Commons:Featured picture candidates/Image:MIT Dome night1.jpg

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

Image:MIT Dome night1.jpg, featured Edit[edit]

Image:MIT Dome night1.jpg, Original not featured[edit]

A HDR image of the Dome at MIT 

result: 1 support, 0 oppose, 0 neutral => not featured. Benh 21:01, 25 November 2007 (UTC)[reply]

Image:MIT Dome night1 Edit.jpg, featured[edit]

A HDR image of the Dome at MIT

  •  Info Original version created by Fcb981, this version uploaded, edited and nominated by --Thermos
  • As the actual edits done for this image might interest some users, perhaps the editing done for this image merit some further discussion. After all, even though composure and such are often discussed, post processing is often left behind. In addition, quite often opposing votes are given on the basis of blown highlights or too much noise. Although I can fully understand these votes, I think that contributors could benefit from more guidance how to address the issues. I hope that this longish explanation will help at least some users.
  • Highlight recovery
To recover the highlights and bring out the surface detail on the dome, I used Adobe Lightroom (which as a dedicated RAW-tools is not even intended for this kind of editing, although it can be used on JPGs) and a procedure described in Evening's book "The Adobe Photoshop Lightroom Book." As a first step, I slightly reduced exposure with dedicated slider for that purpose. After that, I applied a small amount of highlight recovery tool. Then I went to curves tool and reduced highlights still some more. This was done in about one minute.
However, as this procedure is somewhat Lightroom specific (the highlight recovery tool at least, while the exposure tool is probaply found in image editing software intended for photograpers), for the benefit of users of other software, I tried if I could achieve comparable results by other methods.
By working with curves tool alone, by reducing highlight area by some 50% and lights area by some 25%, I think the final result was comparable. Perhaps with little less highlight detail, but the general appearance was "close enough". As curves is available in e.g. Gimp, perhaps this should be considered some kind of generaly available method.
For Lightroom users, I also tried if I could achieve comparable result with simpler method. I found out, that I could achieve "close enough" result by just adjusting exposure or recovery slider alone. However, I think that Evening's method results in better overall result and when you know it, it requires just few more seconds of work and should be used.
However, to avoid false impressions of what can be achieved with the "highlight recovery procedures" that I described above, out of curiosity I tried to apply the procedures to several images in Commons by other contributors. As non-scientific subjective hunch, I would say that with about 50% of the test images I could lift out significant amount of highlight detail, which was hidden in original image while leaving other parts of image relatively original. On the other hand, the other 50% of the images could be improved somewhat, but nowhere near the amount that I think this image benefitted from the procedures. Quite simply, this image was good to begin with and was easily improved. And ofcourse, this is not to critisise Fcb981, who contributed a wonderful image. It is just when I saw the original, I knew what could be easily done.
And as a final note for "highlight problems". Although this edit was done on JPG-image, in my opinion, it is much easier to address highlight issues if the original image is shot in RAW-format. Although properties of RAW will depend on camera, when the image is shot in RAW, at least according to tests I have read, there is usually quite a bit of more dynamic range available, which makes it much easier to avoid blown highlights.
  • Noise reduction
For noise reduction, the situation is quite different and dependant on the software. With this image I used a software called "Neat Image". I simply run the image through that software with basic settings and the result is what we have (when I now think about it, with more advanced settings even better results could have been achieved). Anyhow, I am happy with the results and with the ease it could be achieved.
Unlike the alternative methods for highlight recovery, for noise reduction I am not aware of alternate methods. Perhaps the "smart blur" method that Fcb981 mentions above could be such, but I am not familiar with it. Perhaps some Photoshop or Gimp expert could describe it. --Thermos 01:19, 13 November 2007 (UTC)[reply]
  • The highlight recovery you spoke of is useful. however when I converted from the 64bit uncompressed tiff HDR file I specified a highlight clip of 0.025%... so there weren't really "blown highlights" to speak of. Recovering from RAW files is effective although I wasn't aware there was extra highlight info in JPGs. So, I'm not sure if the highlight reduction was needed but it looks good this way. The nosie reduction is a little heavy for my tastes and the grass lost all texture but whatever. The editing is nicely done. Smart blur gives results similar to these. I think it's results are a little more crude then neat image but in photoshop it's better then median or despecle and "reduce noise" is too weak for me. Regardless, I don't usually mind if people make an edit that addresses noise. While I think the composition and light are more important I'm definitely supporting of people making themselves happy ; ) -Fcb981 02:11, 13 November 2007 (UTC)[reply]
  • Sorry. You are right about highlights. If they were really blown, i.e. without information, it would have been impossible to recover anything. And after thinking this some more, probaply the reason for the results I achieved with other images is that some images may just be adjusted to achieve more detail in light areas, while other images actually lack the information altogether. Would be reasonable conclusion. Anyhow, I hope this helps some users. And for NR part, I agree that on basic setting it is a bit heavyhanded... --Thermos 03:13, 13 November 2007 (UTC)[reply]
Here is the thing that used to be fun for me with GIMP -- perhaps it needs people who are well versed in pixel maths, photography and computer language -- was the decomposition of the technique into its components. The photoshop 'slider' would be controlling perhaps a couple of different constrained channel ops or layer modes or both at the same time. I watched one of them decompose the healing brush that way, for example -- where they figured out that it was just a brush constrained to use specific modes, yadda.
So, I realize, as I watch these images scroll by, that I am probably a fairly typical 'isn't that a pretty picture' kind of critic. That being said, I am also a 'isn't that an incredibly intelligent person' kind of critic and it would be really nice if there were people here who could figure out what the expensive software is doing. -- carol 07:23, 13 November 2007 (UTC)[reply]
result: 11 supports, 1 oppose, 0 neutral => featured. Benh 21:01, 25 November 2007 (UTC)[reply]
Retrieved from "https://commons.wikimedia.org/w/index.php?title=Commons:Featured_picture_candidates/Image:MIT_Dome_night1.jpg&oldid=8708670"