Commons:Featured picture candidates/File:1 taipei sunrise panorama 2015.jpg

From Wikimedia Commons, the free media repository
Jump to navigation Jump to search

File:1 taipei sunrise panorama 2015.jpg, featured[edit]

Voting period is over. Please don't add any new votes.Voting period ends on 8 Jan 2015 at 21:08:34 (UTC)
Visit the nomination page to add or modify image notes.

Original nomination with little chance of success
Pano of Taipei
I will ask Chensiyuan for a new version. I do however also think that the opposes above are pretty harsh for a 48-Mpix image and somewhat reflective of our pixel-peeping disease here. If the image were downsized to 20 Mpix, which is a perfectly acceptable resolution for a landscape, it would be just a bit noisy, probably not enough for anybody to seriously care, so basically high-res-curse strikes again. In addition, I think that those of us who have previously done high hfov rectilinear panoramas know that the quality decreases at the borders due to the projection, which is why patterns like noise become more apparent in those regions, but downsizing etc would remove detail in the center.--DXR (talk) 08:37, 31 December 2014 (UTC)[reply]
+1. It's completely normal for such a wide photo to lose detail and resolution in the corners due to the distortion. And looking at the photo at a slightly reduced resolution makes the issue go away. The quality in the centre is amazing, the quality in the corners is only good. Compare this with any featured non-panorama landscape and it's still far better, so I don't understand an oppose based on this problem. — Julian H.✈ (talk/files) 10:44, 31 December 2014 (UTC)[reply]
I am surprised that even you, Julian do not see these obvious technical issues as a problem. In the past you looked (at least at my noms) very thoroughly, which is good, and critized the tiniest amount of channel overexposure :) For an FP, DXR and Julian one can expect technical perfect processing - this is not the case here. My point is not border unsharpness but very unfortunate (over)sharpening of noisy areas - keep the noise as it is, sharpening does not help to make the noise looking better! BTW: It is not impossible to make high res panos with sharp and noise free borders. I agree that the quality might be OK downscaled to 20 Mpx - but that does not change my point that there should not be obvious processing mistakes for an FP and the image is not nominated as a downscaled version. -Tuxyso (talk) 12:38, 31 December 2014 (UTC)[reply]
The problem is that this approach will mean that we will keep on featuring our own stuff that checks all the formal boxes, but not other images even though nobody seriously denies their wow. I think that this is unfortunate and not in the best interest of "our customers", that is the passive visitors who want to see impressive, wow images in our galleries. I have contacted the author and I hope that we might get a reworked version of the image, but even if that is not the case I think that this panorama is among the best we have. I myself do what I can to upload very high resolution files in good quality and I am fully aware that you and many others do, but there is only so far we can travel and I fear that FPC was, is and will be eurocentric and a pretty closed circle of guys who rate each others' images and - like Colin said in the portrait review below - the best chance to get FPs is doing conservative and technically correct work that doesn't require much wow as long as the technical boxes are ticked.
BTW: your Yosemite image, which is very impressive indeed, is equirectangular (according to the EXIF), which like cylindric projection does NOT distort the borders. Rectillinear projection, which has been used here, does, so your argument not is not quite valid. Of course you can make multi-row panos etc., but for most panoramas rectillinear projection cannot give equal sharpness everywhere. That is why most church interiors are downsized, even though their center sharpness would be great even at 100 Mpix+ --DXR (talk) 12:58, 31 December 2014 (UTC)[reply]
Ok, I'll explain my thoughts in slightly more detail: You might notice that I have hardly ever opposed an image because of noise, because I don't think it's a big problem. Our brains are quite good at looking through noise and seeing the the reality behind it. This is not true for clipping and banding, which look very strange and artificial in my opinion. Especially if clipping exists but isn't at least fully white. But that's all that is, an opinion. It's true though that I'm often concerned about oversharpening, especially if it produces white lines around features with good contrast. Admittedly, that's the case in the corners here, but it's not extreme in my opinion, and most notably, it's not the case in the centre of the frame (at least not to a degree that I find distracting). Most people will view this photo either scaled down to see the whole image or zoomed in to the centre, and in those scenarios, the image isn't flawed. Even at the very bottom, it's still pretty much perfect. Add to that the amazing light, good composition and good resolution and I'm very much willing to accept the flawed quality at the left and right edges of the frame. Sure, I'd like the sharpening to be less there, but this is what we have. — Julian H.✈ (talk/files) 13:50, 31 December 2014 (UTC)[reply]
Tuxyso, your comment "image is not nominated as a downscaled version" is a viewpoint for judging I'm opposed to. We don't create versions of files on Commons that are downscaled simply to pass FP. In what way would that serve our mission? I wish there was an easy way to offer a link to a downscaled version of a file by making using of MediaWiki's thumbnail reduction API, but it appears to be crippled for large images where it would be most useful, and there also seems to be issues with sharpening artefacts in the current software. -- Colin (talk) 10:26, 2 January 2015 (UTC)[reply]
  •  Support I agree that there are processing issues that are very apparent at 100%. But with high-megapixel images, we have the choice to review at more modest dimensions. Reduced 60% as noted above and the noise disappears leaving a remarkably sharp picture that is well exposed, very detailed and full of interesting features. The processing issues are frequent with this photographer and I wish we had better communication so he can improve. But sometimes we just have to judge what we have, rather than what we wish or what we would have done differently ourselves. -- Colin (talk) 11:13, 31 December 2014 (UTC)[reply]
  •  Comment Fortunately, Chensiyuan has given me access to the raw files. I will try to do an edit myself adressing the comments above. --DXR (talk) 09:08, 1 January 2015 (UTC)[reply]

Alternative 1: Reworked in rectillinear projection[edit]

Edit Pano

  •  Comment Happy new year to all of you! I have reworked the image based on the raw files, tried to keep the look similar but address the noise/sharpening issue. I think that the issues pointed out above are now solved adequately. Julian, Tuxyso, Daniel Case, Kreuzschnabel and Colin, please have another look. Of course any further comments are welcome. Keep in mind that this is a 100° view, something one gets from a 15mm lens, so it is very wide and the view is a bit extraordinary.
  •  Support --DXR (talk) 12:59, 1 January 2015 (UTC)[reply]
  •  Comment Can I suggest you try another projection than rectilinear (e.g. General Panini in Hugin) to see if this minimises distortion at the lower corners. You could use the sliders Hugin has for that projection to keep the top half closer to rectilinear while the bottom remains closer to cylindrical. -- Colin (talk) 13:43, 1 January 2015 (UTC)[reply]
Colin, Diliff, here is the Panini general projection version. I am not a big fan of using Pan-gen or any other cylindrical projections in scenes that can be realistically captured with existing wide angle lenses. In addition, the bent street in the foreground that does not exist like this in reality is a EV problem for me. --DXR (talk) 10:59, 2 January 2015 (UTC)[reply]
  •  Oppose I've changed my support to oppose now as it is no longer the finest: the panini is much better. I agree the street curves but the rectilinear is not without its gross distortions: those buildings at the bottom are not really trapezium shaped -- the width/height proportions are all wrong as well as wonky angles. So at an individual-building-level, the panini projection is greatly superior and closer to reality for EV. One of the glorious features of a high-resolution panorama like this is being able to study it in detail, and that cannot be done as realistically with the rectilinear: if I fill my screen with any portion of the panini projection, it is more or less accurate with no curve visible but if I do the same with the rectilinear, then much of it is very unpleasant and not at all realistic. -- Colin (talk) 12:27, 2 January 2015 (UTC)[reply]
DXR, would you consider offering the Panini projection as an alt (and perhaps collapsing the original nomination image, which won't go anywhere now). I think that is the fairest approach (if everyone who voted is pinged) rather than having to go through a possible delist/replace afterwards. Alternatively, if this is too complicated now, just reboot the whole nom with the two projections. -- Colin (talk) 14:52, 2 January 2015 (UTC)[reply]
Colin, while I disagree with your assessment concerning the superiority of the panini projection in this case, I'm not too fussed about which version is promoted (as long as one of them is), so I have followed your advice. -- DXR 19:38, 2 January 2015‎
I understand your point of view. I have tried to match the look, which really is not that easy if you are starting from the raw files, but I don't rule out that somebody else could do a better job at this. One observation I have made, also based on the Toledo images I processed this summer (Original, Edit), is that Hugin imo is better at balancing exposure differences between frames. This does however also mean that nice, but somewhat coincidential, gradients in the sky disappear. --DXR (talk) 10:59, 2 January 2015 (UTC)[reply]

Alternative 2: Reworked in panini general projection[edit]

Edit Pano*

  •  Comment As suggested by Colin. I have expressed above why I prefer rectillinear projection, but of course the reviewers should be free to choose from both options. I'm sorry for the constant pinging, but just for fairness' sake everybody who voted above should be aware of the alt. Tuxyso, LivioAndronico, Diliff, Nikhil, Kreuzschnabel, Julian, Yann --DXR (talk) 19:38, 2 January 2015 (UTC)[reply]
  •  Support Sure! --LivioAndronico talk 19:57, 2 January 2015 (UTC)[reply]
  •  Support Better. Yann (talk) 19:59, 2 January 2015 (UTC)[reply]
  • Weak  Oppose. I don't really like the way it's curved the horizontal lines in the foreground. For this image, I think Panini is unnecessary. If the horizontal field of view was larger, it would be more useful. Diliff (talk) 20:38, 2 January 2015 (UTC)[reply]
  •  Support The comparison of different projections has imho a high educational value. The answer to the question which one of these is better is imho not trivial and a matter of personal taste. Our "naturalists" will argue that those curved lines are not "real" and are bad per se. For me some curvature in those wide panoramic views has an aesthetic merit. --Tuxyso (talk) 20:57, 2 January 2015 (UTC)[reply]
    • Well, I would agree with you if it were only a matter of taste, but we still have a responsibility to represent the subject accurately. I'm not saying rectilinear is more 'correct', but it is the projection that we usually expect to see and I think it is therefore the preferred projection except when excessive distortion makes it unpractical. I don't think the rectilinear distortion is enough that Panini projection is warranted, but that's just my opinion. Diliff (talk) 00:00, 3 January 2015 (UTC)[reply]
      • Rectilinear is not the projection we see since our retina is not a plane, and it is only a classical choice for 2D works for small angle-of-view (Diliff you will remember from our discussions last year that the great renaissance artists discussed this and thought something round 45-degrees (if memory serves me) was reasonable -- hardly 100-degrees). Our eyes in fact only see accurately in a tiny centre area (as I know you know) and outside this area is very blurred and much constructed by the mind. Therefore I'd claim that the Panini in getting the accuracy right in small areas is far more representative of what we see than the Rectilinear which completely fails to get the accuracy right at the building-level. Neither projections are 100% accurate but the gross building-distortion in the rectilinear is far more dishonest imo. Looking again at the rectilinear, the road may be straight in one axis but the buildings actually look like they are tilting to fall off the bottom of the screen like some disaster movie. -- Colin (talk) 11:15, 3 January 2015 (UTC)[reply]
        • I think it was far greater than 45 degrees, probably closer to 60 or 70. 45 degrees is quite conservative. I know rectilinear isn't the projection we see with our own eyes but it is the projection that we usually expect to see in a photograph - that's what I meant. I agree with you that the rectilinear projection has its share of problems relating to the distortion, but I generally prefer it as a projection because it preserves the relationship between objects slightly better and objects retain their true shape, albeit stretched along the plane. With the panini projection, objects become bent, and it gives the illusion that the angle of view is much narrower than it really is. This helps with distortion but comes with its own dishonesty. I accept that no projection is free of limitations though, and clearly I'm in the minority this time. Diliff (talk) 09:04, 5 January 2015 (UTC)[reply]
          • See this reference. "90° is the visual angle accepted since the Renaissance as the outer limit of images projected onto a plane... The practical limit for an acceptable visual cone has historically been a 60° circle of view — a suggestion first made by Piero della Francesca in c.1470 and repeated often since then. In fact, depending on the geometry of the principal form and the location of the vanishing points, a 40° circle of view or less is much more typical. Leonardo da Vinci devoted many pages in his notebooks (c.1490) to the analysis of perspective distortions, and he especially disliked the exaggerated apparent size of the perspective grid as it reached the ground line of the image plane. He recommended painting an object as it appears from a distance of 3 to 10 times its actual dimensions. This is equivalent to placing the figure within a 19° to 6° circle of view. In fact, modern vision research has found that most people say an object "fills their field of view" once it occupies approximately a 20° circle of view; the classical French rule has been to contain the image within a 30° circle of view." I'm not sure about your "true shape" claim -- the individual buildings at the lower corner are surely much closer to their "true shape" in the panini than the rectilinear. Only a mathematician could love the "relationship" those buildings have with reality. -- Colin (talk) 10:04, 5 January 2015 (UTC)[reply]
            • The difference is that painters like Leonardo had the ability to suspend reality and imagine the view through walls to create the perspective of their choosing. Photographers have to work with physical limitations. I particularly enjoyed reading "Only a mathematician could love the "relationship" those buildings have with reality" though. :-) I suppose you're right. The panini view looks better when viewing at 100%, because you just don't see the bending of straight lines at that magnification. It's more of an issue when viewing the image as a whole, whereas the rectilinear view becomes less distasteful, the greater the viewing distance. Diliff (talk) 10:21, 5 January 2015 (UTC)[reply]
  •  Support see above. — Julian H.✈ (talk/files) 22:16, 2 January 2015 (UTC)[reply]
  •  Support Much better. -- Colin (talk) 11:09, 3 January 2015 (UTC)[reply]
  •  Support per Colin. 😄 ArionEstar 😜 (talk) 12:50, 3 January 2015 (UTC)[reply]
  •  Support Kruusamägi (talk) 22:16, 3 January 2015 (UTC)[reply]
  •  Support Preferred. --King of ♥ ♦ ♣ ♠ 01:10, 4 January 2015 (UTC)[reply]
  •  Support--ArildV (talk) 10:39, 7 January 2015 (UTC)[reply]
Confirmed results:
Result: 9 support, 1 oppose, 0 neutral → featured. /-- ChristianFerrer 17:10, 10 January 2015 (UTC)[reply]
This image will be added to the FP gallery: Places/Architecture/Cityscapes
Retrieved from "https://commons.wikimedia.org/w/index.php?title=Commons:Featured_picture_candidates/File:1_taipei_sunrise_panorama_2015.jpg&oldid=611878863"