Wikipedia:Featured picture candidates/Hong Kong Skyline Restitch - Dec 2007.jpg
Appearance
- Reason
- This typical 'postcard' view shows almost all of Hong Kong's skyline (except from the air, this is, as far as I know and have seen, the most complete panoramic view possible) from Victoria Peak and is very detailed and well exposed. Due to the extremes of brightness in the scene, it is not possible to expose correctly for all elements with a single exposure - I took three frames for each of the 26 segments, exposed -2, 0 and +2 stops and merged these prior to stitching.
Taking the photo at night allows a very detailed view and highlights the many architecturally interesting skyscrapers of the city. Each of the buildings stand out far more than during the daytime.
- Articles this image appears in
- Victoria Peak, Hong Kong and Metropolis
- Creator
- User:Diliff
- Support as nominator Diliff | (Talk) (Contribs) 20:08, 23 April 2008 (UTC)
- Support original (per user:antilived below) Superb! — BRIAN0918 • 2008-04-23 20:31Z
- Comment Nice, it's a great view and very detailed - if I can make a suggestion, i think it would benefit from an S curve adjustment (gradient masked to avoid the top half of the image) to push contrast in the top a little. Also, I find that some wide radius USM works well after HDR to bring back a little local contrast, something like 20% at between 10 and 20 pixels for an image this size. It helps reduce some of the flatness that HDR leaves. Just a couple of suggestions, take them as you will, I tried them out on your image but held back from uploading it as an alternate in case you wanted to try it. Mfield (talk) 21:14, 23 April 2008 (UTC)
- Well, I hesitate because I don't know if extra contrast is what it needs in this case... Hong Kong was (and is, in general) quite hazy and that lack of contrast is the result. In any case, I have already boosted the contrast a little. That said, perhaps you're right. As a photographer, sometimes its hard to see when an image needs a subtle adjustment because you're too involved. I am working from a pretty contrasty monitor (Dell 3007WFP-HC) but have noticed that some monitors, particularly on laptops, are quite washed out. Did apply the wide radius USM (hadn't tried that before - interesting) and it certainly brought out a bit of punch but I'm still not 100% convinced it needs much more than it already has. Feel free to upload the edit and we have a look at it though. Diliff | (Talk) (Contribs) 21:47, 23 April 2008 (UTC)
- There you go, its a slight thing, fairly subjective to be sure but i think it is an improvement. The image in general reminds me very much of one of mine I took of Seoul at dusk. Mfield (talk) 22:26, 23 April 2008 (UTC)
- I like it, and I can see what you've done and why you've done it, but to me its six of one, half dozen of the other. :-) Its a subtle edit and I could switch back and forth over and over again appreciating different aspects of each edit. I suppose I'll leave it to the rest to decide which they prefer. Diliff | (Talk) (Contribs) 22:51, 23 April 2008 (UTC)
- There you go, its a slight thing, fairly subjective to be sure but i think it is an improvement. The image in general reminds me very much of one of mine I took of Seoul at dusk. Mfield (talk) 22:26, 23 April 2008 (UTC)
- Well, I hesitate because I don't know if extra contrast is what it needs in this case... Hong Kong was (and is, in general) quite hazy and that lack of contrast is the result. In any case, I have already boosted the contrast a little. That said, perhaps you're right. As a photographer, sometimes its hard to see when an image needs a subtle adjustment because you're too involved. I am working from a pretty contrasty monitor (Dell 3007WFP-HC) but have noticed that some monitors, particularly on laptops, are quite washed out. Did apply the wide radius USM (hadn't tried that before - interesting) and it certainly brought out a bit of punch but I'm still not 100% convinced it needs much more than it already has. Feel free to upload the edit and we have a look at it though. Diliff | (Talk) (Contribs) 21:47, 23 April 2008 (UTC)
Semi-weak supportIt's a stunning picture.But I question the level of smog? above the city. Is it possible to re-take such a photo with less smog?crassic![talk] 21:51, 23 April 2008 (UTC)- Its certainly not possible for me to re-take the photo any time soon, as I don't live there - it was just a holiday. And as mentioned below, this is pretty typical for Hong Kong. In some sense this is more encyclopaedic as a result. ;-) Diliff | (Talk) (Contribs) 22:51, 23 April 2008 (UTC)
- Okay. In that case, Support edit. crassic![talk] 02:13, 24 April 2008 (UTC)
- Its certainly not possible for me to re-take the photo any time soon, as I don't live there - it was just a holiday. And as mentioned below, this is pretty typical for Hong Kong. In some sense this is more encyclopaedic as a result. ;-) Diliff | (Talk) (Contribs) 22:51, 23 April 2008 (UTC)
- Support Great image. Hong Kong generally has about this much smog, or at least when I took a holiday there. See other photos showing the smog: Image:A view of Hong Kong.JPG, Image:Government House rightview.jpg, and Image:2 International Financial Centre.jpg. SpencerT♦C 21:54, 23 April 2008 (UTC)
- Support Edit 2 only. Too much HDR effect on the original; not a realistic image. If this were being used to illustrate High dynamic range imaging, I might support it, but I think it actually detracts from the Hong Kong article, as it is distractingly unrealistic. Kaldari (talk) 21:57, 23 April 2008 (UTC)
- Do you mean Oppose? It isn't actually HDR anyway, or tone mapping. Although its a somewhat similar process, the whole point is that it doesn't have the same 'HDR look' that tone mapping does. Obviously its a subjective thing and impossible to prove absolutely whether a photo is 'realistic', apart from having slightly better shadow detail, a normally exposed image would look quite similar. In fact I can upload a similar image to prove it to you if you'd like. This is not an overprocessed mess of an image as you seem to believe. ;-) Diliff | (Talk) (Contribs) 22:24, 23 April 2008 (UTC)
- I've just uploaded a quick-n-dirty stitch of the image taken only from the neutral exposures. As you can see, there is not a huge difference. It hasn't been edited in any way from the RAW file (hence it is slightly darker), other than colour balanced. The main difference, IMO, between the images is simply the shadow detail is slightly better in the original compared to Edit 2. Oh, and because I rushed it through, the projection of the panorama is a bit different. Diliff | (Talk) (Contribs) 22:47, 23 April 2008 (UTC)
- I think the reason the original version looks "overcooked"/HDRed is because there are no black shadows. For an image taken at night, you would expect something in the picture to be black. Otherwise it looks like the buildings are glowing unnaturally as you typically see in HDR images (example). Your 3rd image looks far more realistic to my eyes. Kaldari (talk) 23:11, 23 April 2008 (UTC)
- I agree that in a lot of HDR images, buildings appear to glow. This is due to the tone mapping, where local contrast is enhanced to provide extra contrast and to make the image appear to have greater dynamic range than it really does. But that is not the case in this image, where no tone mapping has taken place. I can see nowhere in the image where the buildings are glowing (apart from the obvious bright lights causing the hazy sky to glow, that is). And you're right in that usually in night scenes, there are areas of blackness, but this is Hong Kong! If the area truely was black, it would still be black in this image. All I've done is lifted the shadow detail slightly. Some parts of the image are still very dark, but ambient light means its not pitch black. Diliff | (Talk) (Contribs) 23:21, 23 April 2008 (UTC)
- I don't understand how you are claiming this image is not a high dynamic range image. Whether you are blending the exposures manually or by computer algorithm doesn't matter - it's still HDR. All that I'm saying is that by "lifting the shadow detail" you are making the photograph appear unrealistic. It looks almost like an illustration rather than a photograph. As Mfield's previous comment attests, it's obvious to anyone who sees this image that it is HDR (if they've seen HDR images). You may think that it looks great now, but in 10 years, when people get tired of the "HDR-look", the image will look dated and probably very specifically dated to around 2007–2010. Why don't you like the more realistic Edit 2? Kaldari (talk) 01:01, 24 April 2008 (UTC)
- Well, although as an exposure blend is an HDR image, albeit not a tonemapped one, that doesn't mean that it has to look incorrect. HDR images often resemble much more closely what the human eye and brain, with their adapative exposure system, perceive at the time. I would wager that Edit 2 looks less like what Diliff was looking at at the time with his eyes than the Original or Edit1, the skill is in finding the balance that matches what the eye saw with what the viewer of the image sees. A big problem is that viewers are accustomed to seeing low dynamic range images of high dynamic range scenes and perceive them as natural, when in fact they aren't natural either. Some of my edit suggestions are ways I have found to counter the side effects of exposure blending and HDR, I didn't witness the original scene so it's impossible for me to say how much it deviates. I have just noticed however that the blue lights on the Edit2 version on the tower just left of center are blue and pink on the Original. Is that a side effect of some PP or do they change color and that was captured in multiple exposures and reflected in the blend? Mfield (talk) 01:25, 24 April 2008 (UTC)
- You're absolutely right in saying that sometimes a well processed HDR image is more like what the human eye sees than what a typical camera captures with a single exposure, but as you say, a lot of people expect to see a more typical camera exposure (limited dynamic range). Well spotted regarding the colour of the lights. Yes, they changed slightly between frames, but I don't really see it as an issue, as long as they're blended nicely. Give it another 20 seconds and it would have been a completely different colour again. Diliff | (Talk) (Contribs) 02:49, 24 April 2008 (UTC)
- I think this is just a mismatch of terms. This is not HDR; there is no such thing as HDR in jpeg because jpeg is limited to 8 bits. Instead, Diliff took an HDR version and tone mapped it, i.e. he compressed the dynamic range to 8-bits. FWIW, the original does not look at all unrealistic. Thegreenj 02:14, 24 April 2008 (UTC)
- You're right to an extent, but I actually did not tone map it at all. At no stage was it a true HDR image (32 bits). I merely blended each exposure together so that the 'good' bits were kept (in the centre of the exposure curve) and the overexposed and underexposed bits were discarded. I find that this, when it works, does a far better job of maintaining the 'realistic' look of an scene, whereas tone mapping tends to mess things up (at least from my experience with Photomatix). As an example, here are two similar photos that I've taken. This one was turned into a 32 bit HDR image and tone mapped. This one was exposure blended. The tone mapped one could definitely be done better, and I don't claim they are otherwise identical (slightly different time at dusk, different weather conditions, leaves vs no leaves, but my point is that from my experience, you get a much more realistic image through exposure blending when done right. Anyway, I'm sure I'm annoying everyone with conversation in the middle of this nom so I'll leave it there. Diliff | (Talk) (Contribs) 12:31, 24 April 2008 (UTC)
- Well, although as an exposure blend is an HDR image, albeit not a tonemapped one, that doesn't mean that it has to look incorrect. HDR images often resemble much more closely what the human eye and brain, with their adapative exposure system, perceive at the time. I would wager that Edit 2 looks less like what Diliff was looking at at the time with his eyes than the Original or Edit1, the skill is in finding the balance that matches what the eye saw with what the viewer of the image sees. A big problem is that viewers are accustomed to seeing low dynamic range images of high dynamic range scenes and perceive them as natural, when in fact they aren't natural either. Some of my edit suggestions are ways I have found to counter the side effects of exposure blending and HDR, I didn't witness the original scene so it's impossible for me to say how much it deviates. I have just noticed however that the blue lights on the Edit2 version on the tower just left of center are blue and pink on the Original. Is that a side effect of some PP or do they change color and that was captured in multiple exposures and reflected in the blend? Mfield (talk) 01:25, 24 April 2008 (UTC)
- I don't understand how you are claiming this image is not a high dynamic range image. Whether you are blending the exposures manually or by computer algorithm doesn't matter - it's still HDR. All that I'm saying is that by "lifting the shadow detail" you are making the photograph appear unrealistic. It looks almost like an illustration rather than a photograph. As Mfield's previous comment attests, it's obvious to anyone who sees this image that it is HDR (if they've seen HDR images). You may think that it looks great now, but in 10 years, when people get tired of the "HDR-look", the image will look dated and probably very specifically dated to around 2007–2010. Why don't you like the more realistic Edit 2? Kaldari (talk) 01:01, 24 April 2008 (UTC)
- I agree that in a lot of HDR images, buildings appear to glow. This is due to the tone mapping, where local contrast is enhanced to provide extra contrast and to make the image appear to have greater dynamic range than it really does. But that is not the case in this image, where no tone mapping has taken place. I can see nowhere in the image where the buildings are glowing (apart from the obvious bright lights causing the hazy sky to glow, that is). And you're right in that usually in night scenes, there are areas of blackness, but this is Hong Kong! If the area truely was black, it would still be black in this image. All I've done is lifted the shadow detail slightly. Some parts of the image are still very dark, but ambient light means its not pitch black. Diliff | (Talk) (Contribs) 23:21, 23 April 2008 (UTC)
- I think the reason the original version looks "overcooked"/HDRed is because there are no black shadows. For an image taken at night, you would expect something in the picture to be black. Otherwise it looks like the buildings are glowing unnaturally as you typically see in HDR images (example). Your 3rd image looks far more realistic to my eyes. Kaldari (talk) 23:11, 23 April 2008 (UTC)
- I've just uploaded a quick-n-dirty stitch of the image taken only from the neutral exposures. As you can see, there is not a huge difference. It hasn't been edited in any way from the RAW file (hence it is slightly darker), other than colour balanced. The main difference, IMO, between the images is simply the shadow detail is slightly better in the original compared to Edit 2. Oh, and because I rushed it through, the projection of the panorama is a bit different. Diliff | (Talk) (Contribs) 22:47, 23 April 2008 (UTC)
- Do you mean Oppose? It isn't actually HDR anyway, or tone mapping. Although its a somewhat similar process, the whole point is that it doesn't have the same 'HDR look' that tone mapping does. Obviously its a subjective thing and impossible to prove absolutely whether a photo is 'realistic', apart from having slightly better shadow detail, a normally exposed image would look quite similar. In fact I can upload a similar image to prove it to you if you'd like. This is not an overprocessed mess of an image as you seem to believe. ;-) Diliff | (Talk) (Contribs) 22:24, 23 April 2008 (UTC)
- Support edit 1, weak support original Most of this technobabble goes right over my head, and while they're both excellent images I prefer the retouched version.--HereToHelp (talk to me) 23:35, 23 April 2008 (UTC)
- Support edit 1 Mfield (talk) 00:18, 24 April 2008 (UTC)
- Support original or edit 1 Both look great to me. Thegreenj 02:15, 24 April 2008 (UTC)
- Support either, slight preference for the edit Looks like a scene from Blade Runner... Matt Deres (talk) 02:27, 24 April 2008 (UTC)
- And I love Blade Runner. Haha. ;) crassic![talk] 11:39, 24 April 2008 (UTC)
- I think this one of mine looks way more Blade Runnerish. ;-) Diliff | (Talk) (Contribs) 12:31, 24 April 2008 (UTC)
- Ah, yes! Isn't it odd that Blade Runner supposedly presents a "dystopian" future, yet we seem to be using it as a blueprint for designing our living spaces...? Matt Deres (talk) 21:54, 24 April 2008 (UTC)
- No way, Blade Runner has plenty of fog. Papa Lima Whiskey (talk) 15:06, 26 April 2008 (UTC)
- I think this one of mine looks way more Blade Runnerish. ;-) Diliff | (Talk) (Contribs) 12:31, 24 April 2008 (UTC)
- Support either, slight preference for the edit The edit and the original are both absolutely stunning pictures with a remarkable level of detail. Axeman89 (talk)
Support edit. Papa Lima Whiskey (talk) 10:57, 24 April 2008 (UTC) Neutral. In the end, I can't convince myself that HDRI is necessary or beneficial to this particular scene. Edit 2 is probably reasonable (but not officially nominated?). Papa Lima Whiskey (talk) 10:40, 27 April 2008 (UTC)- But given that the HDRI (I even hesitate to use that term, this image has very subtle use of blending techniques, thats all, not HDR tone mapping) improves shadow and highlight detail, I'm not sure how you could say it isn't beneficial. Whether it is 'realistic' is of course open to interpretation, but I suspect that if I didn't mention any editing techniques in the nomination, the issue may not even have been raised as this scene could almost be possible with a single exposure, and I don't know if anyone could look at the photo and know for sure, without having tried. In fact, in a couple of years from now, if cameras start coming out with increased dynamic range capable of capturing this scene, would you decry them for it? This whole issue is significantly overblown IMO. Diliff | (Talk) (Contribs) 11:22, 27 April 2008 (UTC)
- No, I think we can see that it's HDRI. Papa Lima Whiskey (talk) 16:15, 27 April 2008 (UTC)
- I think that I could alter Edit 2 to look very similar to the original without using any HDRI whatsoever. The only significant difference would be the quality of detail in the highlights and shadows, not the brightness or balance of them. I think you have a preconception about what HDRI looks like. I can't really be bothered, but I could assemble a collection of images and ask you to determine if they were edited single exposures or HDRI images, and I think you'd get a significant number of them wrong (as would most people). It isn't as clear cut as you seem to believe. There are far too many factors at play here. Diliff | (Talk) (Contribs) 14:40, 29 April 2008 (UTC)
- I think that I could alter Edit 2 to look very similar to the original without using any HDRI whatsoever. What is your point? This image is a very clear example of HDRI, and your comment in fact exemplifies the increasing slipping of our grasp on reality here at FPC. The gold standard we should aspire to is for images to be passed as FPs without being edited at all. Papa Lima Whiskey (talk) 00:53, 4 May 2008 (UTC)
- My point is that HDRI doesn't automatically mean overprocessed, if it is done well and sometimes it serves to make an image look more realistic, even if you can identify it visually as being HDRI (although I don't think you always can with certainty). As for aspiring to images being passed without any editing at all, I don't think that is the right attitude. As I've mentioned elsewhere, image processing occurs the split second you press the shutter on a camera, whether you like it or not. Cameras essentially edit an image when it turns the captured data into a JPEG file, giving it sharpening, colour balance, contrast etc. Now by your logic, you would say that the output from a camera is the 'gold standard', but as soon as I process the image from a RAW file - essentially doing the same thing that the camera does - you seem to imply it isn't accurate anymore. It has never been as cut and dry as that. Cameras are quite capable of messing up the accuracy on their own. Diliff | (Talk) (Contribs) 08:38, 4 May 2008 (UTC)
- I was hoping you wouldn't bring that up, because it's utterly ridiculous and somewhat embarrassing imho. Of course jpeg is a compressed format. But the algorithm is standardised, at least for each camera line - the same algorithm applied to each image - and is honed to represent the image *as is*. JPEG compression is a necessary evil that camera makers chose to work with, both in early cameras because memory chips were small and expensive, and in some budget cameras currently, where the included chip is, again small. You and I know that no allrounder pro camera would add essential edits to an image out of the box, because it limits the artistic expression you can achieve with the camera (and get me right here so we can avoid another argument - I was talking about *artistic* expression for just a second there, not encyclopaedic faithfulness!). I even suspect that any such behaviour can be switched off if you know how. We also both know that in the age of cameras using film, you had to pick your film to match the lighting conditions, and different films gave different colour characteristics. But you knew that all pictures shot on a Velvia film would come out according to a shared standard, and a trained eye could tell a Velvia from a Provia etc. HDRI photography does not currently have this benefit, as there is no standard rendering, and many images come out looking rather clownish as a result. Your concept of "overprocessed" only serves to substantiate my concern: if you no longer care whether your image is slightly processed (as we've seen with the variety of edits being offered here), you're making a fundamental mistake. Maybe Hong Kong is lit differently to NYC, but after the last person has had their say and their wish granted, it may end up looking exactly like NYC. FPC is not about making a beautiful picture, it's about making an encyclopaedic picture. We have a time-honoured tradition here of sending "beautiful" pictures to Commons. Papa Lima Whiskey (talk) 09:22, 4 May 2008 (UTC)
- My point is that HDRI doesn't automatically mean overprocessed, if it is done well and sometimes it serves to make an image look more realistic, even if you can identify it visually as being HDRI (although I don't think you always can with certainty). As for aspiring to images being passed without any editing at all, I don't think that is the right attitude. As I've mentioned elsewhere, image processing occurs the split second you press the shutter on a camera, whether you like it or not. Cameras essentially edit an image when it turns the captured data into a JPEG file, giving it sharpening, colour balance, contrast etc. Now by your logic, you would say that the output from a camera is the 'gold standard', but as soon as I process the image from a RAW file - essentially doing the same thing that the camera does - you seem to imply it isn't accurate anymore. It has never been as cut and dry as that. Cameras are quite capable of messing up the accuracy on their own. Diliff | (Talk) (Contribs) 08:38, 4 May 2008 (UTC)
- I think that I could alter Edit 2 to look very similar to the original without using any HDRI whatsoever. What is your point? This image is a very clear example of HDRI, and your comment in fact exemplifies the increasing slipping of our grasp on reality here at FPC. The gold standard we should aspire to is for images to be passed as FPs without being edited at all. Papa Lima Whiskey (talk) 00:53, 4 May 2008 (UTC)
- I think that I could alter Edit 2 to look very similar to the original without using any HDRI whatsoever. The only significant difference would be the quality of detail in the highlights and shadows, not the brightness or balance of them. I think you have a preconception about what HDRI looks like. I can't really be bothered, but I could assemble a collection of images and ask you to determine if they were edited single exposures or HDRI images, and I think you'd get a significant number of them wrong (as would most people). It isn't as clear cut as you seem to believe. There are far too many factors at play here. Diliff | (Talk) (Contribs) 14:40, 29 April 2008 (UTC)
- No, I think we can see that it's HDRI. Papa Lima Whiskey (talk) 16:15, 27 April 2008 (UTC)
- But given that the HDRI (I even hesitate to use that term, this image has very subtle use of blending techniques, thats all, not HDR tone mapping) improves shadow and highlight detail, I'm not sure how you could say it isn't beneficial. Whether it is 'realistic' is of course open to interpretation, but I suspect that if I didn't mention any editing techniques in the nomination, the issue may not even have been raised as this scene could almost be possible with a single exposure, and I don't know if anyone could look at the photo and know for sure, without having tried. In fact, in a couple of years from now, if cameras start coming out with increased dynamic range capable of capturing this scene, would you decry them for it? This whole issue is significantly overblown IMO. Diliff | (Talk) (Contribs) 11:22, 27 April 2008 (UTC)
- Support any, prefer Original - Honestly, I trust Dillif's judgment here. de Bivort 12:59, 24 April 2008 (UTC)
- Support edit Well done. Lipton sale (talk) 18:31, 24 April 2008 (UTC)
- Support --Richard Bartz (talk) 18:53, 24 April 2008 (UTC)
- Support either Can't see anything wrong with it. Juliancolton Tropical Cyclone 17:54, 25 April 2008 (UTC)
- Support Any. Even if the first two versions are deemed "unrealistic", a full-size and properly edited Edit 2 would be sufficient. NauticaShades 21:50, 24 April 2008 (UTC)
- Support edit 2 only the other edits do in fact look quite different than the actual conditions shown in edit 2 not to mention the glare has increased in the exposure blended original, it does not look like a natural night scene Thisglad (talk) 08:30, 25 April 2008 (UTC)
- I did mention that Edit 2 is not for voting, just for reference.. At the very least, it is quite underexposed, as I did not post process it for actual use on Wikipedia. Also, I think the 'glare' you think you are seeing is probably just the ambient light reflecting off the haze in the sky. This is not a HDR related artifact, but was very real and is only more visible in the original because it is brighter (not wrong or fake, just brighter). Just my opinion though. If you're intent on supporting edit 2 only, I assume that would make your vote by definition neutral towards the original or edit 1? Or are you opposing? Could you clarify? Diliff | (Talk) (Contribs) 10:51, 25 April 2008 (UTC)
- Actually looking at again the main problem I have with the original is the blue color cast, I think the color balance is inaccurate, at least the sky should be more black than blue Thisglad (talk) 12:26, 25 April 2008 (UTC)
- The colour cast is identical in both the original and edit 2. Again, the only significant difference between them is the brightness and highlight/shadow detail. Brighten Edit 2 and you will see the colour is almost identical. As for the colour cast accuracy itself, because of what I mentioned about the haze in the sky, it will never be a true black because the various lighting of the city (incandescent oranges, coloured lights, intense whites etc). As such, you cannot choose the sky as a grey point (and if you do, the whole scene ends up looking very faded and unpleasant yellow). so in the end it comes down to an attempt to present the scene as accurately and aesthetic as you can, making certain compromises along the way. Its very easy to second-guess but far more difficult to please everyone. Diliff | (Talk) (Contribs) 12:59, 25 April 2008 (UTC)
- Why does this blue cast not show up in other images of Hong Kong? Just do a Google Image search for "Hong Kong at Night". None of those images resemble your original or Edit 1. Kaldari (talk) 16:14, 25 April 2008 (UTC)
- The example you provided does not have the low lying haze that mine does so you cannot compare the sky colour. During that day, the sky was a hazy, murky grey. Does that mean that the colour was 'fake' since the sky is usually blue? The thing is, doing a random search on Google isn't really fair. Most people don't know how to use their camera. Their camera will likely be left on auto WB (which almost always exposes too warm at night), or they'll take a photo with a flash, leaving it severely underexposed. But heres a couple of examples: [1], [2]. In the end, as I said, it comes down to aesthetics as much as it does accuracy, since there is no single light source to balance for. So maybe I adjusted it slightly too cool, but if I didn't, it would look quite murky. Diliff | (Talk) (Contribs) 16:25, 25 April 2008 (UTC)
- Why does this blue cast not show up in other images of Hong Kong? Just do a Google Image search for "Hong Kong at Night". None of those images resemble your original or Edit 1. Kaldari (talk) 16:14, 25 April 2008 (UTC)
- The colour cast is identical in both the original and edit 2. Again, the only significant difference between them is the brightness and highlight/shadow detail. Brighten Edit 2 and you will see the colour is almost identical. As for the colour cast accuracy itself, because of what I mentioned about the haze in the sky, it will never be a true black because the various lighting of the city (incandescent oranges, coloured lights, intense whites etc). As such, you cannot choose the sky as a grey point (and if you do, the whole scene ends up looking very faded and unpleasant yellow). so in the end it comes down to an attempt to present the scene as accurately and aesthetic as you can, making certain compromises along the way. Its very easy to second-guess but far more difficult to please everyone. Diliff | (Talk) (Contribs) 12:59, 25 April 2008 (UTC)
- Actually looking at again the main problem I have with the original is the blue color cast, I think the color balance is inaccurate, at least the sky should be more black than blue Thisglad (talk) 12:26, 25 April 2008 (UTC)
- I did mention that Edit 2 is not for voting, just for reference.. At the very least, it is quite underexposed, as I did not post process it for actual use on Wikipedia. Also, I think the 'glare' you think you are seeing is probably just the ambient light reflecting off the haze in the sky. This is not a HDR related artifact, but was very real and is only more visible in the original because it is brighter (not wrong or fake, just brighter). Just my opinion though. If you're intent on supporting edit 2 only, I assume that would make your vote by definition neutral towards the original or edit 1? Or are you opposing? Could you clarify? Diliff | (Talk) (Contribs) 10:51, 25 April 2008 (UTC)
- Support Orginal or edit 3 - a native Hong Kong resident friend of mine says the original is more faithful to what you would actually see than the edit, which is what I think is well, considering all the smog and possibly fog in the photo (normally the smog isn't THIS bad). This is not what you would expect directly from camera, but it matches what we actually see very well, and to me that is way more enc than the arbitrary camera response curve that we are all used to. --antilivedT | C | G 10:55, 25 April 2008 (UTC)
- Support Orginal - nice though could use some more contrast Capital photographer (talk) 17:48, 25 April 2008 (UTC)
- Support. Ain't this a screenshot from some science-fiction film? - Darwinek (talk) 19:59, 26 April 2008 (UTC)
- Support, with a slight preference for the 2nd version. This is a remarkable photograph. -- Alvesgaspar (talk) 23:16, 27 April 2008 (UTC)
- Neutral I am a native Hong Kong resident, and I can say that Edit2 (quick-n-dirty version) is the most realistic one. Edit2 separates the business district and the mid-levels residential area clearly. -βαςεLXIV™ 14:35, 28 April 2008 (UTC)
- Support for Edit 1 The brightness is much better in edit 1 than the original. --Dcelasun (talk) 11:37, 29 April 2008 (UTC)
- Note: its actually darker in Edit 1, not brighter.. Diliff | (Talk) (Contribs) 11:45, 29 April 2008 (UTC)
- Comment. There's a part of the central tower that is blue in the neutral exposure, but pink in the HDRI version. Why do we see this effect? Papa Lima Whiskey (talk) 14:22, 29 April 2008 (UTC)
- That was answered a bit further up. Basically the neutral exposure was, as described, a single exposure and the HDRI - I still hesitate to use that term, but it was multiple exposures. The tower's lighting was constantly changing, so when the three exposures were blended, it gave a slightly different colour. Thats it, nothing too overt, and not really misrepresenting the scene either, as it isn't a static colour. Taking another shot seconds later would have resulted in a different colour again. Diliff | (Talk) (Contribs) 14:28, 29 April 2008 (UTC)
- Oppose and
Obvious conditional support after fixt. Just noticed this on commons, amazing. Buildings are leaning to the left, which is easy to fix I believe, and which is why I oppose (stupidly one might say). I think one cannot get right exposure of night scene without a bit of tone mapping, so I prefer the retouched version. Blieusong (talk) 19:48, 29 April 2008 (UTC)- Actually I think you will find they are leaning inwards slightly on both sides. I'm surprised, with all of the controversy over this image, that nobody has mentioned this. It was actually kind of delibrate. I found that when correcting the tilt completely, the composition was not as good, because the edges of the mountain became more prominent in the frame, so it was a bit of a trade off. I'm glad you agree with me about the exposure, although as I said originally, it was not tone mapping. I used a program which you may be interested in. Rather than create a HDR file and then tonemap it, it calculates which pixels in a set has the best exposure on a bell curve and combines the image together that way. So it is similar to tone mapping, but quite different in how it works, and it usually results in a much more pleasant photo, without messing with settings too much (although you sometimes have to increase contrast a bit afterwards). I'll put a note on your talk page with more information about it. Diliff | (Talk) (Contribs) 19:58, 29 April 2008 (UTC)
- No, you're right. There is a tiny lean towards the left! I'll see what I can do about fixing it. You need to be pixel peeping to notice! Diliff | (Talk) (Contribs) 20:00, 29 April 2008 (UTC)
- me and my roomate user:sanchezn found the leaning obvious (and I'm surprised we are the only ones :) ). I needn't pixel peep (though I often do that with high res panos such as this one) to notice, I just put a building edge against the border of my monitor, to find out. Blieusong (talk) 20:46, 29 April 2008 (UTC)
- Yeah, its pretty easy to check the lean when you have so many skyscrapers to choose from. Your Paris skyline panorama has a big horizon curve though. ;-) I know, I remember you saying that it was delibrate. I was just looking at your commons gallery, lots of them should really be featured pictures on the En Wiki. I'm surprised they haven't been nominated. Diliff | (Talk) (Contribs) 20:52, 29 April 2008 (UTC)
- yup, unfortunately, there are many buildings eligible to be put against the edge of monitor ;). BTW, my oppose really isn't one, as find this picture beautifuly amazing and I know it won't change the results of this candidature. I just meant to get attention on the leaning. Thank you for the comment on my gallery. Maybe they are not nominated because they suffer from being compared to your panoramas ;) Blieusong (talk) 21:58, 29 April 2008 (UTC)
- I decided to go right back to square one and reprocess/restitch it completely to also address the issue of colour balance that Kaldari raised. While I maintain that colour balance in a scene like this is very subjective and complex with so many light sources, I think this may be a better compromise. And I still maintain that the dynamic range and accuracy of this image is far superior to Edit 2. What do you think? Diliff | (Talk) (Contribs) 22:00, 29 April 2008 (UTC)
- I think it's great !!! I strongly support Edit 3. I agree with you for the dynamic range point, and color balance wasn't a matter to me. Hope other people will notice this edit, now that this section has gone very far down this page ! I'll just put this edit on Commons FPC too. Blieusong (talk) 06:59, 30 April 2008 (UTC)
- I decided to go right back to square one and reprocess/restitch it completely to also address the issue of colour balance that Kaldari raised. While I maintain that colour balance in a scene like this is very subjective and complex with so many light sources, I think this may be a better compromise. And I still maintain that the dynamic range and accuracy of this image is far superior to Edit 2. What do you think? Diliff | (Talk) (Contribs) 22:00, 29 April 2008 (UTC)
- No, you're right. There is a tiny lean towards the left! I'll see what I can do about fixing it. You need to be pixel peeping to notice! Diliff | (Talk) (Contribs) 20:00, 29 April 2008 (UTC)
- Actually I think you will find they are leaning inwards slightly on both sides. I'm surprised, with all of the controversy over this image, that nobody has mentioned this. It was actually kind of delibrate. I found that when correcting the tilt completely, the composition was not as good, because the edges of the mountain became more prominent in the frame, so it was a bit of a trade off. I'm glad you agree with me about the exposure, although as I said originally, it was not tone mapping. I used a program which you may be interested in. Rather than create a HDR file and then tonemap it, it calculates which pixels in a set has the best exposure on a bell curve and combines the image together that way. So it is similar to tone mapping, but quite different in how it works, and it usually results in a much more pleasant photo, without messing with settings too much (although you sometimes have to increase contrast a bit afterwards). I'll put a note on your talk page with more information about it. Diliff | (Talk) (Contribs) 19:58, 29 April 2008 (UTC)
- Comment for closer. I hope I'm not misspeaking for the others, but as this nomination is so far down the list and not attracting any attention since I've added Edit 3, I wanted to point out that as the creator, I think it is the best version (Mfield's edit could equally be applied to it and I'd be happy to support those changes too, if consensus pointed to Edit 1) as the tilt is corrected, issues with colour balance have been addressed (hopefully to Kaldari's satisfaction) and the angle of view is slightly wider. Diliff | (Talk) (Contribs) 14:09, 30 April 2008 (UTC)
- Support edit 1 or edit 2 only, edit 1 has far better contrast in the upper half than the original, but my personal favourite is edit 2. Dorftrottel (warn) 19:27, April 30, 2008
- Out of interest, is there anything specifically you don't like about edit 3? Diliff | (Talk) (Contribs) 22:20, 30 April 2008 (UTC)
- Like the original, the upper half is too bright with fuzzy light, the entire image has an artificial, "HDR overkill" feel to it. Dorftrottel (talk) 23:29, April 30, 2008
- Obviously you've never been to Hong Kong or Mainland China recently, the air had gone down greatly in the last 5 years. My point is: it DOES look like that in real life.--antilivedT | C | G 07:25, 2 May 2008 (UTC)
- Yeah, its quite frustrating. It seems like these days in FPC, everyone is overcritical and just wants to give their unqualified two cent opinion and/or speak before thinking. Nevermind that an experienced photographer believes it is accurate, as do many people who have been there. Furthermore, the whole purpose of HDR is to compress dynamic range so shouldn't that mean that the bright 'fuzzy light' in the background would be darker than in Edit 1, rather than brighter, if it had a stronger HDRI effect? Yes, thats what it means. Oh well, what can you do? ;-) Diliff | (Talk) (Contribs) 08:31, 2 May 2008 (UTC)
- Obviously you've never been to Hong Kong or Mainland China recently, the air had gone down greatly in the last 5 years. My point is: it DOES look like that in real life.--antilivedT | C | G 07:25, 2 May 2008 (UTC)
- Like the original, the upper half is too bright with fuzzy light, the entire image has an artificial, "HDR overkill" feel to it. Dorftrottel (talk) 23:29, April 30, 2008
- Out of interest, is there anything specifically you don't like about edit 3? Diliff | (Talk) (Contribs) 22:20, 30 April 2008 (UTC)
- Doesn't matter to me personally what it looks like in reality. Edit 1 and edit 2 still look far better and also more authentic, and that is what counts. Dorftrottel (warn) 11:04, May 2, 2008
- So you are basically admitting that it doesn't matter what the reality of a scene is, it only matters what you think it looks like? This is Wikipedia, not Dorftrottel's website of things that he thinks look good and authentic. ;-) You might think thats what counts, but I certainly hope that the closer takes into account your disregard for those who have first hand experiences with the subject. Diliff | (Talk) (Contribs) 11:42, 2 May 2008 (UTC)
- I'm 'admitting'? What is that even supposed to mean? Did you accuse me of anything? Do yourself a favour and slow down on the rhetoric. And yes, like everyone else here, I'm providing input from my personal perspective, and I don't feel the need to give other people that kind of talk. Also, are you as strictly opposed to photoshopping tourists out of siteseeing hotspot images because it's not authentic? Dorftrottel (criticise) 12:00, May 2, 2008
- I think you need to look up the word 'admit' in a dictionary. Admissions don't always follow accusations, although they often do. To admit is to acknowledge. As for photoshopping tourists, yes actually, as a general rule I am opposed to photoshopping tourists out of images. Why do you ask? What relevence does it have to this? Are you questioning my consistency? ;-) Diliff | (Talk) (Contribs) 12:12, 2 May 2008 (UTC)
- Then write "acknowledged" when you mean acknowledged. ;-) You see, I lost some assumption of good faith right between "This is Wikipedia, not Dorftrottel's website of things that he thinks look good and authentic. ;-)" and your accusation against me of "disregard for those who have first hand experiences with the subject" totally out of the blue. ;-) Dorftrottel (ask) 13:18, May 2, 2008 ;-) ;-) ;-)
- Support, with preference for Edit 3. The color balance adjustment and wider view are improvements, in my view. The visibility of the haze is part of the image's encyclopedic value.--ragesoss (talk) 18:48, 3 May 2008 (UTC)
- Support all, prefer edit 1 or 3 Very interesting, illustrative picture. Edits 1 and 3 are very well done, although edit 3 shows a bit more mountain than I would like. --SharkfaceT/C 19:39, 3 May 2008 (UTC)
Promoted Image:Hong Kong Skyline Restitch - Dec 2007.jpg MER-C 09:09, 4 May 2008 (UTC)