression. This isn't Fir0002's fault... it's a nice enough sunset. But illustrating HDR in an article viewed on a computer screen is nearly impossible. The more range-compressed images people see in an article that's supposed to be explaining HDR, the more likely people are to make the wrong association.-- moondigger 12:39, 5 July 2006 (UTC)[reply]
        • Hang on, I'm confused now. If a HDR image can never be displayed correctly, what's the point in making them in the first place? I've read the article but I still don't fully understand the point your making moondigger; it seems strange to me that a whole imaging technique would exist if the images can never ever be displayed correctly, or am I misunderstanding you? Is there any way that they can be displayed properly? Raven4x4x 13:21, 5 July 2006 (UTC)[reply]
          • I think we're getting outside the scope of the discussion of this image. A good explanation would take some time to compose, and when I do it I plan on integrating it into the article text. The very short version is that HDR means high dynamic range. i.e., more dynamic range than our capture media (film, CCD, or CMOS sensor) can handle in a single frame, and often more dynamic range than our output devices (computer monitor, printer) can display in a single frame. This is a limitation of the technology we're working with. (Also, due to the nature of digital capture technology, information in the highlight areas is more accurately described than information in the dark shadow areas.) So when we're presented with a scene that exceeds the number of stops of luminance that we can capture in a single frame, we have to make multiple exposures, each one adjusted to capture the maximum detail. Photoshop then allows us to combine each of these exposures into a single file that accurately describes the full range of the scene. The problem is that no monitor can display it all, accurately, at one time. What HDR techniques have done for us is that we've accurately captured and stored all the information in a scene. That doesn't mean we can display it or produce output that matches what we've captured. Now if we want to produce output, we must change all the tonal values from their original, accurate values to new values that fit within the capabilities of our output device by doing tone-mapping or dynamic range compression. When you do that, the result is no longer HDR. Sometimes this process will result in something that seems natural to the eye; more often, it will result in something that doesn't resemble reality as we normally perceive it. This "HDR sunset" falls into the latter category for me; the original base exposure looks like something I might actually see in the real world, while the tone-mapped version, however interesting it might be, appears unnatural. Ditto with the HDR church interior we saw here a few weeks ago (also in the HDR article).
Look, I'll withdraw my opposition to this image. When I have the time, I'll write up a longer article, complete with sample images, and integrate it into the existing article. -- moondigger 14:16, 5 July 2006 (UTC)[reply]
A nit, your statement "information in the highlight areas is more accurately described than information in the dark shadow areas" is factually inaccurate. While brighter content suffers less competition from noise it suffers much more quantization noise because of the perceptual compression (gamma 2.2 curve) applied by LDR capture and inverted by our monitors. Also, most sensors go somewhat non-linear as their photon wells fill up. See Steve Mann's work on comparametrics [1], his work demonstrates that maximum information content is actually found in the mid-tones. "Expose to the right" makes sense for a lot of reasons, but the claims that it maximizes the recorded information with a normal photography process do not stand up to scientific analysis. --Gmaxwell 01:55, 7 July 2006 (UTC)[reply]
I hadn't read Steve Mann's work, so thanks for the link. However, my statement is not factually inaccurate. Information in the highlight areas is more accurately described than information in the dark shadow areas - at least during capture, which is what I was talking about. The quantization noise doesn't come into it until we actually do something with the captured data later on when we manipulate it for output. But the whole point of HDR is to take multiple exposures and combine them prior to doing any perceptual compression; Photoshop HDR works with RAW files that have no perceptual compression applied. At this stage all we're doing is capturing data in multiple stepped exposures so Photoshop's HDR routines can pull the maximum information from each file for a given set of tones in the image. For that purpose, exposing to the right makes even more sense. (BTW, if I'm missing some obvious point here, please correct me.)-- moondigger 05:48, 7 July 2006 (UTC)[reply]
  • Support – beautiful! Morganfitzp 03:08, 5 July 2006 (UTC)[reply]
  • Support very nice image. Pegasus1138Talk | Contribs | Email ---- 04:25, 5 July 2006 (UTC)[reply]
  • Oppose. What's HDR about it? Looks like its just two seperate exposures for foreground and sky. Rather boring and not particularly illustrative of either sunset or HDR. --Dschwen 06:51, 5 July 2006 (UTC)[reply]
HDR is a method whereby you take seperate exposures and combine them to produce an image without blown higlights or underexposed areas. --Fir0002 12:20, 5 July 2006 (UTC)[reply]
Uhm, no, you are describing exposure blending. --Dschwen 06:25, 7 July 2006 (UTC)[reply]
  • Oppose Support comparison- I have to agree with Dschwen, it is a beatiful image although I don't think it demonstrates HDR very well. I look at this image and see a sunset, if there was another image with a standard exposure next to it I would change to full Support (or even better, the three images you used to create this image and the standard exposure image). Froggydarb 11:50, 5 July 2006 (UTC)[reply]
Done. --Fir0002 12:20, 5 July 2006 (UTC)[reply]
I don't think you understand me (or I didn't explain myself properly). I meant: have the two pictures together, like the before and after flood picture you put up on FPC a week or so ago. This way it doesn't just look like a sunset, because you have something to compare the HDR picture with. If someone that didn't know what HDR was looked at the image they might think it's just a sunset, even if it had a caption noting that is a HDR image. Froggydarb 21:31, 5 July 2006 (UTC)[reply]
OK how about that? --Fir0002 11:15, 6 July 2006 (UTC)[reply]
Yep, that's good. Froggydarb 23:41, 6 July 2006 (UTC)[reply]
  • Oppose - There is much more to featured pictures than just being a pretty photograph or diagram. You cannot apply criteria that one would use to judge whether or not one would hang a picture up in one's own house to judge candidacy for featured pictures. In terms of composition, the subject matter is uninteresting, there is nothing being done by taking a photograph of random clouds in a sky that is approaching dusk and labeling it as a 'sunset.' If someone looks at the image the immediately cannot recognize what the purpose is (which is not always a bad thing, but in this case it makes the feel lacking in composition). This image just do not lend itself to improving the HDR article, especially when the other pictures in the article are much more dynamic and have more dimensions, depth, and illustration of what HDR photography is. Sudachi 00:26, 6 July 2006 (UTC)[reply]
  • Support Groovy photo man! Keep on truckin’! TomStar81 02:27, 6 July 2006 (UTC)[reply]
  • Oppose Agree with moondigger. We need something else to illustrate HDR. All the examples (except perhaps the Grand Canyon one) result in a low contrast, murky image. If there are bright highligts in a scene, I want them bright on my screen, and I want the shadows dark, but with detail! So, I'll probably never get what I really want... --Janke | Talk 06:54, 6 July 2006 (UTC)[reply]
    • A lot of my photographs are tonemapped HDR, since I usually shoot on a tripod it doesn't take me any more work to capture that way, and I've setup scripts to automate my processing. When it is done right they just look normal and natural to the eye (and a lot like the scene actually looked). The advantage of doing HDR photography and tonemapping is that you can capture difficult scenes without being unrealistic (I don't know about you, but I can see shadow detail fine without highlights clipping), this is possible because your instantaneous perceptual dynamic range is not that huge, so with the right compression an LDR image can look a lot like the HDR scene. In any case... --Gmaxwell 01:40, 7 July 2006 (UTC)[reply]
  • Oppose I don't think the scene is especially informative on the subject.. The classic "bare lightbulb in a desklamp" picture would be better. Perhaps an image of a desk with a bright light on top and a candle underneath. :)--Gmaxwell 01:40, 7 July 2006 (UTC)[reply]

Not promoted Raven4x4x 05:27, 12 July 2006 (UTC)[reply]