Wikipedia:Featured picture candidates/Water drop animation

Animation of a water drop


A challenge was put up on Featured picture candidates/Detaching drop to have an animation, so I took about 300 pics of my sink and created an 18 pic animation of a falling water drop. Other user suggested me to nominate it, so here it is. I also added it to Drop (liquid), now the images make up more than the text of the article. I reduced the size to 768x1024 and 8MB, as the full size would be around 50MB. The time interval between the pictures is calculated to match the distance to the faucet under the assumption of a free-fall, ignoring surface tension (forgive me for not calculating these ;)
 * Update: Gmaxwell offered to redo the image using a specialized software tool and added a small preview to this nomination. I just sent Gmaxwell 12MB of images to work with, and am looking forward for the results! I would like to request an extension of the voting period to wait for and judge the new version from Gmaxwell.-- Chris 73 | Talk 13:56, 17 February 2006 (UTC)


 * Self-Nominate and support . - Chris 73 | Talk 11:18, 12 February 2006 (UTC)
 * Support clearly superior corrected Version by Gmaxwell. -- Chris 73 | Talk 09:17, 23 February 2006 (UTC)
 * Support: I think it is fantastic.  Nice job. Meniscus 12:53, 12 February 2006 (UTC)
 * I would like to direct my support vote to the new version. Meniscus 06:35, 17 February 2006 (UTC)
 * Support. Now that's a great animation. - Mgm|(talk) 13:07, 12 February 2006 (UTC)
 * Support. Nice job. Mikeo 13:14, 12 February 2006 (UTC)
 * Support --Dschwen 13:45, 12 February 2006 (UTC)
 * Comment I don't like the way exposure seems to differ between frames, as does the position of the tap and tiles. chowells 14:51, 12 February 2006 (UTC)
 * Opppose nice try, but I find the varying exposure and moving of the tap incredibly annoying to watch. chowells 18:55, 12 February 2006 (UTC)
 * Oppose - it's a good pic/animation, but the shadows behind it are too distracting. Is there any way to make sure that the light is constant? Thanks! Flcelloguy (A note? ) 15:37, 12 February 2006 (UTC)
 * I asked someone about this before, and they said that the camera's exposure time and aperture need to be fixed between images. Better digital cameras will let you manually set these options, or choose an "auto exposure lock", although my digital camera doesn't. &mdash; 0918 BRIAN &bull; 2006-02-12 16:42
 * I believe it was taken with a flash. Even with aperture/exposure length lock, the flash is difficult to control, as it is evaluatively metering and firing a burst of light that it deems is appropriate. Each shot would be slightly different in terms of flash output. Diliff  | (Talk)   (Contribs) 22:36, 12 February 2006 (UTC)
 * Still oppose. The newer gif image is too small, and the ogg file can't be displayed on most computers without downloading ogg vorbis. Can't the gif be enhanced? Thanks! Flcelloguy (A note? ) 21:23, 17 February 2006 (UTC)
 * How big should it be? I sized it to work well on the article at a number of display resolutions, but it was just a guess... if you have a concrete suggestion please provide it... It would be trivial for me to resize it (or you can do it, the theora version is suitable for resizing).  Currently at 60Kb in size I would be highly hesitant to inline anything much larger, and frankly would oppose anything too much larger due to considerations for users with slow connections. The old image when resized by mediawiki to the same physical dimensions was over 400K.  Animated GIF is a highly unoptimal format for 'video' like content, leading to huge file sizes and poor quality... Although it is what we have for inline content today, so we have to live with its limitations. As far as the video (Which is not vorbis, vorbis is an audio codec)... it would be foolish to provide the high quality version in animated gif because animated gif is never high quality due to the 256 color limitation. Ogg/Theora is the video codec used by Wikipedia, and is what we use for all videos. Like many other video codecs it doesn't ship with Windows, so many users will need to install it although we're working on adding a java player for Vorbis and Theora to mediawiki. Because Theora is our official codec, I'm going to go ahead and be so bold and advise you that you simply can not oppose media for featured status because of the use of Theora. --Gmaxwell 04:32, 18 February 2006 (UTC)
 * Another datapoint, if I allow mediawiki to thumb the new image to 10px less wide, the rendered file not only looks like crap (with dancing dithering), but is about 260K. Mediawiki scaling simply isn't acceptable for many animated gifs. --Gmaxwell 04:36, 18 February 2006 (UTC)
 * Maybe we could upload two gifs besides the Ogg, one as large as the common allows (up to 30MB), and a smaller version for display? Also, resizing my original version worked very well (see ), but GMaxwells version has display errors. Not sure how Gmaxwells .gif can be changed so Wikipedia resizes it correctly. -- Chris 73 | Talk 10:16, 18 February 2006 (UTC)
 * Your version resized is about 406k, mine is 60k. 406k is unacceptable for an inline image. The reason mediawiki barfs on resizing mine is because I'm not updating the complete image in each version, only the parts that change. If I disable this (i.e. unoptimize-gif under filter->animate in gimp) so that it doesn't get mangled on resize the result is 399k, unacceptable for an inline image. Furthermore MediaWiki resizing makes the video look like crud because it uses a per-frame randomized dither. Just say no to MediaWiki based resizing of animated gifs. :) As far as a large gif, I will not upload a 30MB gif when I've uploaded higher quality Theora file which is under 100k. Gif makes perfect sense for an inline image, but it makes no sense for the high quality version because it's both huge and looks like crud. --Gmaxwell 22:38, 18 February 2006 (UTC)
 * Ok, point taken. You surely do know more about gifs and images than me. Did we ever have a featured ogg? -- Chris 73 | Talk 23:32, 18 February 2006 (UTC)
 * Oppose although I would change my vote if the tap and tiles could be steadied and the background colour could be fixed. Maybe choose one of the backgrounds for all the frames? The water drop looks good!~ Veledan • Talk 18:09, 12 February 2006 (UTC)
 * Support (GMaxwell's edit) ~ Veledan • Talk 13:26, 17 February 2006 (UTC)


 * Oppose because of variable exposure and disturbing background. --CutterX 18:37, 12 February 2006 (UTC)
 * Oppose Seems jerky, and the most interesting part of falling water (the shape that rises up as the drop hits a water surface) is not seen. The varying exposures don't worry me at all - Adrian Pingstone 19:30, 12 February 2006 (UTC)
 * Support Very cool.--Lewk_of_Serthic 19:44, 12 February 2006 (UTC)
 * Oppose Uneven lighting and animation. &#126;MDD4696 21:44, 12 February 2006 (UTC)
 * Support. I think it is featured quality regardless of its imperfections. I know there should be basic requirements that need to be satisfied for all pictures, but I do think the difficulty and the uniqueness of pictures should hold some weight. Then again, if it fails, there is the incentive to go back and re-shoot it again. I know it took 300 frames to get right and the author may not want to do it all over again, but hey, the quest for perfection was never meant to be easy. ;) Diliff  | (Talk)   (Contribs) 22:31, 12 February 2006 (UTC)
 * Pretty cool animations, but I'd like a higher res version --Fir0002 00:40, 13 February 2006 (UTC)
 * The full 2592x1944 animation was 50MB, and the commons has an upload limit of 30 MB, hence I reduced the size to 768x1024 with a manageable 8MB. The image displayed on the image page is already reduced in size, click on the image to see full size. -- Chris 73 | Talk 06:42, 13 February 2006 (UTC)
 * I have a feeling that you'd be able to use the higher res version if you converted it into ogg instead of using an animated gif, although people would start opposing for that instead... enochlau (talk) 03:21, 14 February 2006 (UTC)
 * Oppose. Agree with MDD. - JPM | 03:09, 13 February 2006 (UTC)
 * Support the edited version. - JPM | 04:41, 17 February 2006 (UTC)
 * Comment Can the variable exposures be manually corrected with some kind of Photoshop action? Surely there must be a Photoshop guru out there. –Joke 02:27, 14 February 2006 (UTC)
 * Comment I've offered to correct the exposure and motion but have not heard back yet. I already have the video processing toolchain for this sort of thing setup. --Gmaxwell 03:29, 14 February 2006 (UTC)
 * Sorry for the delay. I tried to adjust the exposure manually using Gimp, but currently I am away from my main computer, and have no access to my photos. I also adjusted the position, trying to match the faucet position between the images. I'll fiddle around more on the weekend -- Chris 73 | Talk 07:36, 14 February 2006 (UTC)
 * What software do you use to correct the exposure between frames? I'm interested because I've done some time-lapse photography myself and needed to use evaluative metering rather than locked metering, so occasionally the camera chooses the 'wrong' exposure for a frame or two which results in a similar thing. Diliff  | (Talk)   (Contribs) 18:31, 14 February 2006 (UTC)
 * I used The Gimp for creating the whole animation, including adjusting the exact position, brightness (i.e. "Exposure"), resizing etc. Not sure what Gmaxwell uses for his video files. -- Chris 73 | Talk 18:57, 14 February 2006 (UTC)
 * Using the gimp will lead to uneven correction because the image colorspace is nonlinear. I answered Diliff's questions to me on my talk page since he wrote to me there...--Gmaxwell 03:01, 16 February 2006 (UTC)
 * Support new version. See my comment below below, 19 Feb. Considering that this is shot with a standard digital camera, the result is great. Not many wikipedians have access to equipment like Harold Edgerton ("Papa Flash") had... I wonder if the strange flutter in the exposure (as opposed to a plain change in exposure) is due to different paths of the flash discharge inside the xenon-filled tube? --Janke | Talk 07:34, 14 February 2006 (UTC)
 * Support Briseis 13:51, 14 February 2006 (UTC)
 * Oppose - Un-even lighting. I also think it's too big; expecially for those people who don't have broadband and are on dial-up. KILO-LIMA 14:02, 14 February 2006 (UTC)
 * Resize it. The pic on this page is much smaller than the original, and the copy on Featured picture candidates/Detaching drop is even smaller, so there is no problem for low-bandwith users unless they want the full size. -- Chris 73 | Talk 14:17, 14 February 2006 (UTC)
 * Support - Gladly change my vote note. Supprot new version. Well done! KILO-LIMA 15:50, 17 February 2006 (UTC)
 * Support Kessa Ligerro 07:49, 15 February 2006 (UTC)
 * Oppose. Yes, I think we can make this better before we give it FP status. - Samsara contrib talk 12:01, 15 February 2006 (UTC)
 * Support. Very nice animation -- Ian &equiv; talk 03:08, 16 February 2006 (UTC)
 * Support Calderwood 07:48, 16 February 2006 (UTC)
 * New version. It hurt me to see this potentially wonderful work being featured while it retained so many obvious and correctable technical flaws. I asked Chris 73 for the originals so I could correct the image, but he didn't seem interested in working with me. Normally I oppose making subjective changes to images which original author opposes, so I hope Chris 73 will support my version... but I feel that the changes I made were technical rather than artistic and that my modified version is objectively better. I have scaled the image to display size because the mediawiki rescaling for animated GIFs is far from optimal, I will also provide a full resolution version in a few minutes and update the entry with a link. This version is 60k when inserted into the page, which is a major improvement over the dialup crushing 400k of the version at the top. I must apologize for the some what poor quality of the full resolution version: I only had the large animated GIF to work with, and the dithering noise is non-linear and thus hard to suppress. I'd also like to ask that this vote be extended a bit to allow people to consider the new version.--Gmaxwell 22:21, 16 February 2006 (UTC)
 * Wow, that's so much better. I'd support that version. Shame it's not a tad bigger. chowells 22:59, 16 February 2006 (UTC)
 * Wish granted, see the link under the image. I *could* do an animated GIF of the higher res version, but the limitation of 256 colors really makes it look like crud. It would still look better had I started from the orignals rather than fighting against the dataloss in the animated gif. --Gmaxwell 23:32, 16 February 2006 (UTC)
 * Support new version. - Samsara contrib talk 23:48, 16 February 2006 (UTC)
 * supportGmaxwell's version Borisblue 04:06, 17 February 2006 (UTC)
 * Oppose Neat, but not all that interesting. I find it creepy.  Therefore, it would be more in line for chinese water torture, or water wastage articles.  Drip Drip -- Colle |[[Image:Locatecolle.png]]| Talk -- 04:11, 18 February 2006 (UTC)
 * And... I can't stop watching it. -- Colle |[[Image:Locatecolle.png]]| Talk -- 04:13, 18 February 2006 (UTC)
 * I think it's creepy too, and that why I thought it would be worth my time. It's a striking illustration. :) --Gmaxwell 22:38, 18 February 2006 (UTC)

Raven4x4x 07:04, 4 March 2006 (UTC) 
 * Support new version. Very cool. WP 09:11, 18 February 2006 (UTC)
 * Comment the playback speed of this animation (either version) is physically inaccurate. I noticed this when creating my own preliminary attempt. The drop spends much more time hanging on the faucet and slowly detatching. While still attached to the faucet, surface tension counters the gravitational pull, therefore acceleration is much smaller. For a FP I'd expect it to be correct. --Dschwen 12:04, 19 February 2006 (UTC)
 * Dschwen's objections are correct. As a consequence of this consideration I change my vote into oppose Calderwood 14:40, 19 February 2006 (UTC)
 * Indeed, I never noticed that, was so enthralled with the other improvements. Great observation, Dschwen! I struck out my support vote until this is fixed, then I'll support again. --Janke | Talk 15:58, 19 February 2006 (UTC)
 * Well, it's easy enough to adjust the timing. For lack of better information I just made the update match the orignal. Dschwen, how do you know your video is more accurate? It looks like it was recorded with the same multiple phase approach that distroys find timing informaiton. --Gmaxwell 19:47, 19 February 2006 (UTC)
 * No, it actually doesn't distroy timing information. I used every picture for my animation, just reordering them. Now simple statistics tells me that in this case the timing mut be correct. When the drop is moving slow at a certain position the probability of capturing it in a frame is higher then when it is moving fast. So more frames of slow moving drops are captured and in the resulting animation the drop moves slower. :-) --Dschwen 20:55, 19 February 2006 (UTC)
 * Does that mean the triggering of the shots were timed totally randomly? If not, then the photographer's choice of triggering the shutter surely destroys any statistic methods of determining the right timing... --Janke | Talk 09:30, 20 February 2006 (UTC)
 * As for my set, I switched to series mode and pressed the button for a few minutes or so, creating about a pic per second. The drop is just too fast for any good timing (the reality being faster than this animation) -- Chris 73 | Talk 09:57, 20 February 2006 (UTC)
 * I semi randomly shot pictures whenever the cam was ready, wit additional random pauses. I had a drop frequency of abt. 5/sec so I'm pretty confident in my sampling being random enough. --Dschwen 11:18, 20 February 2006 (UTC)
 * I'm amused... Getting the timing of this obviously enough right is enough of a challenge that we're unable to gauge the methods any multiple phase method will be disregarded as physically inaccurate but if I recorded this with a 200fps video camera it would be rejected as too low resolution. The featured picture process is a waste of time. --Gmaxwell 14:34, 20 February 2006 (UTC)
 * On the contrary. I really like the new constructive spin on FP (like with the carbon diagrams lately). Why the sudden frustration? The phase approach is the most promising. I believe I've got the timing right in my animation, but it is low quality since my drops were starting from slightly different positions, so it looks jerkey. Basically we'd have to take not 18 selected frames from the original pictureset, but all of them. That would preserve timing. This is totally worth a try. --Dschwen 14:59, 20 February 2006 (UTC)
 * I am also very optimistic. I think this process - and the great work of GMaxwell - will produce a better animation than what any of us could do alone. About using the full set of pictures: I have about 100 usable pics (or 50 MB of data). I removed some because of unsightly drops (i.e. just two drops following the big one instead of three). However, out of the 100 pics about 40% are very similiar with only a small bulge at the faucet and no visible drop. I would much rather favour a nicely distributed set of pictures with the drop at roughly equally spaced positions, and adjust the display time so it feels right. I am definitely looking forward for seeing GMaxwells work. Don't let us hanging (pun intended ;) -- Chris 73 | Talk 15:22, 20 February 2006 (UTC)
 * Support GMaxwell's version is cool. KI 22:50, 21 February 2006 (UTC)
 * Support The gmaxwell version.  psch  e  mp  |  talk  07:40, 22 February 2006 (UTC)
 * Support Gmaxwell's version, wish the animation was a tad smoother, but it is good for a fine example of work. Elle vécut heureuse à jamais  (Be eudaimonic!) 15:13, 22 February 2006 (UTC)
 * Support corrected version. –Joke 23:46, 22 February 2006 (UTC)
 * I can't believe I just sat here and watched water dripping out of a spigot for more than a minute. What is wrong with me? Support Gmaxwell's. -- LV (Dark Mark)  17:23, 25 February 2006 (UTC)
 * Support new version, good stuff. - Eagleamn 07:22, 26 February 2006 (UTC)
 * Support new version. - Mailer Diablo 10:10, 26 February 2006 (UTC)
 * Support new version (Jay 00:58, 28 February 2006 (UTC))
 * Support new version. --James 01:01, 28 February 2006 (UTC)
 * Support new version. Grace notes  T &#167; 17:30, 3 March 2006 (UTC)