Wikipedia:Featured picture candidates/Machu Picchu

From Wikipedia, the free encyclopedia

[edit] Machu Picchu

Machu Picchu Sanctuary and the Sacred Valley, showing the prominent peak of Huayna Picchu.

Downsampled, sharper version.

Restitched, brightened, and slightly downsampled version by Rubyk.

While we wait for Diliff, here's a stitch by me

And here's its edit

Diliff's re-stitch, slight crop, and adjustment of highlight detail

Another stitch which hopefully addresses the glitch issue

Downsampled, darkened clouds

Darkened, converted to sRGB from Adobe RGB instead of just assigning sRGB through save for web option

An awesome Machu Picchu panorama by User:Rubyk, from Ravedave's list on FPC talk - just one random click there, and this came up! (If the wiki server would be faster right now, I'd have a look at some other pics in the list...)

• Nominate and support. Note: The original image is over 2300 by 8300 pixels - and it is not too sharp at full resolution, and shows some artifacts. I downsampled it by 40%, and it became sharper - I prefer that version, even though it is smaller. I also did one other, subtle thing - can you find it, and do you approve? --Janke | Talk 16:27, 3 April 2006 (UTC)
• Support Hah what you removed was one of the things I thought was bad about the pic. I knew what it was before I even looked :) Great job photoshopping it BTW.-Ravedave 17:30, 3 April 2006 (UTC)
• Reluctant oppose. It has a lot of potential but there are some extremely obvious stitching marks in the image. I'd love to get my hands on the original files and stitch it myself. ;) If I could correct the poor stitching, I'd support. For the record Janke, I'm usually pretty strict on touching up images and removing details, but you did a good job and in this case I do approve. ;) Diliff | (Talk) (Contribs) 17:46, 3 April 2006 (UTC)

The uploader/photographer is currently active on WP, so I've asked him if he can provide the original shots. We'll see... --Janke | Talk 20:34, 3 April 2006 (UTC)

• Oppose edit. Not out of spite or plain principle. But when I consider what bothers me less, a duffelbag in picture which is full of people anyways, or having a retouched picture in an encyclopedic article, I'll go for the first option. By the way a slight blue shadow is still visible and it is clearly noticable where you took the replacement stone from. Irritates me! --Dschwen 09:18, 4 April 2006 (UTC)
  • I agree with you in principle. I think there is probably room to simply crop this part of the panorama out. Not ideal, but not a deal-breaker either.. Just a thought. Diliff | (Talk) (Contribs) 09:48, 4 April 2006 (UTC)
• Support downsampled and dufflebagless edit (its removal is clearly marked using Dschwen's template). Diliff can you help me spot the stitching errors you refer to? I'm having trouble finding them and I'm willing to hold my hand up and plead ignorance! ~ Veledan • Talk 17:57, 4 April 2006 (UTC)
  • Never mind, I spotted the repetitions in the vegetation at the bottom. Sorry, I have to withdraw my support, though I think it's great otherwise and I would fully support a re-stitched version :-( ~ Veledan • Talk 18:00, 4 April 2006 (UTC)
    • I did not note the stitching "doubling" either, until it was first pointed out. Shall we withdraw this nomination, while waiting for Rubyk's reply? --Janke | Talk 19:18, 4 April 2006 (UTC)
• I am willing to support the unedited. The stitching problems so far as I can see are mostly in the cloud areas. There are some problems with focus etc. but its got to be very hard to take a panorama like this IN the clouds, they're MOVING! The other blurred interpolated stiches at the bottom are mostly in unimportant areas of forest. The image is huge and the subject is interesting. --Deglr6328 23:32, 4 April 2006 (UTC)
  • I find the stitching more problematic below the clouds, personally. If you follow the stitch marks on the clouds vertically down, you will see duplication of features in the landscape due to poor stitching/feature matching in whatever software was used to create it. To be fair, they do blend in somewhat (mainly because the eye overlooks things that seem to appear normal. They aren't immediately visible, but they are definitely there on close inspection. It helps to know what to be looking for. Diliff | (Talk) (Contribs) 13:24, 5 April 2006 (UTC)
    • That's what I was talinkg about. I see them.--Deglr6328 04:29, 7 April 2006 (UTC)
• Oppose I can't for the life of me find stitch marks on the edited version. However, I am perturbed by the blue left behind where the dufflebag was. I would support the unedited. Nevermind, the stitch marks are indeed terrible! Search4Lancer 00:06, 5 April 2006 (UTC)
• Oppose Poor stitching. ~MDD4696 16:37, 6 April 2006 (UTC)
  • I oppose Diliff's edit. There is significant posterizing in the clouds! I support Rubyk's re-stitch or Fir's unedited stitch. ~MDD4696 23:22, 10 April 2006 (UTC)
• Support. A subject deserving so much detail. The thumb on this page seems too small in relation to the others. David R. Ingham 05:17, 7 April 2006 (UTC)
• Support. I appreciate you nominating my photo of Machu Picchu. I support the nomination, and am uploading a new version of the image without the stitching artifacts, slightly brightened, downsampled, and cropped to remove the duffle bag that has been bothering people. --Rubyk 14:32, 8 April 2006 (UTC)

Sorry, but the brightness increase is not really an improvement. Now the clouds are even more blown-out. --Dschwen 15:49, 8 April 2006 (UTC)

I agree. If you can upload a restitched version with the same brightness/contrast as the original, you'll get my immediate support for that. This is an awesome photo, and I'd really like to see it featured! --Janke | Talk 17:49, 8 April 2006 (UTC)

• • I'm sorry to to say the stitching artifacts are still very much visible. Not as prominent as the previous version but they are most definitely there. Rubyk, I suspect it is the software you are using. I am confident I can create a higher quality stitch if you can provide the originals. I also agree that the brightening is not really an improvement. Diliff | (Talk) (Contribs) 18:23, 8 April 2006 (UTC)

First of seven sequentially numbered source images.

• • • Here are the source images. Maybe one of you can do a better job at the stitching. Cheers, --Rubyk 00:14, 9 April 2006 (UTC)
• Thanks, Rubyk! Suggestion: When Diliff has made a stitch he considers perfect ;-), we could delete this nomination and put up a new one. As the nominator of this one, I'll happily withdraw it when a better one exists. --Janke | Talk 06:22, 9 April 2006 (UTC)
• Comment - Image:Peru Machu Picchu Sunset.jpg is already a featured picture. Although this picture is very good, is there really need to have two FPs of the same subject? Arco Acqua 15:21, 9 April 2006 (UTC)

I see no problem with that. Each FPC should be considered on its own. Cf. the lithographs by Haeckel, a fifth is on its way to FP status. --Janke | Talk 17:32, 9 April 2006 (UTC)

• I'd support one of these, but I'll wait until all versions are in before placing my vote. Current preference for the edit of my stitch --Fir0002 www 10:58, 10 April 2006 (UTC)
  • Uhh.. well. I feel like its almost not worth adding my stitch - it would just get lost in the list! Your stitch seems technically pretty good, but I don't really agree with your edit though. You always tend to over-brighten images (IMHO) and they lose the sense of atmosphere. Sometimes 'muted' is the actual (and intended) vibe. My edit is a slight crop, and attempts to restore some of the highlight detail in the clouds, but is otherwise left as-is. Should we re-nominate this/these images so we can get a proper vote?

Diliff | (Talk) (Contribs) 19:54, 10 April 2006 (UTC)

I went with the brightening because so did the photographer, and I guess that was the "vibe" he wanted - not the dark versions. --Fir0002 www 22:07, 10 April 2006 (UTC)

• Support Diliff version, oppose rest. Diliffs edit resolves two issues I had with the image, the stitching fault in center near the bottom of the frame to the right of the big rock (most likely a touch up, but barely noticable) and the blue bag. Fir's edit is a bit too bright, too big for the sharpness of the original material and still has the glitch. Especially the clouds look much better in Diliffs version. --Dschwen 20:33, 10 April 2006 (UTC)
  • Which glitch? I can't see any problems - prehaps upload a small temp file with the problem circled. And as you yourself have often vociferously said, bigger is better and there is virtually no limit to the space on the server. --Fir0002 www 22:05, 10 April 2006 (UTC)
    • As for the glitch check this magnification. As for the vociferously saying bigger is better, please differentiate. Sharpness matters as I said here. When asking for bigger versions of small but sharp images I tend be optimistic and hope for the larger version to be sharp as well. When I have the choice between an unsharp big version and a sharp slightly smaller version I wouldn't ask for the big version. And I definately wouldn't ask vociferously for that matter. --Dschwen 22:33, 10 April 2006 (UTC)
    • I dunno, it may have been too strong an adjective but "The originals however should be uploaded in maximum quality" is one of your recent remarks --Fir0002 www 08:51, 11 April 2006 (UTC)
      • I hate edit conflicts. ;) I'll leave my comments as is for the record, though: This glitch[1]. The foreground grass on the left segment has separated from the right segment. I agree that perhaps Dschwen has been a little contradictory there. However, I think he has previously said (and I completely agree) that it should as large as possible, but only where it does not visibly impair perceived sharpness. In the case of this image, the original files were quite soft, and downsampling resulted in an equally detailed but more aesthetically pleasing image. The scenario where Dschwen was advocating the 'bigger is better' mantra was where the image was obviously too small. Experienced photographers/editors like you and I can use our experience to dictate how far we can happily downsample. For the average contributor, I think its probably safer to request the original files and they can subsequently be modified to suit if need be, as the original image is always going to contain the maximum possible detail and we will never have to go through the laborious process of getting back to the original author to request a better version. Diliff | (Talk) (Contribs) 22:40, 10 April 2006 (UTC)
        • Diliff, there's significant posterizing in the clouds in your stitch... what happened? ~MDD4696 23:22, 10 April 2006 (UTC)
          • I agree - looks pretty horrible, noticeable banding in the sky. Too much "curves" or level correction in the sky area? (I see you darkened the sky a bit.) The sky in Fir's version doesn't have that problem. --Janke | Talk 05:52, 11 April 2006 (UTC)
          • You have a point there. Its more a matter of taste, though. The sky was very overexposed to begin with, so I had tried to make the most of what detail was available in the sky. I could have left it as it was, and thats fine, but I just tried to get a bit of texture out of the lower part of the clouds. At my end, the transition is reasonably subtle, but you would see a lot of banding if (for example - I'm not implying this is definitely the case) you had your colour settings set to less than true colour, but I don't see banding at all. I admit that due to the enhancement of the highlights in the clouds, there by logic must be more obvious steps in the luminosity of them, but I don't see it in the image at all, to be honest. I've viewed the cloud area at 300% and enhanced the levels on that crop by a factor of about 5, and while that results in an incredibly contrasty cloud ;), I don't see ANY significant banding... If the issue is not to do with your colour display settings, perhaps you could crop what you're refering to and show me? Alternative, I can simply upload the stitch without the cloud enhancement. I always knew that would be subject to taste, but I didn't (and don't) see banding. :) Diliff | (Talk) (Contribs) 06:46, 11 April 2006 (UTC)
            • Yeah I've also noticed the posterization. I've uploaded a temp file with some of the spots identified. --Fir0002 www 07:53, 11 April 2006 (UTC)

Posterization in the sky

• • • • • • • • Interesting. I'm at a work computer right now, and the LCD monitor has 8-bit colour so I can definitely see the posterization here... But as for whether its the monitor or the image itself, its impossible to say for sure (I see posterization on a lot of images with subtle transitions on this monitor). I see your point, however, because I've viewed both your image's clouds and I've viewed mine, and mine is clearly more visible. I'll revert it tonight and re-upload! This is really dragging on a bit but I'm glad we're making the most of the potential of this image through feedback. Diliff | (Talk) (Contribs) 08:30, 11 April 2006 (UTC)

• • • • • • • • • Yeah it certainly makes for a good final result! I mean once we've finished with it, I'm sure it will be as close to perfect as is possible. On that note I've uploaded two more versions, another stitch (to fix the glitch in the original) and an edit to bring out the clouds. Like you did, I experienced problems with posterization but I think I've been able to make a pretty good compromise. --Fir0002 www 08:51, 11 April 2006 (UTC)
                  • I like what you've done with the clouds and I think that is the best version overall, but, nitpicking here, the scene does look slightly washed out and I'd like to see it a little darker and closer to the original. For that matter, I'm not sure that the excess of bushes on the right really helps either, and I'd like to see it cropped on the right. Otherwise, I like it. Diliff | (Talk) (Contribs) 09:29, 11 April 2006 (UTC)
• There's another interesting difference between Diliff's and Fir's stitches - the perspective rendering. Look how the edge of the grass curves more strongly in Diliff's version - also visible as a steeper diagonal edge on the left. How come? (I do prefer the less steep curve of Fir's stitch.) All in all, this has been a very interesting and educational excercise for us all! --Janke | Talk 15:32, 11 April 2006 (UTC)
  • You're right. I did notice the perspective difference after comparing the two. I think I can explain this by the fact that the centre point isn't set correctly in one of the two stitches and the vertical perspective is shifted to different degrees. In viewing the two stitches sequentially, I have to admit that I PREFER the perspective of Fir0002's, but if you look carefully, the tree in the middle of the ruins is not vertical, whereas it is vertical in my stitch. Make of that what you will. Without any other perspective cues, it is very difficult to tell which is the more accurate - perhaps the tree actually does tilt in reality, although it doesn't appear to in the original images! Janke, seeing as you have been following this closely, can you see really posterization in the clouds in my image? I'm looking again on my home PC with what I consider to be a pretty high quality monitor, and I cannot see any banding or posterization at all. As I mentioned earlier, due to the modification I made to bring out the detail in the clouds, there will be a small amount of luminance (eg, approximately 10 steps at most) spread out over a greater range, but on my monitor, it appears quite well dithered and even when enhancing the clouds further and zooming to 300%, I still see dithering, not banding/posterization. I'm not saying that three different people are 'making it up', but I find it puzzling that they are seeing something that I am not. Is your display definitely set to 32bit/true colour? Diliff | (Talk) (Contribs) 17:52, 11 April 2006 (UTC)
    • I've just checked my monitor and it is on 32 bit color --Fir0002 www 22:35, 11 April 2006 (UTC)

• • • • Diliff: I use a G4 Mac, which is supposed to have good graphics... I definitely see the banding in your edit (top left), but not in Fir's (top right). Here is the proof, I cut out the same part from both images, put them one beside the other, and made a contrast enhancement. A pretty obvious difference, I must say. I also wonder where the "granulation" in your version comes from (only visible in the enhanced example, though.) Hope this helps you solve this mystery! (Maybe Fir has some secret trick, and that is the mystery? ;-) Greetings, --Janke | Talk 06:34, 12 April 2006 (UTC)
        • I finally see it, and it could be that my LCD screen IS a little washed out at the top end of luminosity, because if I tilt the screen at an extreme angle (which essentially darkens the entire image slightly) I DO see the banding, but when viewing it straight-on, I see very subtle variances of the almost-white level and no discernable banding. Seems to be only in the 250-255 range of luminosity, but that is a little worrying.. Something to be aware of in future! As for the "granulation", I'm not sure. It almost appears that his version has had noise reduction applied, as the tip of the mountain's detail seems a little softer, too. That could be as a result of the stitching though, or the fact that he didn't 'pull' the sky's detail as far as I did. I have no idea. Diliff | (Talk) (Contribs) 08:32, 12 April 2006 (UTC)
          • Ok i've uploaded a darker more saturated image. I support that version]] --Fir0002 www 09:22, 13 April 2006 (UTC)
• Comment I propose closing this as a failure, then opening a new FPC for the final edit, anyone agree? -Ravedave 16:15, 13 April 2006 (UTC)
  • Agree. Diliff | (Talk) (Contribs) 13:22, 14 April 2006 (UTC)
  • Agree. My dream candidate: Fir's dark stitch but without the saturation increase and with the bag just cropped instead of retouched. --Dschwen 15:30, 14 April 2006 (UTC)
  • I'd also like to see the bush on the right significantly cropped as it contributes nothing of significance to the view and makes the panorama's proportions unwieldy. Diliff | (Talk) (Contribs) 22:11, 14 April 2006 (UTC)

No result per above comments. Nomination would have failed based on vote count, but the image has obviously changed and improved a lot. Someone please select/create the new candidate soon — I want to see this an FP too! ~ Veledan • Talk 13:35, 15 April 2006 (UTC)