Commons:Featured picture candidates/File:Lockheed SR-71B Blackbird, NASA 831, over California (LCD).jpg

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

File:Lockheed SR-71B Blackbird, NASA 831, over California (LCD).jpg

[edit]

Voting period is over. Please don't add any new votes.Voting period ends on 10 Mar 2013 at 08:25:15 (UTC)
Visit the nomination page to add or modify image notes.

  1. The current FP is of a smaller size. My current nomination is 2.89 1.7 times larger (thanks to Colin for the correction).
  2. The current FP image's source (dfrc.nasa.gov) has the sRGB color profile. My larger file source's (nasa.gov Dryden Image Gallery) image did not have a color profile assigned at all, giving it a brighter and slightly whiter hue. Irrelevant, striked through.
  3. The current nominee is sharper. When I scaled the larger image down to match the FP image, the FP image was still softer. I can provide screenshots of pixels to show this. Not only this, but half of the FP image is stretched by one or two pixels.
  4. I also enlarged the FP image to 5100 x 3996 and found similar results: this larger image is not an enlargement, it's not blurred or soft.
I assigned the image above the Color LCD color profile, a Photoshop option that matched the colors and brightness of the original file. As Colin points out below, this may not be a standard profile. This was then converted again to sRGB. I did this to keep the image consistent. The alternative image below was immediately assigned the sRGB color profile, and its color and brightness matches the current FP image.– Kerαunoςcopia◁galaxies 08:25, 1 March 2013 (UTC)[reply]
  • I'm no expert but I don't think the colour profile stuff you are doing is right. Both images here are set to "AdobeRGB". This profile should not be used for web images. 99% of Wikipedia users will be using a browser, OS and monitor combination that are only capable of handling sRGB, and many will display the wrong colour if AdobeRGB is used. I will have a look at the source JPGs later if I get a chance. Converting a JPG from one profile to another risks colour artefacts like banding as it is not a lossless operation -- setting the profile should be done when first saving a JPG based on a raw or TIFF file. I don't believe there is such a thing as "Color LCD" profile -- this sounds like a name of the profile your PC has defined for your monitor, and is not standard. Colin (talk) 15:38, 1 March 2013 (UTC)[reply]
As I explained in the image descriptions, in the RetouchedPicture templates, after changing the images to either Color LCD or sRGB, the image was then up-converted to Adobe RGB (1998), which was just an automatic thing I have Photoshop set up to do (my working space). What I did was download two copies of the full image, opened one and assigned it Color LCD (I had no choice, if i didn't assign a color profile, the colors went berserk—this is the only option that retained the original look); the second image I opened and assigned sRGB, which darkened the image. Both images were then automatically "upconverted" to Adobe RGB (my working space). I can change this to be sRGB and re-do this process.
I've never noticed any problems viewing Adobe RGB images in my browser, and I didn't know browsers were limited to sRGB; all my other uploads look exactly the same when viewed in my browser as they are when viewed in two different applications I have. But if this is a concern, I have absolutely no qualm re-doing the above steps (the only reason it takes me a while is because i have to re-type up some of the metadata, which was not included in the original file).
As for Color LCD, I'm not claiming that's a real profile or not, but it's what Photoshop recommended because there is no color profile on the JPG. (I had the option of not assigning a color profile, and you should have seen the colors then!) Assigning Color LCD, whether it's a real profile or not, made the image look exactly as it does on the NASA site—on my computer, as I stated above—, so that's why I used it. If you think I should upload the original downloaded file without changing anything, then again, I'm willing to try that. I figured my assigning a profile would simply keep the image appearing consistent since it would finally be Adobe RGB.
Finally, you're absolutely right (about everything, I'm not arguing with you on any point, just clarifying), but if the larger image has no color profile, that's not my fault. Shouldn't one be assigned to it? I found that by assigning sRGB to the larger image (the alternative below), suddenly the colors and brightness matched nearly precisely (color sampler-checked) with the smaller image that is currently FP. In short, I had no choice but to assign a color profile to at least one image, however ultimately destructive it may've been. This larger image is still sharper, even after assigning a color profile, and I just wanted to update the FP. Already, someone on the SR-71 article confirmed the RGB image looks the same (color-wise, at least). – Kerαunoςcopia◁galaxies 16:51, 1 March 2013 (UTC)[reply]
That's a lot of talk. In short, here's what I'm willing to do (with your/anyone's approval before I go through all the work):
  1. Replace the above image (currently titled "LCD") with the original file from NASA.gov. No color profile changes (no changes at all).
  2. Replace the alternative image (currently titled "sRGB") with the color profile assigned as sRGB, and left as-is (no upconversion to AdobeRGB). – Kerαunoςcopia◁galaxies 16:56, 1 March 2013 (UTC)[reply]
Both images are now in sRGB. The image above matches the color and brightness of the original file photo; the image below matches the color and brightness of the FP image. – Kerαunoςcopia◁galaxies 18:28, 1 March 2013 (UTC)[reply]
  • I've now done some investigations. The original FP appears to be based on the image labelled "3000x2668 JPEG Image (4,406 KBytes)" on this dfrc.nasa.gov webpage and (according to the history log) is a lossless crop of that file to remove the footer banner. The above/below nominations are based on the image labelled "Download Full Size" on this nasa.gov webpage but have had some colour space fiddling done to them. I have come to the conclusion that the original FP is fine, the "larger" Nasa image is crap and there is something wrong with Keraunoscopia's colour setup on his PC. Firstly the original source image is 4,949KB and after cropping is 4,174 KB on Commons. This is 3,000px wide. The source of the new images is only 2,557KB yet is 5,100px wide. As Keraunoscopia notes, the former has an sRGB colour profile whereas the latter has no profile (which isn't uncommon on the web). Examining both closely with Photoshop shows no difference in colour to me but clearly shows the latter to have horrible JPG artefacts due to the high compression. A good example of this is the cockpit or pilot's helmet where the bright white is surrounded by loads of JPG gnats. There is no extra detail on the "larger" image at all. Someone has blown it up and saved it with high compression without indicating the colour space (which is almost certainly sRGB anyway). So I propose that actually the new images should simply be deleted from Commons as inferior versions of the FP.
As for why Keraunoscopia is seeing colour problems, see this (slightly out-of-date) article. I wonder if you are using a Mac or a PC, if you have a standard or wide gamut monitor, if you have installed the correct profile for your monitor, what browser & version you are using and what version of Photoshop you are using. In your Photoshop File Handling Preferences, have you got "Ignore EXIF Profile Tag" checked (it should be unchecked). -- Colin (talk) 19:08, 1 March 2013 (UTC)[reply]
 I withdraw my nomination above image only then (not the alternative). I'm on a Mac, Firefox 16 (and I see an school bus yellow car). Do you really not see any color difference between this and File:Lockheed SR-71 Blackbird.jpg? My "Ignore EXIF Profile Tag" is (and was) unchecked. This is both embarrassing and really depressing. I'm so sorry for having wasted your time (and wasting my own, I had another thing I could've been working on). – Kerαunoςcopia◁galaxies 20:01, 1 March 2013 (UTC)[reply]
No difference, in Firefox or Chrome or in Photoshop. I have Window7. However, I think the difference for you might be how the Mac interprets JPGs with no colour space. If you have the mixed joy of having a wide colour gamut display then this is closer to AdobeRGB than sRGB (often there are presets to switch between these on the monitor setup). According to that article I linked the Mac is perverse when it comes to JPGs with no profile -- it doesn't assume they are sRGB (which they almost certainly are) but assumes they are the colour profile of your monitor. If your monitor is set to AdobeRGB then it will display the wrong colours for such images.
But even with JPGs that are set to sRGB there can be problems with wide colour gamut monitors. The problem with some of the "better" browsers that are colour managed is that they appear (to me) to do the transform in a naive "nearest colour" mapping, whereas Photoshop/Lightroom will dither the colours when transforming and displaying the file. Although AdobeRGB has a wider colour palette than sRGB it still has the same 8-bits of limited range in the JPG, so those colours are further apart. A particular shade of red in sRGB may well have no counterpart in AdobeRGB -- Firefox will simply pick the closest and Photoshop will dither. This leads to banding in the sky with the browser for example but not in Photoshop. So if I use a colour managed browser on my display set to AdobeRGB, I can see posterisation in some images on the web. I don't get that problem if I view the same image in Photoshop which nicely dithers the sRGB JPG for my AdobeRGB display. So for web browsing and FP reviewing, I switch my monitor to sRGB emulation and use Firefox which I've set to be not colour managed -- no banding and the correct colours provided the image is sRGB. It is a bit of a pain. Colin (talk) 21:01, 1 March 2013 (UTC)[reply]
You know what, though, I think I went about this the wrong way. I should have never uploaded the above image or even brought up the color profile issue, because I think I created an unintentional distraction. So assume for now I was only nominating the alternate version below.
(When I compare the two images I linked to just above your last reply, I don't see any difference either. I only ever had a problem when I opened the no-assigned-color-profile image in Photoshop and had to assign one—my mistake in assigning the Color LCD profile. If I assign it sRGB, then it still looks exactly like the current FP, so I don't think I have a problem—and I'm certainly not seeing any banding in the browser.)
Ok next, you mentioned file size. (Sorry for getting my math wrong, you're right, it's 1.7 times larger.) I took the lossless crop from the FP image, and enlarged it to 5100 x 3996, and flipped back and forth between the large Dryden Gallery image and the enlarged FP image, and the large Dryden image was still sharper. I don't think the large Dryden image was enlarged, like you suggested above. If it had been, the noise in the desert background (at the very top, it's extremely easy to see) wouldn't be crisp, it would be blurred. Are you absolutely positive this 5100 x 3996 image is an inferior product? The pilots heads look like regular noise that I've seen on many other images on Commons; artifacts, sure, but crisp and clear, not what an enlarged artifact would look like, right? – Kerαunoςcopia◁galaxies 22:38, 1 March 2013 (UTC)[reply]
Well there's the size in pixels and the size in terms of bytes. The "larger" image is only 0.6x of the filesize of the smaller one. Yet is has 2.9x more pixels. How can that work? It is fairly obvious that it has been compressed to death. (BTW I'm talking about the source pictures -- the ones you have saved and uploaded were saved with a low compression setting -- and you get a large filesize as the new JPG faithfully records all the artefacts in the old one). Now open up both source pictures. Enlarge the smaller to 5100 wide using a good quality resampling algorithm. The look at the front cockpit. The triangular window has a nice white edge. The nearby edge of the aircraft that goes from grey to black is also nicely smooth. Compare with the new picture and you see loads of JPG gnats round those edges because JPG can't handle contrasting edges when highly compressed. Now look slightly up at the blue-tinged area and compare the two. The newer picture clearly shows the 8x8 structure of a JPG (it is composed of 8x8 tiles) with very sharp edges on the tile borders. The older picture shows none of the 8x8 tile artefacting because it has been saved with a low level of compression. I'm afraid any extra "detail" in the newer picture is just JPG artefacting and an illusion. Colin (talk) 11:59, 2 March 2013 (UTC)[reply]
Thinking a little more, it would be odd (but not totally surprising) for someone to enlarge the smaller picture. What is certain is that the smaller picture is not derived from the larger one, because it is so damaged by the high compression that it wouldn't generate the clean artefact-free smaller image. Therefore both images may have a common large ancestor. This is an old photo, pre-dating digital cameras. So the ultimate source is a negative or a print. There is a regular pattern noise on the images that suggests it might be a scan from a print. So there may be a TIFF on someone's hard drive that is large and uncompressed. I'm absolutely certain, that the large image we have here is so damaged by the compression that it contains no extra detail. Even if it wasn't badly compressed, it still might not contain more detail as the original might be somewhat soft or have artefacts from the printing process or film noise -- we won't know. Colin (talk) 12:55, 2 March 2013 (UTC)[reply]
I definitely see what you're talking about—I did as you write, which is what I did yesterday also. That extra sharpness was what I thought was simply noise, but maybe it's a bit too crystal sharp (there's no distinguishing mid-tones between certain pixels). Since the source is from actual film (which I was aware of), I was expecting to see film grain too, but I suppose I don't know what I'm looking for. So the sharp JPG pixels are artifacts and not a clearer image then. Well, I think I'm convinced. Definitely, both images must be derived from some master file somewhere, and I suppose that would be the only file to actually be worth locating. Ok, sold! I'm withdrawing everything. Once again, thanks so much for your help ;) – Kerαunoςcopia◁galaxies 22:17, 2 March 2013 (UTC)[reply]

Alternative

[edit]
Retrieved from "https://commons.wikimedia.org/w/index.php?title=Commons:Featured_picture_candidates/File:Lockheed_SR-71B_Blackbird,_NASA_831,_over_California_(LCD).jpg&oldid=92715730"