Wikipedia:Featured picture candidates/MIT Dome

From Wikipedia, the free encyclopedia

[edit] MIT great dome

Reason

I haven't been doing HDRs that long but this was definitely a situation where it couldn't have been done any other way. The highlights on the dome were like 8 stops brighter than anything else... There is also an edit where some noise is reduced, from COM:FPC where this is passing pretty handily.

Proposed caption

A HDR representation of The Great Dome at MIT, which sits atop Building 10 and is featured in most publicity shots, is modeled on McKim, Mead, and White's Low Memorial Library at Columbia University.

Articles this image appears in

MIT

Creator

User:Fcb981

• Support as nominator Fcb981^{(talk:contribs)} 22:47, 12 November 2007 (UTC)
• Comment Is there an image without any noise reduction? The tone mapping just doesn't look right, and playing with the original, I get the impression that it itself was under a fair bit of noise reduction, and the edit, well, looks like plastic... thegreen J Are you green? 01:23, 13 November 2007 (UTC)
  • There is no noise reduction on the original. The edit is not mine. I can see bad noise anyway so people tell me things are noisy and with my calibrated monitor I don't see why they get worked up... what exactly doesn't look right? If you mean the posturization around lights in the inside, that is just a result of the HDR compression which sorts luminosity values and around light sources you sometimes see it. -Fcb981^{(talk:contribs)} 01:38, 13 November 2007 (UTC)
• <sarcasm>Oppose You can't see stars in the sky! It's sooooo underexposed!!</sarcasm> Support Either with slight preference to Edit. Good tone mapping, looks weird at thumbnail but realistic at full resolution. However, why is there a sharp border between the sky and the dome? --antilived^{T | C | G} 02:30, 13 November 2007 (UTC)
  • If you are wondering why it doesn't look like some HDRs where there is a soft boarder that is just PS3s merge to HDR thing, it is sharp. If you are wondering why there is a small halo... I don't know. -Fcb981^{(talk:contribs)} 03:16, 13 November 2007 (UTC)
    • There is a sharp edge, almost aliasing, between the dome and the sky. It looks like a badly done mask or something had been applied to it. PS: You used PS3 to merge it? --antilived^{T | C | G} 05:08, 13 November 2007 (UTC)
      • Yeah, File > Automate > Merge to HDR. it merges the files into a 64bit PSD file, which I converted into a 32bit tiff and then an 8bit JPG. and even though the metered difference was like 7 2/3 stops I only took 4 exposures. From the metered average: +5 stops, +2 stops, -1 stop, -4 stops -Fcb981^{(talk:contribs)} 12:43, 15 November 2007 (UTC)
        • • Oh you mean Photoshop CS3? PS3 is an acronym for something else you know... --antilived^{T | C | G} 22:25, 15 November 2007 (UTC)
            • =) ha ha, I meant to write CS3 not PS3, CS3 I didn't even notice. Now that you mention it, I should convert a PS3 to run CS3. I don't know where I am going with this... : ) -Fcb981^{(talk:contribs)} 23:21, 15 November 2007 (UTC)
• Oppose both I think a better picture could be taken. The original is not terrible, but the patchiness/graininess of the trees and shadows leaves it a bit short. The edit *is* terrible - the trees and foreground are blocky green smears - not one of our "best". Matt Deres 04:05, 13 November 2007 (UTC)
  • Stop being so pedantic. Of course a better one could be taken. If we were to say that if a better picture can be taken the picture is not FP then we would have a few high resolution scan of some ancient tapestries and nothing else! Instead of getting worked up over the grainyness of the sky or some loss of detail on some leaves, ask yourself if you think this is a visually impressive picture that shows some of wikipedia's best work. I think FP is about stunning photographic contributions. Something on a white BG is not stunning. An HDR where a bright dome is accented against some orange clouds and the building is properly exposed as well in my opinion is stunning. If you disagree that this is a beautiful picture, but if you just don't like that the leaves are a little smeared, I think you should look at this for what it is. Its a photograph, not an SVG graphic. -Fcb981^{(talk:contribs)} 13:23, 13 November 2007 (UTC)
    • Your reply is an excellent example of why people shouldn't nominate their own work. Calm down. Your picture is good, but it doesn't match the high standards set by pics like this or this, both of which are also night shots of large buildings. Look at them honestly - does the MIT dome have the same kind of clarity or detail? With the utmost respect for your ability (which far outweighs my efforts behind the lens), it does NOT. Hence, my opposition. The edit took the minor drawbacks of your photo and made them much worse. Matt Deres 04:28, 14 November 2007 (UTC)
      • • "...people shouldn't nominate their own work...". Please note that one of the two pictures you link to was a self-nom, and in fact Diliff who did both of them, along with many (probably most) of the best contributors we have, all usually self-nominate. --jjron 04:06, 15 November 2007 (UTC)
          • Completely true, Jjron and if Diliff responded to criticism of his photos with an ad hominem attack, I would say the same to him. But he doesn't. Matt Deres (talk) 05:38, 18 November 2007 (UTC)
      • Ok, it is true that the pixel dimensions of this are not as high as either of the diliff shots that you mentioned, those are stitched though. This one isn't so there is no way I could have gotten them that big. You are entitled to your interpretation of the FP standards, which is fine. But eventually, it comes down to this question: "Is this as good and artistic picture of the 'thing' that we can reasonably get and does it have visual appeal to the point that it would make a good main page picture?" I mean, I can go back, or diliff could go, during the day, and take a 15 segment stitched panorama and all that, and it would probably be less grainy, it would have more pixels but would it be better? I mean, how much to pixels matter anyway? sure, you can make larger reproductions, but when you blow it up on screen, you are lost. Sure, you can see every little rivet, but what does that have to do with the actual appeal of the picture. Do the colors evoke emotion? does it show the glow of the dome that could never be captured without HDR? does it make we look? stare? You yourself opened it up to full size. Presumably because it caught your eye, you liked the look of the dome, or the tone mapping or the subtle clouds. But once that full size opened, you were no longer looking at it for beauty. You were no longer looking at the colors, or the composition, or thinking about art. You were looking for noise, for lack of sharpness, for ghosting, for lens flare, for blown highlights, for crushed shadows, for stitching errors, for all those things that actually don't have much at all to do with photography. I believe that you know art, and I believe you know the technical aspects of photography. If you truely think that the technical aspects of this photograph get in the way of enjoying its beauty, I can live with your oppose vote. If not, I urge you to think about this picture for what it is. Does firs temp file with green lines showing god knows what and red lines show I don't know what else matter, no. Not unless it gets in the way of the picture. -Fcb981^{(talk:contribs)} 20:26, 14 November 2007 (UTC)
        • 8 Shots at 1 stop interval for HDR, times 6 for a 2x3 stitch, times at least 2 for errors/people in the building, and you shall have the über-stitched FP you want! --antilived^{T | C | G} 22:58, 14 November 2007 (UTC)
          • Aesthetics tho important are only one aspect of an FP - and to be quite honest it having a HDR look isn't overly stunning. I've seen plenty of good (and bad) HDRs and have made a few myself (eg Image:Yarra and bank by night.jpg) and so I'm not overly wowed by this example. So no it doesn't make me look. It doesn't make me stare. The colours evoke no emotion. Also if you took the time to read my comment, the green areas highlighted some areas of particularly poor quality due to what I assume was some very aggressive lifting of shadows. Red areas show sections where I believe digital manipulation and/or stitching/HDR merging faults have occurred. --Fir0002 07:25, 15 November 2007 (UTC)
            • You clearly looked long enough to come out of your wikibreak to comment on it and upload a temp file.<sarcasm> Tho I suppose you look at every fpc candidate in full size. </sarcasm> -Fcb981^{(talk:contribs)} 23:25, 15 November 2007 (UTC)
              •  ? I do view all FP candidates that I comment on at 100%. As for the wikibreak I'm glad to say that it's over! Y12 is finished so I'm back for a few months and will be reasonably accurate. However when I did comment on this I still had one exam to go and wouldn't have commented except that I feared it might receive a similar response to COM:FPC - big on looking at thumbnail but not so big on evaluating the image at full size. Having had a reasonable amount of experience with HDR I know that most HDR's do not come out particularly well at 100% - which is why whenever I see a HDR (and usually it's pretty easy to tell) I automatically assume it's not of high technical quality. And so if I see it up for FPC I'll view it at 100% and confirm/deny my fears. In this case my fears were realized and so I felt compelled to vote and point out the flaws to voters who might not notice them (as appears on commons). Without wishing to offend you, I'd have to say that stunning imagery was not the root of my interest. --Fir0002 11:25, 17 November 2007 (UTC)
• Oppose Very poor quality - noise is rife and the edit is hopeless, it makes the image worse if anything. I also think that this isn't a particularly good HDR - it's attempted to lift the shadows too aggressively resulting in very poor quality and artefacting (refer to highlighted green areas in my temp file). Also of concern are the red areas where it seems there has been a few stitching errors - although I have no idea what was going on in the right most red area - seems like a very poor clone job? PS: I would hesitate to nominate an image just because it is doing well on commons - they seem to have much lower standards in terms of IQ over there... --Fir0002 08:07, 13 November 2007 (UTC)
  • • Stitching errors? There is no stitching in this picture. I've had pictures that pass here and fail commons. So, one could reverse the argument. There is no cloning on the wall, only cloning of a sign that was on the grass out. one of the few areas you failed to highlight in red. -Fcb981^{(talk:contribs)} 12:36, 13 November 2007 (UTC)
      • Well what's going on with that area on the RHS - as far as I can tell it's either a clone job or some serious artefacting. No stitching? Perhaps it's coming through from a misalignment in the HDR sequence because that looks pretty dodgy. Could you provide the longest exposed frame so we can see whether it actually looks like that or if it is some kind of error. --Fir0002 07:25, 15 November 2007 (UTC)
  • But seriously, why are you looking at the grass and the tree? FWIW as long as the encyclopaedic parts (the building) is okay we shouldn't pick faults in the non-critical areas. Of course, there are way better pictures of grass and trees and other assorted foliages, but is it the subject of this photo? --antilived^{T | C | G} 09:16, 13 November 2007 (UTC)
    • • Ummm... because they're part of the picture. Photos have been slammed on quality issues much more minor than that - such quality is unacceptable even if it is not the "main" subject. We're not just featuring a small segment of the photo but the image as a whole. Hence it must be of high quality throughout. --Fir0002 07:25, 15 November 2007 (UTC)
    • They do take up about half the image. Matt Deres 04:33, 14 November 2007 (UTC)
      • So is the night sky, which is awfully underexposed to see the stars. Therefore this and all the other night architecture shots can't be FP! Seriously, complaints on the grass and the trees are totally irrelevant: They AREN'T the subject! They're simply minor elements, and it doesn't really matter if they're blocky or blotchy or whatever, unless it really cause a problem and distracts from viewing the subject (which doesn't apply in this case). --antilived^{T | C | G} 04:59, 14 November 2007 (UTC)
        • If the grass and trees were in the background, it would be perfectly fine for them to be blurred out somewhat. In a related way, we kind of expect the backgrounds in Fir's macro shots to be wiped out; at times, I think it actually makes the bug or bird look that much clearer and cleaner. We're okay with it because the bug could really be anywhere; we just want to see it well. The MIT dome isn't just anywhere - it's sitting on the lawn and is surrounded by the trees - they're part of the subject. IMO, while they don't need to have quite the clarity we demand for the building itself, as part of the subject they can't just be dismissed. At least, IMO. Matt Deres 13:09, 14 November 2007 (UTC)
          • Well the other thing with macro it is a technical impossibility due to the affects of diffraction to have much more in focus - this limitation does not apply here. And yes in a macro and wildlife shots bokeh is very useful in isolating the subject. But that isn't to say that noise or artefacty bokeh should be excused. It must be high quality bokeh. The "minor" elements here do not help isolate the photo and are of unacceptable quality. --Fir0002 07:25, 15 November 2007 (UTC)
            • Hmmm I wasn't the one that said your macro's background aren't good so I'll refrain from commenting on that, but are you saying that you can't tell the difference between the grass and the stairs/building? Or by "isolate" do you mean making the photo stand out? Isn't that the subject's job? Your HDR contains what, 3 stops of dynamic range? Everything is well lit (even the sky), and the effect of HDR is not that apparent. In this HDR, however, the difference between the dome and the grass is 8 stops (as said by Fcb somewhere above), and in real life it would be quite a dark place, without any direct illumination on the grass, versus a brightly illuminated dome. In effect it encompasses the dynamic range of the dark shadows of the trees and the street lamps in your HDR (bright light + white reflector ~= street lamp) without blowing out either, and retains a nice transition between the two. In 8-bit RGB this means 16 values for each stops for the texture, which is very little; personally I can't see the difference less than 5 values (also maybe because of the cheap 6-bit LCD panel), and this is probably the reason why the grass and the trees look smudged. If you want grass and tree textures, there are lots of high quality, high resolution, properly exposed pictures of them around, but are they the things that make this picture encyclopaedic? Are they the things that the picture is trying to illustrate? FWIW we can simply crop out all the grass and trees, does that mean that it's better? --antilived^{T | C | G} 10:54, 15 November 2007 (UTC)
              • I mean by isolate stand out - as in Image:Young grasshopper on grass stalk02.jpg. My HDR contains 4 stops (-2, 0 and +2) combined to make a realistic image. And in your comment (inadvertently I think) you've raised an issue which I thought might have been discussed before now in a different context. Because the fact that it portrays the a dim scene completely unrealistically - if you went there you wouldn't see it anything like that. Which IMO - unless this is on the HDR article - is a negative rather than a plus. Furthermore I think you'll find that the tonemapping from to 8-bit has little or nothing to do with the quality - I suspect it's merely the HDR engine attempting to remove all black areas (as they tend to do) and in the process lifting the shadows aggressively. Perhaps Fcb could upload the longest exposure image he took to confirm/deny this as I asked earlier? The smudging in the edit is solely to do with (fairly poor) noise reduction - the original suffers from noise and artefacting rather than the smudging of the edit (except in the RHS red highlighted area which looks really odd). True you can get high res images of trees and stuff, but that doesn't excuse the failings of these components - as I mentioned earlier even pure gradients like in the grasshopper image I linked need to be free of noise, artefacting etc (in short of high technical quality). The trees are part of the scene and must be judged as such. Cropping might remove some of this, but will destroy the composition. --Fir0002 11:25, 17 November 2007 (UTC)
• Support alternative, i supported this on Commons and i will here, too. I honestly can't see any of the problems mentioned by the opposers - the noise, which was a problem in the original is mostly gone in the edit. The grass in the foreground doesn't look very nice, i agree, but the crop in the edit makes the grass area less dominating. I like the composition a lot, and love the HDR/tone-mapping effect. --Aqwis 12:17, 13 November 2007 (UTC)
  • Perhaps the noise is gone in the edit, but so too has any texture - it looks smudged and plasticy. --Fir0002 11:25, 17 November 2007 (UTC)
• Oppose original and edit Looking at the article now and with the old image [1], I actually prefer the article with the old image. I think it's the fact that, in the thumbnail, there is more contrast in the dome itself than the new night shot. This by itself wouldn't be enough to get me to oppose the image (as Fcb981's image really is of much higher resolution and really does show more detail than the daytime shot), but the ghostly trees and lack of contrast between the outer column and the building do really kill it for me. Enuja (talk) 05:12, 18 November 2007 (UTC)
• Oppose a shame, as it's obviously taken a lot of work, but I can't see a version which doesn't look over-processed and artificial. At first, I thought this was only in the foliage, but I've just discovered that the dome is actually spotlit stone, not backlit glass as it first appeared... I love mapping but it's way too easy to go OTT with, and I'm afraid this is an example of exactly that. --mikaul^talk 18:34, 18 November 2007 (UTC)

Not promoted MER-C 02:39, 19 November 2007 (UTC)