Jump to content

File talk:Lenna (test image).png

Page contents not supported in other languages.
From Wikipedia, the free encyclopedia
(Redirected from File talk:Lenna.png)

Fair use rationale

[edit]

The image was being used to illustrate what a standard test image looks like in the article standard test image, and in that context it is replaceable. However, when used to illustrate the particular image it cannot be considered replaceable. --Oden 21:40, 12 January 2007 (UTC)[reply]

This is a historic image, and important when recounting the history of image file formats and test images. It's irreplaceable and should be kept in both articles.—Chowbok 19:00, 19 January 2007 (UTC)[reply]

Note: I have removed the replaceable template - there was a strong consensus to keep all of the Standard test images at Wikipedia:Images and media for deletion/2007 January 14 --BigDT 20:37, 19 January 2007 (UTC)[reply]

PNG?

[edit]

I know that PNG is the preferred standard non-lossy format for use on wikipedia, but wouldn't it be more appropriate to upload the original TIF file Alexander Sawchuk first scanned in 1973? As a reference image, you'd think you'd want the most accurate reproduction of the original possible; and since PNG hadn't been invented in 1973, recompressing it as PNG can only make it worse.24.68.145.153 (talk) 08:48, 23 December 2007 (UTC)[reply]

The PNG should store exactly the same pixel bits as the TIFF, losslessly. Dicklyon (talk) 00:09, 30 October 2011 (UTC)[reply]

Tag removal

[edit]

Fair use tags removed - Fair use has been established, re-established, re-re-established. JPEG tag removed - anyone adding this tag needs to read and understand the fair use explanation. This picture consists of a string of bits. Alter one bit, and it is no longer valid. PAR (talk) 13:11, 17 October 2008 (UTC)[reply]

Moved from talk:PAR

[edit]

You undid my removal of your Lenna image from her biogaphical page. The use of her image does not qualify as fair use on her biographical page, as any image of her could do here. There fore on the Lena Soderberg page it fails WP:NFCC 1. No free equivalent. The use of the image on the other pages is justified as the image is not replacable on the Lenna page, for example. However just because fair use is allowed on other pages doesn't mean fair use is allowed on any page. That is why no biographical pages of living persons have a fair use image. Rambo's Revenge (talk) 14:41, 12 December 2008 (UTC)[reply]

P.S. My original edit summary for removing it: "remove rationale for biographical person (as a free image can be made), just because they aren't cracking down on the use of this image does not make it public domain" was wrong. I didn't mean public domain, i meant fair use. 14:44, 12 December 2008 (UTC)

I restored the image and the fair use rationale, for reasons indicated on each page. Basically, the image was not presented on the biographical page as simply a picture but "as pictured in the Lenna image". As such, its use there and anywhere else is irreplaceable. Also, just because they aren't cracking down on the use of this image does not make it NOT fair use. PAR (talk) 19:42, 12 December 2008 (UTC)[reply]
Can we move this conversation to File talk:Lenna.png? —Chowbok 19:53, 12 December 2008 (UTC)[reply]

My two cents

[edit]

Here's my opinion, FWIW. This image unquestionably needs to be in Lenna and Standard test image. I'm not sure why it was deleted before, but it needs to be retained at least for those two articles. I think a reasonable argument could be made for including it in Playboy as well, since this is at least somewhat relevant to Playboy's history.

That leaves Lena Soderberg. Should this be used there? Obviously, in most cases we don't want a fair-use photo for identification purposes of a living person. In this case, however, since her chief claim to fame is this exact photo, I think it should be included in the article; not for identification but for illustrating that claim. I would want to underscore that this is not being used for identification, though, by not including it in the infobox in that article but lower down, in the text.—Chowbok 20:00, 12 December 2008 (UTC)[reply]

That would be ok if the Lenna-image notoriety wasn't discussed anywhere but the Lena Soderberg article. But this isn't the case. The image notoriety is discussed in an article by itself, and this article is linked from her bio. We should restrict the use to the article that's more relevant. --Damiens.rf 20:03, 12 December 2008 (UTC)[reply]
I don't feel very strongly about this one way or the other, but I will point out that File:Sharbat Gula.png is used both at Sharbat Gula and Steve McCurry. I dunno, maybe that's a mistake too.—Chowbok 20:34, 12 December 2008 (UTC)[reply]
Let's don't get into WP:OTHERSTUFFEXISTS. But where is the policy that implies "We should restrict the use to the article that's more relevant"? Shouldn't each use be considered on its own merits? Dicklyon (talk) 15:59, 13 December 2008 (UTC)[reply]
It fails the criteria (for the Lena Soderberg article). The criteria ask "Could the subject be adequately conveyed by text without using the non-free content at all". For the Lena Soderberg, the answer is no yes. The Lenna article can easily be summarised in text on the biographical page without the image, and a link to the Lenna page is provided for those that want to know more. Rambo's Revenge (talk) 21:50, 14 December 2008 (UTC)[reply]
I assume you meant to say "the answer is yes". Could the picture of Lena Soderberg "as pictured in the Lenna image" be adequately conveyed by text without using the non-free content at all? The answer is no. May I ask, in what way are you improving the content of Wikipedia by removing this image? In what way have you made this a better page to visit? What are the negative consequences of leaving the image as it is? PAR (talk) 03:40, 15 December 2008 (UTC)[reply]
But see my comments at http://en.wikipedia.org/wiki/Talk:Lena_Soderberg#Use_of_File:Lenna.png_in_this_article. If the photo is used in a way that is clearly NOT replaceable, then the policy you cite is not applicable. That's why I tried to edit the article into that form. Dicklyon (talk) 17:50, 15 December 2008 (UTC)[reply]

Improper deletion and reduction

[edit]

The 512x512 test image has been improperly deleted and replaced by a 220x220 image. The 512x512 digital test image itself is not copyrighted, and does not need to be reduced. Reducing the file means it is no longer a standard test image, which is now in conflict with the supporting text. The scanning of the original, copyrighted photograph to produce this test image is a historical event, which occurred in 1973. The 512x512 standard test image itself qualifies for fair use, as has been determined numerous times in the past. Please read the file description. I am unable to revert to the proper image, since it has been deleted. Do I have to create an entirely new upload, or can this deletion be reversed? PAR (talk) 22:22, 29 October 2011 (UTC)[reply]

I agree. And Playboy has stated that they're OK with the use; not exactly a license, but an approval of the fair use rationale, I'd call it. Dicklyon (talk) 00:10, 30 October 2011 (UTC)[reply]
No and no. Anything short of a free license or public domain makes it non-free for our purposes. Doesn't matter what Playboy thinks short of a release into the public domain or under a free license because it does not affect how our policies dictate that we treat non-free content. WP:NFCC#3b requires a low resolution image, and this image was reduced to the smallest size that we need - 220px width, the size that the articles the image is used in render it. There is no need for a larger non-free image. SchuminWeb (Talk) 00:49, 30 October 2011 (UTC)[reply]
There is a need if a person wants the test image itself. Why should it not be provided? The 220x220 image does not have the status of a test image, with all of the history behind it. It is simply a reproduction of a copyrighted image, and should not be allowed. The 512x512 image is that and more, it has been used in many, many published articles on image science, it has a history all its own and is unique. Can you cite a wikipedia policy that says that an image of this sort must be downsized to whatever size is being used in present articles? Can you cite a wikipedia policy that defines your term "smallest size that we need"? PAR (talk) 03:35, 30 October 2011 (UTC)[reply]
Some comments:
  1. Any trivial reproduction of a copyrighted work, including the 512x512 scan under discussion, is certainly still subject to copyright.
  2. I agree that it can be used on Wikipedia under a claim of fair-use since it is important for a reader to see the picture. Note though that our criteria for non-free content are stricter than standard US fair use.
  3. There is no need that we host the full-sized image as part of an encyclopedic article. A down-scaled version is sufficient for a reader to get an impression of the image and thus the topic, which is all we're aiming for, and the minimal extent clause from WP:NFCC#3b requires that we then limit ourselves to the reduced version. The above argument "a reader might want to use the full image" is in my opinion equivalent to "a reader might want to hear the full song" as part of an article on a song, or "a reader might want to read the complete novel" as part of an article about a novel.
  4. 220x220px is too small. Thumbnail size is configurable, and while 220px width is the default, it can be configured to up to 300px. Square images should thus typically be reduced to 300x300px (as most music cover images are).
Amalthea 11:27, 30 October 2011 (UTC)[reply]
But its important to realize that it is the original picture in Playboy that is copyrighted, NOT the test image itself. When a reader looks at the test image, they are not looking at "the picture", they are looking at a very reduced scan of a portion of the copyrighted picture. The test image is not a "full-sized image". To say "a reader might want to hear the full song" is not the same as "a reader might want to hear the one percent of that song that has been established as fair use" (equivalent to the test image). What you are saying is that since the original song is copyrighted, we must reduce that 1 percent to one quarter of a percent, which makes no sense. It is the test image that Playboy will not contest, since it serves them better as an advertisement. Any other reproduction of their copyrighted photograph is questionable. PAR (talk) 12:33, 30 October 2011 (UTC)[reply]
As a derivative work of the original, the test image would not create a new copyright. If you can submit proof that said image is in fact not copyrighted, then we can entertain the matter of copyright. SchuminWeb (Talk) 18:10, 30 October 2011 (UTC)[reply]
Amalthea, I have no issue raising it up to 300x300, since I was not aware that thumbnail size was configurable that high. SchuminWeb (Talk) 18:10, 30 October 2011 (UTC)[reply]
A 300x300 is equally useless; you two are missing the point. Even at 511x511 it would be useless for its purpose. Dicklyon (talk) 18:15, 30 October 2011 (UTC)[reply]
I strongly agree. This image is used everywhere in the graphics industry as a free image and has been used in this way for decades. Playboy have clearly said that this is OK. Changing the image in any way whatever makes it no longer representative of the test image - every single list bit of the file is utterly crucial to what it is. We've been through this kind of nonsense at least a couple of times before and the answer is the same. The image - as used here - is 100% OK. No material facts have changed since the last time we had this debate - so I refer the originator to those discussions and request that we cease re-re-re-debating this each time someone else who doesn't understand the meaning of "Standard test image" comes along. SteveBaker (talk) 21:19, 30 October 2011 (UTC)[reply]
If the image is useless for our purposes at the low resolution required by WP:NFCC#3b, then we should just nominate it for deletion, then? SchuminWeb (Talk) 16:06, 31 October 2011 (UTC)[reply]
Will you please read and address this statement directly: The test image is already reduced as per WP:NFCC#3b. The test image is not copyrighted, and does not need to undergo a second reduction. PAR (talk) 17:49, 31 October 2011 (UTC)[reply]
Perhaps SchuminWeb is lacking the history here: This image comes from a double-page spread centerfold of some ancient Playboy magazine. The image was scanned at what would now be considered to be very low resolution - and heavily cropped. The resulting 512x512 low-rez crop that we are using is NOT copyrighted in it's own right - the original university department that scanned it have been giving it away into the public domain forever and Playboy have stated that they have no interests in it. Hence, any possible copyright infringement can only be in relation to the original double-page centerfold. But we are only using a small part of that double-page image - and at incredibly low resolution. Hence we are already fully compliant with WP:NFCC#3b with respect to the image that Playboy originally published. To the degree that cropping and scanning might be considered an artistic work in it's own right, Wikipedia has long said that we do not consider such things to be 'creative' so that no new copyrights have been created by so doing. Furthermore, WP:NFCC#3b points out that so long as we use the minimum image resolution necessary to convey the message, we're OK. In this case, the minimum is the 512x512 image because this is a test image for image processing applications. Issues such as the sharpness of the image and the noise created by the photographic processes and the precise color of each pixel are all CRITICAL to understanding the importance of this image as a standard test. By reducing the resolution by even one pixel, you destroy the ability to look at the image and understand it as a standard test, which is the context in which we're using it. We're not concerned that this is a picture of some porn actress - we're concerned about the relative brightnesses of the 342nd and 343rd pixel from the left on the 399th scanline from the top. The minimum resolution IS 512x512 or the message behind the image is quite utterly destroyed.
Furthermore, the scan of the original article is an especially poor one. The color balance is all off - the skin tones are quite wrong. This is a severe distortion of the original Playboy image - as well as being low-rez and heavily cropped.
Use of the image in the Lena Söderberg article is much more tricky. I can see that if we're talking about Ms Soderberg's contribution to the world of image processing, then there would perhaps be justification in using the image at full resolution...however, if it's just being used as a photograph of her in general - then I would agree that using even less resolution than 512x512 would be tenable. However, in the "Lenna" article, we absolutely must have 100% of the 512x512 image and it must be presented in a non-lossy file format such as PNG and not in a lossy format such as JPEG or GIF. SteveBaker (talk) 19:42, 31 October 2011 (UTC)[reply]
Well, Schuminweb refuses to acknowledge, read, understand or respond to any objection to his error, he only repeats the error. See User_talk:SchuminWeb#Lenna.png. I am certain there is an administrator somewhere that will at the very least acknowledge and respond to the argument. PAR (talk) 22:13, 2 November 2011 (UTC)[reply]

Sorry for not following up sooner.

PAR, Steve, let me reply first to a common point you make: you say that the 512x512px scan is in the public domain, not copyrighted. I doubt that this is true. In fact, our own article hints at the image being copyrighted, and that Playboy is merely not taking action and has "decided to overlook" it. By default, a mechanical reproduction of a copyrighted image is still subject to the same copyright. It does not matter who scanned it and what they decided to do with it. Can either of you offer proof that the copyright holder, Playboy, has released this particular image into the public domain, and is not merely tolerating the widespread use? Amalthea 10:25, 3 November 2011 (UTC)[reply]

That might be arguable - but the fair-use claim here is strong enough to stand by itself. Please re-read my previous post and reply directly to the points I made there - that is the explanation you need for why 'fair use' (at the very least) applies here. But just in case you still don't get it, let's run down the WP:FAIRUSE bullet list:
  • No free equivalent. True. Obviously - because this is a standard test image, no alternative that differs from the original by even one bit could perform the task of describing this image. Obviously, if the image were to be used in other ways in other articles (eg as a representation of Lena Soderberg herself) then this might not be the case - and a separate fair-use claim would be needed for that use. I'm only talking about Lenna here.
  • Respect for commercial opportunities. The original image was sold as pornography, this usage is as a standard graphics test - nobody will come to Wikipedia to look at this cropped, blurry, miscolored image instead of buying a copy of Playboy magazine! (Aside from anything else, all of the 'naughty bits' had been cropped out before promulgating this as a standard test image!) Also, Playboy have publicly stated that this is simply not an issue for them. Nobody sells the cropped/lowres version - it's given away free everywhere it's used - so no possible commercial loss can come from displaying the image on the Lenna page at Wikipedia.
  • Minimal usage. Yes. It's only used in the article about the standard test image that's derived from the Playboy centerfold...if it appears elsewhere (eg in the Lena Soderburg article) then a separate fair-use claim might be needed. But that is not our concern here.
  • Minimal extent of use. Yes. This is a tiny crop out of a double-page 'centerfold' picture. It's scanned at about 75 dots per inch - incredibly low resolution compared to glossy-magazines which are typically printed at 1000 dots per inch or more. Displaying it here cropped still more or at yet lower resolution would utterly destroy it's value as a description of this standard graphical test because doing so would alter things like the sharpness of edges and the color of parts of the image that are crucial to understanding this in the context of being a standard test image. So, yes, this is the utterly minimal extent of usage. Once again, if the image were to be used in other ways in other articles (eg as a representation of Lena Soderberg herself) then this might not be the case - and a separate fair-use claim would be needed.
  • Previous publication. Yes. The entire image processing community uses this image - it's been published thousands of times in prestigious places such as the SigGraph proceedings and in scientific papers and magazines about computer graphics - you can see the image on myriad web sites and other Wiki's. It's also been displayed in several TV shows such as the old BBC "Horizon" program about computer graphics from back in the 1980's.
  • Media-specific policy. Yes. We meet Wikipedia:Image use policy in every regard.
  • One-article minimum. Yes, it's used in Lenna.
  • Contextual significance. Yes, it's impossible to discuss this important graphical test image without showing it in the article. To discuss the issues of red-tint in the skin tones without showing this exact image would be impossible...or at least incredibly ridiculous.
  • Restrictions on location. Yes, it's only used in the body of the article, not in a gallery or any other impermissible place.
  • Image description page. Yes, we have an appropriate description on the image description page.
Hence we clearly meet every single bullet with respect to the copyright of the original Playboy centerfold. Hundreds of other books, magazines, posters, fliers and other publications have used this image under fair use without any problems whatever - and Playboy have even publicly endorsed this usage. Wikipedia certainly qualifies!
The remaining issue is whether the act of scanning and cropping the image created an additional copyright for Alexander Sawchuk or the University of Southern California Signal and Image Processing Institute (SIPI). It has long been Wikipedia policy that we do not consider scanning and cropping to add artistic value to the image - and that is supported by US law. So no ADDITIONAL copyright beyond that owned by Playboy needs to be entertained here. I would cite National Portrait Gallery and Wikimedia Foundation copyright dispute where it was claimed that scanning/photographing and cropping those images was an artistic act that created an additional copyright for the gallery. Bridgeman Art Library v. Corel Corp. also indicates that no artistic input comes from scanning and cropping. Hence Wikipedia's legal folks decided that this was a non-issue and we have used those images throughout the encyclopedia. So scanning and cropping of the Playboy image by SIPI didn't add a new copyright (at least as far as Wikipedia is concerned). Thus we don't need to defend our use of the SIPI scanned/cropped image so long as we may use fair use to defend our use of a part of the original, copyrighted, Playboy image.
So this image meets 100% of the criteria for fair use versus the only copyright we care about. Deleting it is incorrect.
QED.
SteveBaker (talk) 13:10, 3 November 2011 (UTC)[reply]
Per sense of this discussion restored 512x512px version of this test image unless and until some firm consensus is achieved to reduce it. Centpacrr (talk) 16:12, 3 November 2011 (UTC)[reply]
The test image is more than just a picture, it is a string of bits which are used to compare the string of bits generated by various compression algorithms. For its main purpose then, a reduced image of this reduced (but special) test image is a totally inadequate illustration of the test image. Also, I think it's not a question of why WP needs the 512x512 image, its a question of why does it not? PAR (talk) 18:56, 3 November 2011 (UTC)[reply]
Agreed, and this would apply for any placement of the image at Commons. However that argument has to be argued on those grounds.
Here on Wikipedia, our only justification for images is to illustrate articles (SchuminWeb, for one, deletes any that he thinks don't). To illustrate an article, we don't need bitwise comparability, nor 512 resolution. Andy Dingley (talk) 20:02, 3 November 2011 (UTC)[reply]
No - that's not true. Wikipedia does not use images purely as decorative touches. They may also serve as an actual part of the description of the thing we're discussing. That happens (for example) in the case of diagrams. In an article about (say) a car, we use the photographs to describe the shape of the bodywork - because doing that adequately in words is impossible. "The front kinda curves down a bit, then bends more sharply at the front"...doesn't help the reader at all. The image in that case is more than just a pretty picture to liven up the article...it's a description (albeit in pixels rather than in words) of the subject of the article.
In this case, the overall appearance of an image of a porn star is utterly unnecessary as an illustration for decorative purposes. The people who use the Lenna image don't give a damn that it's a picture of some porn star. They care about the nature of the edges in the scene, the degree of noise in the smooth areas, the gradation of color, the purple hue of the feather, the (pictorially incorrect) red-tint of the skin tones and so forth. Reducing the size of the image by even one pixel (or storing it as a JPEG or anything of that nature) destroys the ability of this image to carry an explanation of the standard test image because it would alter those characteristics and imply something about the standard test image that just isn't true.
Imagine using photoshop to change the shape of the hood of a photo in Corvette Stingray. Doing that would destroy the ability of that photograph to describe the shape of the car and make the image be a pictorial lie...an untruth that cannot be supported by WP:RS.
In this case, describing those subtle features of the image in text would be beyond ridiculous ("The signal-to-noise ratio of the region between (123,34) and (150,72) is around 1:2.3"...crazy!). Using a reduced resolution image would introduce new artifacts into the image that would constitute incorrect information about the "Lenna" image. So I'd have to demand the complete removal of a reduced version of the Lenna image from the article on grounds that it is introducing incorrect information that cannot be backed up by reliable sources.
Furthermore - there is no need to further reduce the resolution. The copyrighted image is NOT the 512x512 crop that we're discussing here - the copyright only exists for a double-page centerfold from Playboy magazine. We are taking a very reduced and tightly cropped piece of that image and using it under fair use provisions. We are required to use a reduced resolution - and we're already doing that at 512x512 - and we're required to use the minimum amount necessary to support our description - which happens to be 512x512.
Those of you who are not graphics/image analysis folks need to understand that standard test images are a very special case for Wikipedia...these cannot be hacked around for the sake of making a political point. This image is already a heavily cropped and reduced-resolution version of the magazine centerfold from which it was scanned - so we are already well within the requirements of fair use...there is no need to reduce the resolution further in order to be on firm legal grounds - and reducing it by even one pixel (or enlarging it by one pixel for that matter) would totally destroy all value as a description of the standard test image called "Lenna". SteveBaker (talk) 13:29, 4 November 2011 (UTC)[reply]
"Illustrating an article" does not imply "purely decorative".
Illustrating an article in this sense means illustrating an article. An article that is read by humans. The human reader needs to see the image and recognise the lady portrayed, but there is no need for this image to actually be the test image itself. Our consumption of this article is just by reading it, not by feeding it to a compression algorithm. Yes, the article should link to an exact copy of the test image, but that's a linkage question, not one for embedding on the page. We achieve the clear need for making the real file available by linking, not by embedding.
There is some reasonable claim that printing the article requires something bigger than 300px. However that applies to all free-use images in all articles, and as yet I don't believe that's an argument that makes it past the image deleters.
There is also a view (not supported by policy, but certainly implemented by some of the admins in this thread) that WP shouldn't host any images that are large, aren't used in articles immediately or are freely licensed: it's either Commons, grudging acceptance under fair-use, or gone for good. Andy Dingley (talk) 13:56, 4 November 2011 (UTC)[reply]
No, the human reader does NOT need to see the image and recognise the lady portrayed. That's the entire problem here. We don't care that this image happens to be a photograph of a woman. We care that the original scanning of the image introduce color shifts and created noise and softened some of the original hard edges. This cannot be described in words - it can only be depicted in the image itself. This is a DESCRIPTION of the test pattern - that merely happens to be the test pattern itself.
To a practitioner in the art of computer graphics - this is an incredibly critical matter. Consider the case of the article Lorem ipsum which is about a standard latin text used in the printing industry. In that article we quote the entire text of lorem ipsum which starts: "Lorem ipsum dolor sit amet, consectetur adipisicing elit, sed do..." and so forth. Would it serve the purposes of that article equally well to write "Lore...ipsum color...yadda..yadda...adipicising editing, god do..."? No! That would be awful! Outrageous! Unsupportable by WP:RS - a downright lie! Please understand that for someone from the computer graphics business, reducing the resolution of the Lenna image is precisely the same thing - every bit as bad as the example I give above. SteveBaker (talk) 14:33, 4 November 2011 (UTC)[reply]

If you're seriously arguing that the image is in fact a bitstream and that only unaltered reproduction of that bitstream adequately conveys the required detail to our readers, then the simple argument would be that unaltered reproduction of the bitstream is not permitted by our guidelines on fair use and thus we can't use the image at all. I would argue that there is no ground for interpretation on that point at all. Chris Cunningham (user:thumperward) - talk 14:53, 4 November 2011 (UTC)[reply]

@Chris: I've explained this several times - but one more time won't hurt: You are confusing what it is that is copyrighted here - and that is the crux of this matter. The copyrighted work in question is a double-page Playboy centerfold containing a full-length photograph of a naked woman wearing a hat. That image has been scanned at very low resolution and cropped down to just her head and shoulders to form the 512x512 test image. The unaltered reproduction of the original centerfold would (of course) be inadmissable under fair use. However, displaying a cropped/de-rezzed version of the centerfold is not a problem for fair use against that copyright. That is why this image has been used tens of thousands of times in mainstream books, magazines and professional journals under the provisions of "fair use".
Now, there remains the question of whether the cropped/de-rezzed 512x512 image bears it's own copyright as a derived work. Wikipedia's rules quite clearly say "No". Merely scanning/photographing and cropping/aligning/de-rezzing an original work does not create a new copyright because no artistic/creative input was required. This was argued at great length and agreed (with advice and complete agreement from the Wikimedia foundations' legal team) in the matter of National Portrait Gallery and Wikimedia Foundation copyright dispute. So the only copyright we have to care about is the original Playboy copyright - and for that, fair use rules apply perfectly with a 512x512 cropped/de-rezzed image. Hence we have no need to drop the resolution yet further in order to obey the law and Wikipedias' fair use guidelines. SteveBaker (talk) 15:40, 8 November 2011 (UTC)[reply]

Size of non-free image

[edit]

There has been a discussion at File talk:Lenna.png regarding the size of File:Lenna.png, which is currently tagged as non-free, and whether a 512x512 image (original size) is acceptable under WP:NFCC#3b, or whether a 300x300 version of the image (the largest size that the image is rendered in articles) is the largest size that Wikipedia should retain. SchuminWeb (Talk) 20:07, 3 November 2011 (UTC)[reply]

For anyone reading User_talk:SchuminWeb#Lenna.png as part of their decision process, please note that certain comments by User:Schuminweb were removed before it was saved as an archive. Please check the history page to recover the full discussion. My apologies, the modifications are only stylistic. PAR (talk) 21:29, 3 November 2011 (UTC)[reply]
100 dpi (512 x 512) The image is already a low-resolution crop of the original and the copyright holder has stated they are effectively okay with the image being distributed at this particular resolution. I don't see any pressing (legal) need to further reduce the image to the somewhat arbitrarily chosen resolution of 300 x 300. —Ruud 23:53, 3 November 2011 (UTC)[reply]
The reduction is to comply with WP:NFCC#3b, which requires low resolution in our non-free images. In this case, the largest that the image will render in articles is as a thumbnail, and the maximum that a thumbnail will render is 300px across. Thus the reduction to that size. SchuminWeb (Talk) 01:50, 4 November 2011 (UTC)[reply]
Nothing in WP:NFCC#3b requires the use of 300x300 or less, only that the
To reproduce this image in print at 1 by 1 inch (that's a very significant reduction from the original) you would need a source image of at least 600 x 600 and then it's probably still going to look unsharp. I can't find any mention of the particular dimension of 300 x 300 in WP:NFCC and I doubt it is based on solid legal arguments. —Ruud 08:28, 4 November 2011 (UTC)[reply]
Scanning through the archives it seems that no one really has a definitive answer on this. The 300 x 300 apparently is a standard for album covers. In general the reduction in resolution should be seen in comparison to the original (that is, the centerfold, not the test image, which as you correctly point out is a derivative work) and can be as high as necessary to retain important details such as text that need to be legible. —Ruud 08:47, 4 November 2011 (UTC)[reply]
(512x512) - Please see SteveBaker's detailed explanation in the above section. A few points that need to be stressed:
  • To the casual observer, this is a pretty picture. To the imaging community it is much more: it is a 512x512 pixel test image - a particular string of bytes which are generally used to compare to a string of bytes resulting from the application of various image compression algorithms to the original test image. It is ubiquitous in the image processing literature, being used in many peer-reviewed journal articles. Any modification of the image destroys its use as a test image for the community, leaving only a pretty picture. The Lenna article is not an article about a pretty picture, it is an article about a famous test image.
  • The test image is not copyrighted, it is a low-quality, low-resolution scan of a copyrighted image. It should not be subject to the rules which apply to a copyrighted image, but rather, at most, to the rules that apply to a scan of a copyrighted image. In other words, the idea that a copyrighted image must be in reduced form to be displayed on Wikipedia is not applicable. It is already an extremely reduced scan of a copyrighted image, with no commercial value. PAR (talk) 00:50, 4 November 2011 (UTC)[reply]
Please provide written documentation of this permission from Playboy that we have the right to distribute this image freely. The idea that Playboy is not going to legally pursue such illegal uses of the image is not the same as a permission to distribute the image freely. In any case, it is a derivative work of copyrighted material, and a new copyright is not created, i.e. the derivative work is still the intellectual property of Playboy. SchuminWeb (Talk) 01:50, 4 November 2011 (UTC)[reply]
Agreed, and Playboy has not given any such permission, as you well know. This does not prohibit the use of a low resolution scan on Wikipedia. Now will you please provide documentation which permits the use of a 300x300 image yet prohibits the use of a 512x512 image? The oft-quoted WP:NFCC#3b does not.PAR (talk) 17:34, 4 November 2011 (UTC)[reply]
As I'm quite sure you are well aware, we don't need any permission to use this image under the provisions of fair use. The fact that Playboy have said that they aren't concerned about this kind of use of this particular image should be regarded merely as corroborative evidence. My point-by-point addressing of all of the bullets in WP:FAIRUSE should suffice to allow us to continue to use this reduced-resolution/cropped scan of a Playboy centerfold in this Wikipedia article. If you could tell us where you think I made an error in that point-by-point addresssing - we'd all be able to focus this discussion more tightly. SteveBaker (talk) 14:15, 4 November 2011 (UTC)[reply]
  • I don't really have an opinion on this one way or the other. On the one hand, it doesn't seem particularly relevant that the actual test image is 512px; the point of having this is for people to learn what it looks like, not to provide an exact copy for image compression testing purposes. On the other, the idea that we have to reduce fair-use images to some arbitrary dimensions never made a lot of sense to me; I totally agree that we should eliminate fair-use images whenever possible, but in the cases where we do have to have them, it doesn't make them more "fair" to reduce them. I do wish PAR would quit saying the image is not copyrighted, as that is simply not true.—Chowbok 03:08, 4 November 2011 (UTC)[reply]
Ok then, we both agree on the situation, but have been using different terminology to describe it. Let me restate, using some of User:Schuminweb's terminology: It is a derivative work of copyrighted material, but a new copyright is not created. The derivative work is still, however, the intellectual property of Playboy and covered by that copyright with the usual exceptions which allow for its use under certain conditions. PAR (talk) 17:34, 4 November 2011 (UTC)[reply]
    • The reason that 300x300 was chosen was because 300px across is the largest size that MediaWiki will render a thumbnail image. The image was originally sized down to 220x220 (default thumbnail size), until Amalthea mentioned that the largest size that users can configure thumbnail images for is 300px across. This was done to comply with WP:NFCC#3b. SchuminWeb (Talk) 04:36, 4 November 2011 (UTC)[reply]
WP:NFCC#3b does NOT require the use of a 300x300 or less image. It requires the use of a reduced image. Being very generous and assuming the copyrighted image is the equivalent of 600 dpi, the reduced, cropped test image is less than three tenths of one percent of the original. WP:NFCC#3b is not violated by a 512x512 scan.PAR (talk) 17:34, 4 November 2011 (UTC)[reply]
    • "the point of having this is for people to learn what it looks like" - but the problem is that "learning what it looks like" isn't a matter of seeing a photograph of a porn star - it's about seeing the subtle issues of noise, edge sharpness, scanning artifacts, color incorrectness, etc - and those are destroyed by resizing it. Even if we were legally allowed to - we would have zero interest in putting the entire centerfold at 1600x2000 resolution into the article...that would be a matter of mere curiosity with zero bearing on the matter of this standard test image. SteveBaker (talk) 14:42, 4 November 2011 (UTC)[reply]
  • (512x5122) per my previous posts, which I won't repeat here. SteveBaker (talk) 14:38, 4 November 2011 (UTC)[reply]

Clearly some people here need some instruction on the nature of image resizing.

The following image has (on the left of the red line) some noise - just random dots. The image on the right is what happens if you take that and resize it by one pixel using GIMP on 'high quality' resize settings:

Notice that some very odd things have happened. The center of the image on the right seems blurrier than the edges. The middles of the sides of the image seem streaky. The corners are completely OK. In an ordinary photographic depiction of someone's face (say) this is not important. Our eyes and brains can still see that it's a face and generally cope with the artifacts that resizing produced and recognize the person in question. So reducing the resolution of a typical fair-use image makes perfect sense.

However, if you are using an image to describe a test pattern of 1's and 0's that are used to test out software that does things like compressing images or enhancing them or some other numerical operation - then the presence of some noise in the image is deeply relevant to how that software operates. Signal to noise ratio (for example) is critical here. These standard test images (of which Lenna is only the most popular of dozens) provide an agreed-upon metric upon which practitioners in the field may judge the relative quality of various software algorithms.

These people do not give a damn that this is a photograph of a porn star - or a space ship or someone's cat. They care that the pattern of noise on Lenna Soderbergs' bare shoulders is reproduced more or less well - that there are or are not contouring artifacts, edge blur, color shifts. It is important that this pattern of noise is present in the image and no new noise introduced - or blurriness or anything else changed compared to the standard.

When we describe this image in Wikipedia, it is imperative that we don't lie to our readers and suggest that there is either more or less noise right there...that would not be a fact that could be supported by WP:RS. As you can see from my example above, resizing the image introduces blurriness in some parts of the image and not others - it does all sorts of very VERY complicated things to the nature of how you would approach compressing or enhancing it. If we resize this image by so much as one pixel, we utterly destroy it's ability to describe the standard test pattern and we go from a useful description of the standard test to a mere photograph of a porn star. I repeat - the image we're using is ALREADY a dramatically reduced resolution and cropped image of the copyrighted Playboy centerfold. We quite clearly pass the bar required by fair use. Reducing the resolution by so much as a single pixel destroys the ability of this image to describe the test pattern and (IMHO) it could be considered a violation of fair use to include it into the article because to do so would be to add it for purely decorative purposes - since we'd have destroyed any validity as supporting information to the descriptive text.

SteveBaker (talk) 13:58, 4 November 2011 (UTC)[reply]

What is lost on you and the other no-resize people is that we're not rendering it at full resolution in articles in the first place. So it's funny that you argue that every pixel is necessary when we're not even using it at such a resolution. SchuminWeb (Talk) 14:02, 4 November 2011 (UTC)[reply]
Its not lost, its just not relevant. If the article rendition were the only version availiable to a Wikipedia user, then you would be correct. But its not, the full resolution test image is availiable by clicking on the image. PAR (talk) 17:34, 4 November 2011 (UTC)[reply]
Also, upon further consideration, I am dropping this RFC, since nothing here will stop people like Centpacrr from just restoring the image from elsewhere in violation of policies. See you at FFD. SchuminWeb (Talk) 14:03, 4 November 2011 (UTC)[reply]
Please be sure to inform me when your proposed FFD is entered - I will certainly want to be a part of that discussion. SteveBaker (talk) 14:37, 4 November 2011 (UTC)[reply]

As I said in the section above, if the argument is that only unaltered reproduction is permissible because this is really a bitstream which happens to look like a picture of a pretty lady when interpreted in a certain way, then our fair use obligations cannot possible be met (as an unaltered reproduction is not permitted under our fair use guidelines). Chris Cunningham (user:thumperward) - talk 14:57, 4 November 2011 (UTC)[reply]

As one of the very points of the article is that this specific standard test image must be exactly 512 x 512 px to illustrate what it does and how it is used, it now appears in the article at that size and in its own section entitled "The 512 x 512 pixel digital test image" to provide that context. To accurately illustrate what the article is about, its display at 512x512px is the smallest (and only) size at which this image that can be displayed to achieve that end. Centpacrr (talk) 15:54, 4 November 2011 (UTC)[reply]
That's precisely the point. We're talking about a digital media file which happens to take the form of a 512x512 bitmap. We are not "exerpting" it: we are copying it bit-for-bit. So the argument is really about whether that original 512x512 bitmap falls under fair use. Anyway, let's take this to Wikipedia:Files for deletion/2011 November 4#File:Lenna.png rather than discussing it in two places. Chris Cunningham (user:thumperward) - talk 16:17, 4 November 2011 (UTC)[reply]
What YOU are missing is that the 512x512 image is not copyrighted - it's the original Playboy centerfold that is copyrighted. The act of scanning and cropping is not considered to add anything creative to the image - so the 512x512 image is not copyrighted - EXCEPT in that it bears the Playboy centerfold copyright. Since the latter is defensible under fair use, we have no further issues. SteveBaker (talk) 19:06, 5 November 2011 (UTC)[reply]

"differs greatly in color balance and contrast"

[edit]

Isn't the aspect ratio ridiculously off as well? --BjKa (talk) 18:30, 28 January 2021 (UTC)[reply]