Commons:Featured picture candidates/Image:Fulmer Falls Closeup 3000px.jpg

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

Image:Fulmer Falls Closeup 3000px.jpg, featured[edit]

Fulmer Falls

Well the exposure time (~1/3 of a second) is stylistic, I won't disagree. If people don't like it for that reason, so be it. I probably have others with a faster speed. Update: I don't. If they don't like the long exposure, they can oppose, but it's the only one I have or can get. I would likely never be able to reproduce the exact conditions shown in this image: specific time of year, time of day, weather, naturalistic conditions in a failing Eastern hemlock forest. Within a few years, the hemlocks shown in this picture may all be gone due to the Hemlock woolly adelgid. It is not uncommon for dead hemlock trunks to clutter the image. See this image for an example. -- Ram-Man 12:32, 1 May 2007 (UTC)[reply]
I suppose you're referring to the water being slightly overexposed? I did bracket the exposure and this one was the best. Any darker and you lose the detail in the darker regions. I even had ideal lighting to eliminate even more contrasty lighting: taken in the evening near sunset on an overcast day. If I put up a lower exposure shot, someone else would complain about the underexposure. I've made large prints of this waterfall and the white water looks more than natural in this case. The water of the falls already spans ~65% of the image's range. Perhaps I'll see if I can reprocess the image. -- Ram-Man 20:46, 30 April 2007 (UTC)[reply]
Actually on a closer inspection, the picture looks quite noisy in the dark parts too. FP should be almost flawless. Lycaon 09:10, 3 May 2007 (UTC)[reply]
Are you serious, what magnification are you looking at? Poor image quality? It's a 7MP image, and the minimum requirements are 2MP. If we applied this standard to all images, barely a one of them would pass. There is at most 1 or 2 pixel widths of CA anywhere in the image. This is invisible. You'd have to make huge prints just to notice what you are talking about. Maybe I should downsample to 1600x1200 and apply photoshop filters? At that resolution the CA wouldn't be visible because there wouldn't be enough resolution to show it, but the lost resolution would then be noticeable at the magnfications you're talking about. And what artifacts are you referring to? -- Ram-Man 12:16, 1 May 2007 (UTC)[reply]
I'm looking at 100% which is what all voters should be doing. You are falling into the common error that more megapixels is more quality. Your image at 100% is poor quality. To illustrate the point, you can take a crop of a high grade image such as this and increase the res by 200% and you'll be getting similar quality to what your image is at 100%. That's not good. It's much better to resize it to a point where it is crisp and detailed at 100% - full res as I mentioned earlier of your current image has poor level of fine grade detail --Fir0002 www 06:34, 2 May 2007 (UTC)[reply]
  • Not even slightly am I falling to that error: If anything, blaming the camera (cheap?) for the "problems" indicates a false belief that a better camera will yield better pictures. If there are flaws that actually matter, they are mine alone as the photographer. I'm referring only to the standards for an FP, which states that 2MP file resolution is sufficient (saying nothing about the effective spatial resolution). This image has more than 2MP of spatial resolution as you've proven here. It's great that some images have more resolution than others (for a number of technical reasons), but unless we've changed the FP standards, it shouldn't matter. I don't care what it looks like at 100%, because either you are looking at an image crop (which makes it a different image) or at a large print size at an unreasonable viewing distance. If I wanted a crop of the image to look good, I would have taken a different picture. If you want to view a large print size, you may be better off with the extra spatial information that you lose from resampling. At least users can make that choice for themselves. I strongly disagree that voters should use 100% crops as their viewpoint, as 99% of the time that is not how the end-user will view it. -- Ram-Man 12:35, 2 May 2007 (UTC)[reply]
  •  Comment Just to be clear, are you saying that a down-sampled image, that is then enlarged to the original size will be better quality than the original at that size? --Tony Wills 08:17, 2 May 2007 (UTC)[reply]
  • Not at all!! Dunno how you came to that conclusion. I'm saying that the full 3000px image has very poor quality at 100% (a result of a cheap camera I suppose). I used the example of Big Ben to show what a high quality image is - not necessarily because it's high res, but because at 100% there is heaps of detail. So much detail, that I could blow up the image to 200% and get as good or better quality than this photo at 100%. I'm saying this image has to be downsampled to solve the problem of poor IQ at 100% --Fir0002 www 09:06, 2 May 2007 (UTC)[reply]
  • It was not a conclusion, it was a pointed question :-). So we have established that down-sampling looses information from the picture that can't be re-gained. Yes, of course we can hide deficiencies by down-sampling, and get the image past the eagle eyed FP/QI analysis. But do we end up with a 'better' image file? As the guidelines say, we can't predict what people will want to use the file for, by down-sampling we have destroyed a little info and restricted how users view the file (What if they want to print out a large poster, are they not better off with the original? A solution is to upload two versions, one for long term use, one to pass FP/QI scrutiny - but that's a wasteful solution given that the wiki has adequate software for scaling the image on the fly. Perhaps a better solution is to just specify the optimum resolution for the image (and set the FP display pages to serve it up at that resolution). --Tony Wills 11:23, 2 May 2007 (UTC)[reply]
  • I looked at that resampled image and find it oversharp with way too much contrast. I could take a similar picture with harsh contrast any day and adding photoshop sharpening later. However, getting a nice low-contrast image of a falls like that is very difficult without losing detail in the falls themselves or resorting to film or HDR/Photoshop tricks. Modifying it takes the photographer out of the photograph. I could have heavily post-processed the orginal image, but I don't like how that loses data through generational changes and restricts user's ability to use it. I don't see the purpose of tweaking an image to look good at an arbitrary resolution at 100%. Who even looks at images at 100%? The FP process is nice because it allows us to find and catalog some of the best looking pictures, but I'm not so obsessed that I will do anything I can for a FP. If it falls to been a FP, who cares? At least the picture is still useful. Also due to Bayer interpolation all digital cameras suffer from poor sharpness at 100% unless they are artificially sharpened or downsampled (possibly in camera). Some have slightly better spatial resolution than others, but they can never technically achieve 100% spatial resolution. -- Ram-Man 12:20, 2 May 2007 (UTC)[reply]
  •  Comment - I agree with Fir0002 when he says that downsample should be used to improve image quality, when necessary. As a matter of fact that is a common practise among our best photographers, like Dilif and Fir himself, and the only way I know to get the kind of detail and sharpness of images like this, when using digital cameras. If an image looks poor in the screen at full resolution, we may be sure it will look alike, or worse, when printed. Then, I believe that we won't loose important information when we downsample a picture in order to improove its quality. In my opinion, all pictures nominated for FP or QI should be reviewed in full size and declined whenever they don't look good enough at that resolution. Alvesgaspar 14:37, 2 May 2007 (UTC)[reply]
You make so many points, it's hard to address them all. I'm going to ignore the fact that many monitors are so poor that photo-sensitive printing will always look better and assume that everyone has sharp, high contrast, high quality monitors. Even so, it is total nonsense that downsampling improves overall image quality. Perceived image quality depends solely on pixel density (dpi) at a specific viewing size and a specifc viewing distance. The reason that downsampling appears to work is because it takes information that is only visible at higher pixel densities and tries to cram it into a lower density to make it visible. This is great if you want to optimize for web viewing on a 72dpi to 100dpi monitor, but it is very bad if you want higher dpi photo-sensitive prints used in a hardcopy format. This is the reason that this image looks great printed at ~220dpi at 11"x14" (I have the print right here), but not as good at 11"x14" on my 100dpi monitor viewed from the same distance. If I printed that same downsampled image at the same print size, the quality would be no better than 100dpi on my monitor even if though the print medium is capable of 300dpi. There would absolutely be a partial quality loss and we should not mandate that for our users who may not use a monitor for image reproduction. Downsampling removes the subtle gradations between pixels by eliminating some of those pixels and introducing changes in contrast or sharpness, either more or less depending on the algorithm used. The other problem with reviewing at "full size" is that it isn't standardized. Unless you require a specific pixel density, viewing size, viewing distance, and identical eyesight, you can't compare them all the same way. Change any one of them and the evaulation criteria changes. It's not the fault of the image that a 1024x768 CRT display at 72dpi doesn't look very sharp, but you can downsample it to look great under those specific conditions. Viewing at "100%" doesn't mean the same thing for everyone just looking at one of the factors! Viewing at 100% is the same as cropping. You're not comparing what it would look like if it was printed at, say, 300dpi, unless you change your viewing distance accordingly. You're looking at a different image. If my image was printed at 3 feet wide and you viewed it from 6 inches away, sure you'd see problems, but who does that? I believe it is common practice because these photographers know that it is usually required to achieve FP status. It just makes it appear to have more image quality when viewed under certain specific conditions. Some are just unaware of the misconceptions. I suspect some professional photographers intentionally downsample their images to prevent commercial usage by not looking as good when sold as prints. -- Ram-Man 15:55, 2 May 2007 (UTC)[reply]
result: 13 support, 4 oppose, 0 neutral => featured. Simonizer 06:40, 15 May 2007 (UTC)[reply]
Retrieved from "https://commons.wikimedia.org/w/index.php?title=Commons:Featured_picture_candidates/Image:Fulmer_Falls_Closeup_3000px.jpg&oldid=347424078"