Wikipedia talk:Manual of Style/Images

WikiWorld cartoons

WikiWorld cartoon currently illustrating the Fake Shemp article

Was there ever a consensus for or against using Wikipedia:WikiWorld cartoons in article space?

It was a comic strip that ran from 2006 to 2008 in the Wikipedia:Wikipedia Signpost, where the artist quoted parts of a Wikipedia article as it was at the time, adding cartoons. I don't know if they were ever intended to be used in mainspace.

Obviously, the text that they're quoting may not be present in the article two decades later. I took the Hammerspace cartoon out of that article last year, because the article text no longer speculated about "the nature of Hammerspace", or mentioned trees and tent poles. (The actual drawing of the monkey with the hammer is perfectly illustrative in isolation, though, and I kept it in the article as a cropped version.)

But even when the text of these cartoons is up to date, a full panel cartoon doesn't really seem like the type of image used for similar purposes in high-quality reference works, and the text is (or should be) redundant to the lead section. Belbury (talk) 16:26, 14 August 2025 (UTC)

• I see this ... thing has been removed from the Fake Shemp article, and rightly so. This is the sort of thing that belongs in children's books, or the For Dummies series. EEng 21:03, 26 March 2026 (UTC)

  With no feedback in favour of mainspace comics, I've added The cartoons were designed for use in the Signpost newsletter, and the full comics should not be used in articles. to the Wikiworld page, and removed a few which were present. In the Fake Shemp and Thinking about the immortality of the crab cases, I don't think it's worth cropping the text to keep just a cartoon character. Belbury (talk) 10:11, 27 March 2026 (UTC)

Expand AI upscaling definition of "historical images"

The MOS is currently that:

Original historical images should always be used in place of AI upscaled versions.

This occasionally gets pushback on the grounds that somebody doesn't consider a 2024 photo of a celebrity to be "historical" per se. Would there be any objection to expanding it to:

Original historical images, and original photographs of people, should always be used in place of AI upscaled versions.

Belbury (talk) 16:57, 19 August 2025 (UTC)

I would support that change. I'd prefer just dropping "historical" from the status quo. Firefangledfeathers (talk / contribs) 17:00, 19 August 2025 (UTC)

With no objections to this, I've updated the guideline to include and original photographs of people. (I'm less sure about dropping "historical" to expand the scope to all images, which would include even simple diagrams.) Belbury (talk) 10:22, 13 March 2026 (UTC)

Discussion at Wikipedia:Neutral point of view/Noticeboard § Discussion at far-left politics

 You are invited to join the discussion at Wikipedia:Neutral point of view/Noticeboard § Discussion at far-left politics. —Alalch E. 10:24, 16 October 2025 (UTC)

Mobius (TV series) upright scaling factor

Hi, The upright factor is getting revert for the article image. The image width is not at default table width, anc visibility is negligible. Enable to adjust from my end, keeping policy in mind. Support required to adjust the picture size, also understand the process of upright properly. SakuraSmart (talk) 03:20, 3 December 2025 (UTC)

You shouldn't be using more than upright 1.2 in the inforbox or lead, see MOS:IMAGESZ. -- LCU ActivelyDisinterested «@» °∆t° 15:02, 3 December 2025 (UTC)

Thanks for clarifying SakuraSmart (talk) 08:11, 4 December 2025 (UTC)

Appropriateness of arbitrary cast images in an article

A lot of articles have images of cast members or other production members, of varying degrees of relevance to the article, often from a time period very far removed from the article's subject. These images are often accompanied by a vaguely relevant somewhat arbitrary caption, sometimes duplicating pieces of the article. My personal feeling is that if it doesn't inform the reader about something relating to the article in any way and isn't even taken near the same time, the image is out of place. Is there any consensus regarding this? Tduk (talk) 23:17, 15 December 2025 (UTC)

Image sizes & zoom

Howdy. At 175% zoom, an image seems to fit in an article, quite well. But at 200% zoom (for some like I, with weaker eyesight), the image is too large in an article. How can this be fixed? GoodDay (talk) 23:22, 28 January 2026 (UTC)

How an image fits in the article will depend on how you're accessing the article and your personal setup. But zoom is usually something done by the browser, could you clarify what you mean? -- LCU ActivelyDisinterested «@» °∆t° 02:25, 29 January 2026 (UTC)

You can change your default image size in your preferences. Nikkimaria (talk) 02:27, 29 January 2026 (UTC)

Related:

• User talk:Robminchin#Image sizes - 23:20, 28 January 2026 (UTC)
• Wikipedia talk:WikiProject Olympics#c-GoodDay-20260128232600-RfC on the placement of 2026 Winter Olympics and similar templates - 23:26, 28 January 2026 (UTC)
• Wikipedia:Village pump (technical)#200% Zoom & sidebar appearances - 00:48, 29 January 2026 (UTC)

There seem to be others. --Redrose64 🌹 (talk) 00:08, 30 January 2026 (UTC)

Color images

I've been looking at MOS:COLORIZED:

"An image that was originally published as a monochrome image, such as black-and-white or sepia photographs, should not usually be colorized. A Wikipedia editor who adds color to such an image is likely engaging in original research. However, images that were colorized by their creator, an expert, or a reliable source may sometimes be acceptable. Colorized images may also be used where the colorization is an encyclopedic aspect of the image and relevant to the article."

I think firstly that this is about photographs (rather than line art or diagrams), so that second word, image, should probably be changed to photographs. Nobody cares if you change the color on a pie chart, because the color isn't important there. Everyone cares if you change the colors on a BLP's photo.

Second, I think the motivation for colorizing is often a personal/aesthetic belief that monochrome photos are inherently "worse". What do you think about adding something like this?

"Although color images are often readily available, Wikipedia does not have a preference for color photographs over black-and-white photographs."

WhatamIdoing (talk) 20:25, 17 February 2026 (UTC)

I've actually been asked to colorize a number of black and white images, usually images which were used for illustrative purposes, such as File:Catherine Elizabeth Middleton (colorized).jpg. I even wrote a simple guide to colorizing. As such, I've engaged a bit of thought about the subject. Thought that I would like to lay out before directly responding.

First off, I think we should all be clear that, even if you have immaculate references, colorizing a black and white photograph is absolutely WP:OR. Combining those references with the original image is as straightforward a case of WP:SYNTH as there is. Black and white photos have different curves and treat colors and shading differently than color photographs. The end result is never going to be exactly what you would have gotten had you just taken the photo with color film (or with the full-color mode of your digital camera).

With that in mind, colorized photos aren't useless here. They can be quite good for illustrating things, such as providing a possibly-better-recognized likeness to go with a name, which is how the file I linked above is typically used on en.wp. Just as we allow editors to engage in OR for talk-page discussions and publish plenty of OR in wikispace and allow certain specific kinds of OR (WP:SUMMARY and WP:CALC and MOS:LEAD all allow for certain, strictly-controlled degrees of it) in article space, there's nothing wrong with improving an image to illustrate the subject better.

Now, getting down to the actual question posed here, I agree that changing the word 'images' to 'photographs' is best. I also prefer wording that is unambiguous: the world 'likely' should be removed from that, so that it clearly states that colorization is OR. If others are amenable, a note that sums up my third paragraph above (that colorized images still have a place here) could be added, though I don't believe that's strictly necessary. Those of us who do this kind of work are all old hands at this, and we don't need a wikispace link to give us permission to do what we've been doing (mainly at the request of others) for years. ᛗᛁᛟᛚᚾᛁᚱPants Tell me all about it. 20:58, 17 February 2026 (UTC)

If it's OR, then there's no place for it here. OR is banned in the mainspace, full stop. There is no "It's OR, but OR is acceptable under the following conditions" option. WhatamIdoing (talk) 21:11, 17 February 2026 (UTC)

WP:5P5 is literally a core tenet of this project. Try again. ᛗᛁᛟᛚᚾᛁᚱPants Tell me all about it. 22:17, 17 February 2026 (UTC)

There are limits to the "no firm rules" principle. We have no firm rules, but we also have no copyvios, no fair-use photos of BLPs, and – relevantly – no original research. That policy begins with the words "Wikipedia articles must not contain original research" (emphasis in the original). It does not then go on to say "Oh, but it's okay to have original research for colorizing photos".

Therefore, only one of the following statements can be true:

• Colorizing photos is not always OR (within the defined meaning of the policy), and therefore they can sometimes be used.
• Colorizing photos is always OR (within the defined meaning of the policy), and therefore they can never be used.

Pick one – I don't much care which one – but it's got to be one of those. WhatamIdoing (talk) 00:43, 18 February 2026 (UTC)

Except images (usually photos) created by volunteers, which are always OR, and make up a noticeable fraction of our images. Yes, that is a hammer in his pants, and he is also glad to see me.--GRuban (talk) 02:13, 18 February 2026 (UTC)

Most images created by volunteers are defined as being non-OR in the WP:OI section of WP:OR. Therefore they are not OR. WhatamIdoing (talk) 02:56, 18 February 2026 (UTC)

The fact that the section exists should be enough for you to figure out that it exists because original images are an exception to the policy. ᛗᛁᛟᛚᚾᛁᚱPants Tell me all about it. 03:02, 18 February 2026 (UTC)

I'm always happy to see you. ᛗᛁᛟᛚᚾᛁᚱPants Tell me all about it. 03:03, 18 February 2026 (UTC)

Quoting a policy which one of the five pillars explicitly says may be ignored in order to improve the project is not a rational counter-argument to me pointing out that one of the five pillars is that we ignore policies when doing so allows us to improve the project. ᛗᛁᛟᛚᚾᛁᚱPants Tell me all about it. 02:57, 18 February 2026 (UTC)

WP:5P isn't a policy. WhatamIdoing (talk) 03:08, 18 February 2026 (UTC)

That is the single most ridiculous response I could have possibly imagined. You're right, the five pillars aren't policy. They're the "fundamental principles" of the project. They are not only more important than our policies, the policies exist to facilitate them. ᛗᛁᛟᛚᚾᛁᚱPants Tell me all about it. 03:51, 18 February 2026 (UTC)

Wikipedia:Five pillars was written as an expansion of Wikipedia:Trifecta, several years after the core content policies were written. Whether they're important is a matter of personal opinion. WhatamIdoing (talk) 06:32, 18 February 2026 (UTC)

I think it's important to be clear what "OR" in WP:OR means: "material ... for which no reliable, published source exists". This is particularly relevant to the construction of diagrams that illustrate the text of an article. The diagram itself is "original" but it's not original "research" if it's clearly supported by reliable sources, whether text or other diagrams. In the same way it's possible that colouring a monochrome photograph could be "original" but not "research". It's not relevant that it's not exactly the same as a colour photograph taken at the same time would be. What is relevant is that there are reliable sources that support the added colouring. Peter coxhead (talk) 09:20, 18 February 2026 (UTC)

I think WhatamIdoing's "it's got to be one of those" principle is fundamentally incorrect. Wikipedia's rules only exist to further its purpose of being a free encyclopedia, not for themselves, they can and do bend when necessary. Arguing about principles in the abstract is not useful, we need to see specific cases. In some cases specific colorized images will be fine. In other cases specific colorized images will be wrong. In other words ... all generalizations are wrong. --GRuban (talk) 14:08, 18 February 2026 (UTC)

Pardon the wall of text, I'm making a point with several examples and some nuance.

For the colors in colorized images or even the various elements in original illustrations (not all are charts), there are rarely reliable, published sources to support them. See the main image at grey alien, which I created for the purpose of being the main image there. There are exactly zero reliable sources establishing that, for example, gray aliens have collarbones or pectoral muscles. How could there be, when the subject is a class of fictional beings? Sure, there are nominally reliable sources discussing, for example, their large, black eyes or lack of a protruding nose or visible genitalia, but how big are the eyes? Are they fully black (as depicted), or a very dark gray color? Is the depicted shape the same kind of 'almond shape' which is described? Is there a flap of skin that might be obscuring genitals? Are there nostrils? Even when we have reliably-sourced answers to certain questions, there are deeper questions which don't.

As for the example I gave above, is there a reliable source for Middleton's skin tone? What about her hair color? What about the background?

Using other photos as a reference for the colors in that one leaves me at the whims of the lighting and post-processing of those photos. It's a textbook case of SYNTH. I had to do research in order to find answers to those questions. I could not find a source which laid out, for example, the exact shade of her skin or hair under various lighting conditions. I had to rely upon my own judgement and creativity. Despite that, the Middleton image was the main image on her page for years, and is still used in mainspace here, as well as on 22 other wikis. And there's no harm to the project from it. Indeed, it improved the project to have that image. There would be harm in deleting it.

This can apply to undoctored photographs as well, if I take a photo of a caterpillar and call it a Danaus plexippus (monarch butterfly) in the 3rd instar, is there anything but my word to verify that it's not actually a Danaus gilippus (queen butterfly) in the 4th instar? Sure, we almost certainly have a few butterfly experts here who could confirm that and lay out the evidence for it, but I could also write in mainspace that it is possible that a tipler cylinder of sufficient non-infinite length could, in fact, produce closed timelike curves and have a physics expert editor come along and mathematically confirm it, complete with proof. It would still be OR (unless either expert editor then published their results, of course).

This is why the fundamental requirement over at commons is 'encyclopedic value', and not 'established provenience' or 'described in reliable sources'. Because over there, where image use is front and foremost, editors understand that originality is not necessarily a bad thing. And here on en.wp, we accept that if an image passes the bar over at commons, it is acceptable for us to use it here, subject to certain unrelated rules, such as not putting irrelevant images in articles or overusing copyrighted images.

If editors like WhatamIdoing need to have another exception explicitly carved out for colorized photos or original images, then I suppose that's a discussion we can have. But as things stand, the editors making these OR types of images which are actually used by other editors in mainspace are already aware of the norms and practices here, and don't need such guidance. I'm surely not going to be asking WhatamIdoing for permission the next time I'm asked to colorize an image, for example, because I know that fulfilling the request will only help improve the encyclopedia, and that is (or should be, rather) the ultimate goal of all of us editors. ᛗᛁᛟᛚᚾᛁᚱPants Tell me all about it. 16:16, 18 February 2026 (UTC)

In re on en.wp, we accept that if an image passes the bar over at commons, it is acceptable for us to use it here: Nope. See, e.g., WP:AIIMAGES which exists because AI-generated images pass the bar over at Commons but are not acceptable for us to use. WhatamIdoing (talk) 18:53, 18 February 2026 (UTC)

See C:Commons:AI-generated media, which holds AI images there to much the same standard we do here. And of course, there are proscribed exceptions to our image use policy here. See AI images for multiple examples. ᛗᛁᛟᛚᚾᛁᚱPants Tell me all about it. 20:29, 18 February 2026 (UTC)

I think that guideline holds AI images to a much lower standard than we do. It needs to be useful for any realistic educational purpose, which would include things like putting the image on a slide for a classroom presentation or in a social media post intended to explain something. That suggests that certain non-educational content ("Make a picture showing me and my boyfriend in a romantic location") won't be acceptable, but that an AI image of (e.g.) any real-world object would be. WhatamIdoing (talk) 20:35, 18 February 2026 (UTC)

That suggests that certain non-educational content ("Make a picture showing me and my boyfriend in a romantic location") won't be acceptable, but that an AI image of (e.g.) any real-world object would be. I can't imagine that a lot of commons editors would find encyclopedic value in an AI generated image of a real-world object, as that image is likely to contain inaccuracies. I think that the language over there might be looser than we use here, but in effect, they are fundamentally the same policy. ᛗᛁᛟᛚᚾᛁᚱPants Tell me all about it. 21:20, 18 February 2026 (UTC)

A spot-check of FFDs mentioning the guideline by name indicates that most such FFDs close with the image being kept. WhatamIdoing (talk) 23:20, 18 February 2026 (UTC)

Umm, you might want to take a closer look at that list. Out of 28 unique images on the 19 standalone MfD pages 16 (57%) were deleted (some were for deletion of more than one image). That's more than half. The rest of the results are archive pages, which contain entries for the standalone pages (which is why they showed up in your search). Of the discussions on the archive pages, the only entries that weren't duplicates of the standalone MfDs were deleted. Indeed, many of those that were kept were kept because they were either C:COM:INUSE as examples of AI-generated imagery, or because they were not actually AI generated according to the uploader, at least not to an appreciable degree (and they were also INUSE). ᛗᛁᛟᛚᚾᛁᚱPants Tell me all about it. 23:55, 18 February 2026 (UTC)

But some of the deletions weren't necessarily for being AI-generated, and half of the deleted files were from just two users. And perhaps more importantly, there are thousands of labeled AI images on Commons, and it looks like only a couple dozen were deleted last year. I think that enwiki's standards rejects quite a lot more than 1% of AI images, don't you? WhatamIdoing (talk) 02:05, 19 February 2026 (UTC)

Ignoring the fact that you were flatly wrong about the numbers, I fail to see how, even if I were to concede that the policy on commons is looser than ours, that contributes anything to your argument thus far. I mean, it really reads like you're suggesting that me generating an AI image wouldn't be any kind of OR, and I fail to see how that makes any sense. ᛗᛁᛟᛚᚾᛁᚱPants Tell me all about it. 17:26, 20 February 2026 (UTC)

It would depend on the image contents, wouldn't it? I don't think that an image such as File:1 out of 100 baseball and 99 brown circles.png would be OR no matter how it was generated. WhatamIdoing (talk) 18:01, 20 February 2026 (UTC)

That would be an example of a chart or graph. It's a different beast than, for example, File:Greylien.png, and I agree that it's not OR.

That being said, I think we've gone on more than long enough, and I would appreciate it if you'd respond to the quoted sections in my latest response to Colin, below, where I spoke my thoughts on what should be done at this page. ᛗᛁᛟᛚᚾᛁᚱPants Tell me all about it. 18:29, 20 February 2026 (UTC)

That would be an example of "an image", as in the guideline says "An image that was originally published as a monochrome image..." WhatamIdoing (talk) 20:51, 20 February 2026 (UTC)

So, if you're not interested in discussing changes to this page, I don't really have anything else to say to you. I enjoyed the argument, but is has long past the point of WP:FORUM behavior, and I'm happy enough to be the bigger person and back out first. Go ahead and get the final word, just please don't ping me. ᛗᛁᛟᛚᚾᛁᚱPants Tell me all about it. 21:13, 20 February 2026 (UTC)

I am interested in improving this page – just busy elsewhere, and therefore not finished reading and thinking about Colin's comment.

For example, maybe that section shouldn't say "An image that was originally published as a monochrome image...". Maybe it should specify photographs and videos, because people mostly don't care about colorizing black-and-white charts or graphs. WhatamIdoing (talk) 00:44, 21 February 2026 (UTC)

On the general question of whether combining information from multiple sources is always SYNTH:

I disagree.

This is not a rare argument, but it's wrong. SYNTH has two necessary components. They are:

• Multiple sources (or, less commonly, distinct parts of the same source) being combined.

• A result (in the Wikipedia article) that does not fairly reflect the meaning of any source.

In text, this mistake leads to the "If it's not COPYVIO, then it's OR" class of claims. When writing an article, editors should combine the words from different sources to produce words that aren't exactly and precisely in any of the sources. They should take a source that says "Feeling sick is a common side effect in children", another that says "Nausea is a known adverse effect of this antibiotic in adults with a prevalence 15–24%", a reference work that says "Adverse effects that exceed more than 10% are considered common", and a dictionary that says "Nausea is a word for feeling sick" to produce a statement that "Nausea is a common side effect" or "Some people experience nausea on this drug" in the article about that antibiotic. This is "combining information from multiple sources" but it's not prohibited Wikipedia:Original research because this is a fair summary of what all the sources say about the nature and prevalence of this drug's side effects.

I think that some kinds of colorizing are similar: An artist or a printer might say that her hair has many distinct shades of brown with a variety different appearances from moment to moment, but the typical person is looking at this and saying "Her hair's brown, right?" If someone looks at your photo, and looks at a variety of photos of the same person from the same time period, and they think it looks like a fair representation of her, then that's not prohibited original research.

Editors can produce something that isn't taken directly from a single source. What editors shouldn't do is produce something that doesn't fairly reflect the meaning in any source. WhatamIdoing (talk) 19:42, 18 February 2026 (UTC)

I think that some kinds of colorizing are similar: An artist or a printer might say that her hair has many distinct shades of brown with a variety different appearances from moment to moment, but the typical person is looking at this and saying "Her hair's brown, right?" You are arguing that it's not SYNTH if the result is 'close enough' to something sourced, which is just carving out an exception from the OR policy.

As I said above, if you want to carve out an explicit exception here and add text to that effect to the page, that's fine, but I've never needed that, because I know the work is going to improve the project.

If someone looks at your photo, and looks at a variety of photos of the same person from the same time period, and they think it looks like a fair representation of her, then that's not prohibited original research. This is literally a post-hoc rationalization. By this logic, all OR is acceptable, because we might one day find out that it's accurate. Or because a reader or another editor might confirm it.

Editors can produce something that isn't taken directly from a single source. What editors shouldn't do is produce something that doesn't fairly reflect the meaning in any source. The photo of Middleton was deliberately taken in black and white. That is a necessary component of its meaning. A component that was changed via colorization. ᛗᛁᛟᛚᚾᛁᚱPants Tell me all about it. 20:21, 18 February 2026 (UTC)

I'm not carving out an exception from the OR policy; I'm saying that the OR policy itself says that if the result is close enough to one or more reliable sources, then it's not OR in the first place.

OR, as defined by that policy, is material that does not (sufficiently) match any reliable sources.

If it matches the reliable sources, then it's not OR as defined by that policy.

This isn't any sort of "exception". It's the main definition. WhatamIdoing (talk) 20:28, 18 February 2026 (UTC)

This isn't any sort of "exception". It's the main definition. A definition that includes an exception still has an exception in it, even if that exception is in the definition.

Look, I think there's a good deal of talking past each other going on here. I'm telling you, with absolute confidence, based on a great deal of both experience and prior thought, that colorizing an image or producing an illustration constitutes original research as the term is defined at the top of WP:OR. I'm not arguing that they represent a violation of the policy as a whole, but rather an exception which is not explicitly spelled out.

You seem to be telling me that the process doesn't involve doing anything that would place them under that definition, except I know for a fact that they do. I've done this work countless times over the years, including a number of times for this very project.

However, you also seem to be implying that they're widely understood not to be a violation of the policy as a whole, and on that point, we agree. The only difference is that I'm saying there's no language permitting this exact thing in the policy, and you're trying to (frankly) shoehorn it into other exceptions already carved out in the policy.

You stand almost no chance of convincing me that colorizing an image isn't WP:SYNTH, because I've colorized a number of images and I know for a fact that it does involve the exact same logic as synthesizing text, which I've also done, just not on WP.

However, I am offering to help work up a way of making this explicit in the policy now, which I don't believe it is. ᛗᛁᛟᛚᚾᛁᚱPants Tell me all about it. 21:18, 18 February 2026 (UTC)

The definition of OR doesn't include "an exception". The definition is: OR = anything that isn't in any reliable source. Literally any reliable source in the whole world. Quoting the first paragraph of the policy, if the material in question is in any reliable source(s) that has been published and still exist—somewhere in the world, in any language, whether or not it is reachable online—even if no source is currently named in the article, then it's not OR. There is no exception.

Perhaps this is the point of confusion: OR isn't about the individual editor's personal process for acquiring the information or creating the article content. If I go into a lab and run an experiment to determine that trinitrotoluene is an explosive chemical, and I write "TNT has explosive qualities" in the article TNT, then that's not prohibited by the Wikipedia:No original research policy. That is non-OR, because there are thousands of reliable sources that have been published and still exist—somewhere in the world, in any language, whether or not it is reachable online—even if no source is currently named in the article that say TNT is an explosive. The fact that I didn't personally consult a reliable source doesn't change the fact that this information can be found in a reliable source, and that Wikipedia's definition of OR is material—such as facts, allegations, and ideas—for which no reliable, published source exists and not material—such as facts, allegations, and ideas—for that admittedly is in all kinds of reliable, published sources, but she says she didn't directly consult one of them at the time she edited the article. WhatamIdoing (talk) 01:34, 21 February 2026 (UTC)

You may want to check the last part of that trifecta again. And I find myself wondering if you understand what a 'fundamental principle' is.

All rules exist to serve a purpose. Community and founder-endorsed statements of purpose (including both the five pillars and the trifecta) are the purposes all policy here exists to serve. It is a wildly irrational and self-defeating position to insist that a rule, which was created in order to serve a purpose, is more important than the purpose itself. ᛗᛁᛟᛚᚾᛁᚱPants Tell me all about it. 14:22, 18 February 2026 (UTC)

Sure, we could talk for hours about the inconsistent ways that we handle principles. For example, on the day that 5P included IAR, that policy's contents were different than they are now. Do you take an original intent POV? Or consider that the community has repeatedly rejected calls to label 5P as a policy. Does that rejection mean that 5P is more important than a policy? I'd call that POV something close to "wildly irrational", but maybe others believe that the rejection was caused by a feeling that changing from WP:NOTAG to {{policy}} would somehow downgrade 5P's status. Or let's discuss the job of a summary: Does the summary become more important than the summarized information? For example, if you're looking at a typical peer-reviewed journal article, and the abstract (akin to 5P) and the body of the article (akin to the policies and other pages linked in 5P) use slightly but meaningfully different language, will you rely on the summary in the abstract or on the full explanation in the article? WhatamIdoing (talk) 19:03, 18 February 2026 (UTC)

It doesn't really matter what exact language another page uses, nor does it really matter what language a page had at any time in the past. This is the encyclopedia of today, we are discussing.

The five pillars are currently, by community consensus, the fundamental principles upon which WP is based. We don't change them based on policy, we craft policy so as to achieve and maintain those principles. ᛗᛁᛟᛚᚾᛁᚱPants Tell me all about it. 21:31, 18 February 2026 (UTC)

For some time I've considered writing an essay to be titled "WP:IAR is not a get-out-of-jail-free card", or similar (see wikt:get out of jail free card#Noun). It would cover the fallacy that WP:IAR allows you to do absolutely anything you like. I could extend it to include 5P5 as well. --Redrose64 🌹 (talk) 22:16, 18 February 2026 (UTC)

The whole IAR argument came about because I pointed out to WhatamIdoing that WP's policies are not immutable, carved-in-stone rules that must be followed even when doing so hurts the project. It's a tangent to the actual discussion, which is us essentially using this page as a borderline (but not quite) WP:FORUM to argue over whether colorizing a black-and-white photo or creating a new illustration (not including charts and other such graphics that serve exclusively to convey reliably-sourced information) is OR or not. (I think it is, but that it's an exception to our policy. WhatamIdoing insists that either it isn't, or every example that's been done so far is a policy vio.)

I don't disagree with the notion of an essay explaining that IAR has its limits, however. The key component is the clause "if a rule prevents you from improving the project", and I've seen IAR cited in cases where the concern wasn't improving the project, but making WP speak WP:THETRUTH. But if you'd like to discuss that, we should probably take that over to Talk:IAR or one of our user talk pages. That is a discussion I'm happy to have, just not here. And I'd probably be happy to help write the essay, as well. ᛗᛁᛟᛚᚾᛁᚱPants Tell me all about it. 00:12, 19 February 2026 (UTC)

Of course I agree that policies and guidelines are not immutable or carved in stone; have I not made well over a thousand changes to them myself? I've made dozens of changes just to this one guideline over the years. If that's not enough to convince you that I believe policies and guidelines can be changed, then note that I wrote WP:PGBOLD c. 2010, which is the policy on making bold edits to policies and guidelines. Of course I believe in changing policies and guidelines.

What I don't believe is that the community holds "the principles" (which ones?) as the basis for Wikipedia, or that the community then writes the policies and guidelines to implement those sacred principles. That would be a nice, tidy world to live in, but it's not the one that we actually live in. What mostly happens is that policies and guidelines get written from the ground up, to solve concrete problems, and then a couple of us go around trying to make them not have obvious WP:PGCONFLICT problems and predictable unintended consequences, despite all kinds of editors wanting to special-case their pet peeve or using the wrong WP:UPPERCASE or insisting that terms like "original research" mean anything except the actual definition in the first paragraph of said policy, because WP:Nobody reads the directions. It has been so long since I saw a proposal that began from any of the principles that I don't remember when the last time was. WhatamIdoing (talk) 04:10, 19 February 2026 (UTC)

What I don't believe is that the community holds "the principles" (which ones?) as the basis for Wikipedia, or that the community then writes the policies and guidelines to implement those sacred principles. Well, I guess you could test that by rewriting WP:5P to say that they're "a shorthand that will kinda give you the feel for the general shape of our policies" instead of saying that they're the 'fundamental principles' of the project and see if you get reverted.

Also, since you enjoyed pointing out what these pages said when they were first written, you might be interested in what WP:5P looked like when it was first created. Although, to be fair, I stand by what I said earlier that it doesn't really matter what they said then, only what they say now. And the 'fundamental principles' language has been there since October of 2009. So it seems pretty inane to suggest that the community doesn't buy that.

or insisting that terms like "original research" mean anything except the actual definition in the first paragraph This legit got a laugh out of me. If this was intended to be a jab, you missed by a mile and made it clear you haven't been reading what I'm writing. ᛗᛁᛟᛚᚾᛁᚱPants Tell me all about it. 01:17, 20 February 2026 (UTC)

What makes you think that calling something a principle, or even "fundamental principles", means that they're more important than policies? Or even that they're in a hierarchical relationship of any type with the official policies? WhatamIdoing (talk) 04:20, 20 February 2026 (UTC)

The direct answer to your question is, of course, the definition of the words 'fundamental', 'principle' and 'policy'. ᛗᛁᛟᛚᚾᛁᚱPants Tell me all about it. 17:27, 20 February 2026 (UTC)

Sure, but that's just an old bureaucrat's trick: Always dispense with the difficult bit in the title. There's no agreement in the community that principles outrank policies. In fact, invoking 5P in ways that conflict with policies and guidelines is generally a losing argument (e.g., 5P says Wikipedia is a gazetteer, but we delete some articles about geographic features anyway). WhatamIdoing (talk) 01:38, 21 February 2026 (UTC)

I think the above discussion got a little side tracked. Citing IAR is a bit like citing NOTCENSORED: it's the weakest argument you can make. Sorry this is a bit long...

I see the COLORIZED section was added in 2023 after this discussion. I think it is incorrect that "A Wikipedia editor who adds color to such an image is likely engaging in original research." and the clause "images that were colorized by their creator, an expert, or a reliable source" is just word salad. All such images are colourised by their creator: the act of colourising creates an image that is different to it source. The requirement for an image to have come from an "expert" is just another example of Wikipedian prejudice against user created images. We don't apply "must have come from an expert" (or published reliable source) to any image on Wikipedia.

One could make an argument about OR if no colour images of a deceased subject exist, but that same would also be true of a user-generated painting of such a person for whom no colour images exist, so maybe that point needs made more general. And the likely error of such claimed OR may be not importantly greater than the error of one photograph vs another: it is quite easy with bad or deliberate lighting to radically alter someone's skin tone, make them look fatter or thinner, highlight their grey hairs or hide them, and so on.

Photographs are altered on Commons (lightened, contrast changed, colour enhanced) in ways Wikipedians never notice because we aren't alerted. It is also common to retouch old photos to fix damaged images and often the tone/colour is altered then too. We use such images in large numbers and frankly I think Wikipedians just don't know the degree to which the images we include in articles are different to the source images from museums or archives. Even the concept that "A Wikipedia editor who" reflects a misunderstanding about how the vast majority of our images are created or altered. It is mostly by Commoners, and they see themselves serving a wider educational purpose than the lead image of a Wikipedia article. Wikipedia P&G has no authority over what Commoners do to images that are referenced in our articles. All we can do as Wikipedians is choose one or another filename (or pop over to Commons to fork variants or whatever, subject to their rules).

As a photographer, I have a surprising degree of control over the image I get. And modern cameras (especially smartphones) have a surprising degree of computational generation going on in any image they make. Everything from the brightness to the colours to the contrast to the degree of detail or softness is controllable. Many modern photos are the result of combinations of photos aligned and merged. And many modern photos have some degree of AI adjustment. I feel the "user generated" or "hand touched" aspect gets a Dunning-Kruger reaction (we all made terrible paintings of mum and dad in primary school so we are experts in painting portraits, whereas very few of us ever worked in a portrait photography studio or in retouching such images).

I think also we repeatedly see editors making subjective judgements over which kind of images they like or think most suitable, but feel somehow they can't say "I don't like..." or "I prefer..." and instead feel the need to cite policy to make their argument. It is fine to not prefer colourised images just as it is fine to not prefer paintings of living people vs photographs. I don't think it is helpful to try to use P&G to make these arguments. Does an image serve a useful encyclopaedia purpose in the article and does it do it better than that other image. That's all that matters. Not how it was made or who (or what) made it.

I don't think the suggested sentence claiming Wikipedia is entirely neutral when it comes to favouring colour over b&w is true or helpful. I think overwhelmingly if editors were asked to choose between otherwise equivalent original-colour or b&w photos they would pick the colour one. Certainly for a lead image, where identification and illustration purposes are key. Later in an article, one could more readily admit a photo in artistic or b&w style or even the artificial monochromatic colours of concert lighting, say.

I don't think there are fundamental in-principle or policy forbidding colourising or indicating it is OR any more than 101 other image and photo creation/alteration activities. Is Wikipedia going to insist that I am not allowed to alter the brightness levels of my photograph in Lightroom, vs what the camera took, because that's "OR"? Colourising is one tool of many in generating an image for which the community can judge on talk pages over whether they serve an encyclopaedic purpose in an article.

We can judge whether any image (including user generated ones) is a fair likeness of the subject, and use sources of other images to make that judgement just as we judge whether the original words we write in our articles are a fair representation of the facts and claims that our sources make.

I think the current section of colourising images should be scrapped. Possibly something saying that Wikipedia has no automatic preference for a colourised version of a photo is sufficient. -- Colin°^Talk 10:24, 19 February 2026 (UTC)

Citing IAR is a bit like citing NOTCENSORED: it's the weakest argument you can make. Nobody's citing IAR to suggest we can do something here. I have said that, for a long, long time, editors have IAR-ed rules against adding novel information (read: Original Research) by colorizing images and creating original illustrations (other than infographics, which are sourced). In over a decade on this site, during which time I colorized multiple images, this is the first argument I've had about it, unless there was some minor back and forth so meaningless I've since forgotten it.

I do appreciate your points about other forms of image editing. I've had them in mind throughout this, but my comments have been wordy enough without making those points.

I think the current section of colourising images should be scrapped. Possibly something saying that Wikipedia has no automatic preference for a colourised version of a photo is sufficient. I would get behind this. Colorization is far from the most common request over at the graphics lab, but it's also far from uncommon, and it usually is done with little fanfare and the image quietly used with little to no objection. So a short discussion (even one which references a very long, prior discussion) that resulted in language being added to turn a relatively common practice into a policy vio seems ripe for removal. ᛗᛁᛟᛚᚾᛁᚱPants Tell me all about it. 01:30, 20 February 2026 (UTC)

I dunno. There was somebody using your account above talking about "The whole IAR argument", so I think somebody's citing IAR to claim something here.

@Colin, in suggesting that Wikipedia doesn't have a preference, I was thinking of official preferences, rather than the average view. For example, we have a lot of discussions over whether men who died in the middle of the 19th century should be represented with a high-quality, full-color oil painting from their younger years or with a grayscale daguerreotype from shortly before their deaths (e.g., Henry Clay). We have no official position on the choice; editors are free to prefer the color vs non-color options, just like they are free to prefer the painting vs photograph options. WhatamIdoing (talk) 04:27, 20 February 2026 (UTC)

What is the difference between "official preference" and "average view"? Surely the former is just whatever preference someone managed to insert into a P&G, which may or may not be aligned with the average view of participants in this collaborative project, but is supposed to be aligned if our consensus forming process actually works. Thus it is a fair assumption that something you propose adding to the P&G will be assumed to align with the communities views. I don't think it does. Sometimes if we perceive Wikipedia has no view on something, it is best then to just not include it in P&G rather than making that neutrality concrete.

MjolnirPants, can we please stop being distracted by IAR. You will convince nobody to alter a P&G page with an IAR argument no matter what your intention of using it. If you are saying actual practice on the project is not aligned with the documented P&G text, that is a different thing, and valid argument for removing or fixing said P&G.

I think MP, you screwed up with the claim "I think we should all be clear that, even if you have immaculate references, colorizing a black and white photograph is absolutely WP:OR. Combining those references with the original image is as straightforward a case of WP:SYNTH as there is." If you believe that, then WAID's response that OR is absolutely forbidden, is not going to change. But, well, I don't believe that, and it doesn't look like WAID does either. You seem to think it is OR because it can't possibly produce a result equivalent to what would have been generated if colour film was in the camera instead. That's not a definition of OR. I'm not sure the argument for including colourised images on Wikipedia will succeed if you continue with such a strong view that it is apparently a textbook example of OR.

Whenever Wikipedian's get OR wrong about images, I think it is good to remind ourselves that we are required by policy to write original text in articles: WP:OUROWNWORDS. We can use words that do not appear in the source text. Our writing can take a different tone than the original text (formal, encyclopaedic, vs informal perhaps, or journalistic). Our writing can make a point neutrally when the source made the point using loaded language. Our writing can be respectful of the subject when our source was mocking, insulting even. Our writing can be accessible to a lay reader and simplify complex points, perhaps by missing out some of the hardest bits, when the source intention was to convey all the nuance and complexity of the subject that only a PhD could understand. I could go on.

Unless a photograph is being used on Wikipedia as an example of a specific photo (e.g. one an astronaut took on the Moon) then its purpose is typically to provide visual encyclopaedic information about the subject. What someone looks like, say. If I wrote a sentence about what Middleton looks like, combining sources that discuss her hair colour with sources that discuss her make up and sources that discuss her skin tone, that isn't SYNTH. If I claimed her choices were clashing or bad combinations, when no source did about her, that would be SYNTH, even if you can find sources claiming that about those combinations in general. I think you've imagined a problem and would benefit from rowing back on it. Because unfortunately a lot of people think OR applies to user created images as a starting point, and completely forget about the entirely novel user created sentence they just added to a Wikipedia article. -- Colin°^Talk 09:27, 20 February 2026 (UTC)

MjolnirPants, can we please stop being distracted by IAR. You will convince nobody to alter a P&G page with an IAR argument no matter what your intention of using it. Strange, I was under the impression that "Nobody's citing IAR to suggest we can do something here." could not possibly be phrased more clearly. To repeat: You're tilting at a straw man and failing to read what I've actually said about IAR. I've never suggested that we can use it to change anything throughout this discussion, and continuing to put words in my mouth in the face of me explicitly and repeatedly disclaiming them so that you have something to argue with is a prime example of a battleground mentality.

I think MP, you screwed up with the claim... Your lack of any argument against that and, indeed, your comments above about post-processing which directly support it, seem to belie that.

That's not a definition of OR. The definition of OR is "material—such as facts, allegations, and ideas—for which no reliable, published source exists." The colors of objects and people are ideas (arguably facts as well). Colorizing an image produces such ideas, and does so without a source to confirm them.

Whenever Wikipedian's get OR wrong about images, I think it is good to remind ourselves that we are required by policy to write original text in articles: WP:OUROWNWORDS. Yes, I've already raised that point as an example.

Unless a photograph is being used on Wikipedia as an example of a specific photo (e.g. one an astronaut took on the Moon) then its purpose is typically to provide visual encyclopaedic information about the subject. Which is why colorized black and white photos and original illustrations are allowed.

What someone looks like, say. If I wrote a sentence about what Middleton looks like, combining sources that discuss her hair colour with sources that discuss her make up and sources that discuss her skin tone, that isn't SYNTH. No, but no such sources were used in that colorization, nor indeed, in any other colorization I've done. I did, in fact, use a source which stated that her dress was blue, but 'blue' encompasses a lot of different shades (for example:        versus       ), and the shade I chose was based on what would look the most natural in that image, not from sources. What you're suggesting here is akin to suggesting that it's okay to state that a water bear is 'huge' in wikivoice, because it is large in comparison to bacteria. In fact, a tardigrade is quite huge in comparison to a bacteria, so it is fair to use 'huge', even though doing so is highly misleading and against the spirit of an encyclopedia unless we are specific about what it is huge in comparison to. Similarly, the shade of blue I chose might be highly misleading to the viewer, however that doesn't matter in this case, as the image serves to illustrate what Middleton looks like, so we don't care that the dress isn't a 1:1 match to reality.

I think you've imagined a problem and would benefit from rowing back on it. Because unfortunately a lot of people think OR applies to user created images as a starting point, and completely forget about the entirely novel user created sentence they just added to a Wikipedia article. I think you're not responding to anything I actually wrote, because I have made the point several times that there is no problem and nothing needs to be done to address this lack of a problem, because work is continuing regardless of what editors may think of the issue.

I think you've come into this discussion, having only read a few parts of it, and formed an idea of what is being discussed. Except that idea is wrong. The only things you are saying which actually disagree with what I've written here are either unsupported assertions (that I got some part wrong) or disagreements with assertions I have never made, and indeed, have explicitly disclaimed making. To help, I will quote what I actually said above about what should be done here, which has been completely ignored thus far.

Now, getting down to the actual question posed here, I agree that changing the word 'images' to 'photographs' is best. I also prefer wording that is unambiguous: the world 'likely' should be removed from that, so that it clearly states that colorization is OR. If others are amenable, a note that sums up my third paragraph above (that colorized images still have a place here) could be added, though I don't believe that's strictly necessary. Those of us who do this kind of work are all old hands at this, and we don't need a wikispace link to give us permission to do what we've been doing (mainly at the request of others) for years.

All of what has follows has been the result of WhatamIdoing arguing with me about what I said in preface to that, which has been an enjoyable -if often confusing- argument that really has no business here (a point which I have already made), and my multiple efforts to steer the conversation back to the topic of actually making changes to this page have, thus far, gone ignored in favor of continuing to argue with me (and GRuban) over the stuff that doesn't actually matter.

If you would like this to be a productive discussion, you could, instead of putting words in my mouth to continue arguing with me about something that won't result in any changes here, respond to my last paragraph above:

I would get behind this. Colorization is far from the most common request over at the graphics lab, but it's also far from uncommon, and it usually is done with little fanfare and the image quietly used with little to no objection. So a short discussion (even one which references a very long, prior discussion) that resulted in language being added to turn a relatively common practice into a policy vio seems ripe for removal.

ᛗᛁᛟᛚᚾᛁᚱPants Tell me all about it. 17:13, 20 February 2026 (UTC)

I support the idea of changing images to photographs (and videos, too, if that's a request that appears more than about once a year or so). The reason that I added likely is precisely because colorizing non-photograph "images" usually isn't going to be OR, so it'd be stupid to say or imply "is always" when we know that it's definitely not always.

Now: If we narrow this to "photographs", is someone going to claim that colorizing say, a woodcut illustration from a 19th-century book is problematic? Are we leading editors into trouble if we narrow specifically to photographs?

But: If we change "A Wikipedia editor who adds color to such an image is likely engaging in original research" to say "A Wikipedia editor who adds color to a photograph is engaging in original research", then we are effectively banning your colorized photographs. Is that what you want? It'll be a short trip from "MOS:COLORIZE says it's always OR and MP confessed on the talk page that it was definitely OR" to "WP:OR is banned, so kill the photo". I can think of a few editors who would be happy to start an RFC to remove the image you keep talking about, if that's how you want to define things.

Also: Is colorizing a photo really, truly, always OR? False color gets added to photos (e.g., (e.g., micrographs, or to highlight specific features in aerial photos). That's "colorizing". Do we think that's OR? WhatamIdoing (talk) 03:38, 21 February 2026 (UTC)

I dunno. There was somebody using your account above talking about "The whole IAR argument", so I think somebody's citing IAR to claim something here. This is too childish and pointless to warrant any response further than this sentence. ᛗᛁᛟᛚᚾᛁᚱPants Tell me all about it. 16:42, 20 February 2026 (UTC)

Images on the left

Hello! I am hoping for some clarity about placing most images on the right, per WP:IMAGELOC. It clearly states that left-aligned images can disrupt bulleted lists and other similar structures, so that should be avoided. But if there aren't any such structures nearby, and there are no risks of sandwiching, why should "most" images be placed on the right? The endnote reads "A consistent left margin creates a stable anchor for tracking through lines of text, making it more readable or accessible." I must confess I don't understand what that means. If someone could clarify this, that would be awesome. I would also like to see such clarification introduced into the IMAGELOC section. Thanks in advance. OrdinaryOtter (talk) 16:55, 20 February 2026 (UTC)

But if there aren't any such structures nearby, and there are no risks of sandwiching, why should "most" images be placed on the right?

The easy answer is that it's a matter of maintaining a uniform appearance to articles, which can help readers navigate more easily.

The endnote reads "A consistent left margin creates a stable anchor for tracking through lines of text, making it more readable or accessible." I must confess I don't understand what that means. If someone could clarify this, that would be awesome.

That line is saying that keeping images to the right helps readers always know where a new paragraph will begin: all the way on the left side of the page (ignoring the sidebars). It prevents, for example, text from 'jumping' to the left following an image, mid-paragraph.

Here's an example of an image which is floated to the left and which, consequentially, will disrupt this rather long paragraph once this paragraph extends past the bottom of the image (you may need to adjust the width of your browser to see what I mean). I'm going to type a lot of stuff here, just to fill it up a bit and make sure that it will illustrate my point even on a 1920X1080 screen. To do that, I'm going to use the Lorem ipsum text, which is quite common in printing and copysetting work. It's really just a wall of text written in Latin that became something of a meme due to its common usage in establishing what a large amount of text might look like. Lorem ipsum dolor sit amet, consectetur adipiscing elit, sed do eiusmod tempor incididunt ut labore et dolore magna aliqua. Ut enim ad minim veniam, quis nostrud exercitation ullamco laboris nisi ut aliquip ex ea commodo consequat. Duis aute irure dolor in reprehenderit in voluptate velit esse cillum dolore eu fugiat nulla pariatur. Excepteur sint occaecat cupidatat non proident, sunt in culpa qui officia deserunt mollit anim id est laborum.

Notice the way the text jumps further to the left? That can be confusing to some people. You can also read about some technical issues with left-aligned (floated) images at Help:Extended image syntax.

Hope that helps, and feel free to ask any follow up questions you may have. ᛗᛁᛟᛚᚾᛁᚱPants Tell me all about it. 17:40, 20 February 2026 (UTC)

Thanks, that is helpful! Do you know if there's a way to get a few words of your explanation into the MOS page? OrdinaryOtter (talk) 17:57, 20 February 2026 (UTC)

Yeah, it can be edited in by any editor. But I think, with any policy decision, it's a good idea to discuss the exact wording, first.

So what's currently there is:

Left-aligned images may disturb the layout of bulleted lists and similar structures that depend on visual uniformity, e.g. by pushing some items on such lists further inward. Hence, avoid left-aligned images near such structures.

And off the top of my head, what I would replace it with is:

Left-aligned images may make articles more confusing. They can make lines of text jump to the left mid-paragraph and may disturb the layout of bulleted lists and similar structures that depend on visual uniformity, e.g. by pushing some items on such lists further inward. Hence, avoid left-aligned images near such structures.

It's really just adding another example of things that can go wrong with left-aligned images, but it's an example that doesn't rely on the reader having a basic understanding of how html handles lists. ᛗᛁᛟᛚᚾᛁᚱPants Tell me all about it. 19:57, 20 February 2026 (UTC)

This is a good start. I would change "jump to the left mid-paragraph" to "jump to the left in the middle of a paragraph" for clarity, since when I first read it, it sounded like the text was jumping into something called the "left mid-paragraph".

Do you think we could also incorporate that endnote into the body of text? OrdinaryOtter (talk) 21:50, 20 February 2026 (UTC)

"More confusing" will be misunderstood. Editors who interpret "confusing" as a property of the word choice rather than the layout will reject the advice because "obviously" we don't know what we're talking about. Consider something like "less legible" or "harder to read". (But not "less readable", because readability is a property or words and sentences rather than size and placement.) WhatamIdoing (talk) 03:43, 21 February 2026 (UTC)

"More confusing" will be misunderstood. No, it won't, and your explanation is just this side of word salad. Your suggestions are fine, for whatever that's worse. ᛗᛁᛟᛚᚾᛁᚱPants Tell me all about it. 02:14, 22 February 2026 (UTC)

I would change "jump to the left mid-paragraph" to "jump to the left in the middle of a paragraph" for clarity, Agreed. ᛗᛁᛟᛚᚾᛁᚱPants Tell me all about it. 02:15, 22 February 2026 (UTC)

What about:

Left-aligned images may distort the layout of an article and make it harder to navigate. They can make lines of text jump to the left in the middle of a paragraph, and may disturb the layout of bulleted lists and similar structures that depend on visual uniformity, e.g. by pushing some items on such lists further inward. Hence, avoid left-aligned images whenever possible, and always near such structures.

If we use the above version, I think we could cut the endnote out. Its mention of sandwiching seems unnecessary, because sandwiching is discussed right below.

WhatamIdoing and ᛗᛁᛟᛚᚾᛁᚱPants, please be nice to each other. 😊 OrdinaryOtter (talk) 06:14, 22 February 2026 (UTC)

I just had a rather enjoyable argument with WhatamIdoing and don't harbor any ill will at all, I just meant what I said about word salad.

I'm totally okay with your suggestion, and say we run with it. If there are no objections by the time I check this page tomorrow, I'll add it (assuming nobody else did, first). ᛗᛁᛟᛚᚾᛁᚱPants Tell me all about it. 01:37, 23 February 2026 (UTC)

Sounds good. What we drafted here is an explanation of existing policy; is there a more rigorous process for adding brand new policy to the MOS? OrdinaryOtter (talk) 02:06, 23 February 2026 (UTC)

If you're making a (really) major change, then you should make a proper WP:PROPOSAL. If you think the change will produce significant irritation, then you can start an WP:RFC. But for everything else, an ordinary talk page discussion like this one is plenty good enough. (For very small changes to wording, even a short discussion can be overkill.) WhatamIdoing (talk) 04:22, 23 February 2026 (UTC)

This is exactly right. ᛗᛁᛟᛚᚾᛁᚱPants Tell me all about it. 16:37, 23 February 2026 (UTC)

I just made the change. Let's see how long before it gets drive-by reverted with the edit summary "such major changes to the MOS should be subject to an RfC or community-wide discussion first", or words to that effect, heh. ᛗᛁᛟᛚᚾᛁᚱPants Tell me all about it. 16:41, 23 February 2026 (UTC)

I'll preemptively support the change before that happens. -- LCU ActivelyDisinterested «@» °∆t° 20:51, 23 February 2026 (UTC)

I'll preemptively support your support of the changes. Now, if we can get someone to preemptively support my support of your support for the changes, we might be able to go back in time and get this added back around 2004 or so. ᛗᛁᛟᛚᚾᛁᚱPants Tell me all about it. 17:36, 24 February 2026 (UTC)

@OrdinaryOtter: What sort of situation are you referring to when you write that a left-aligned-image can create unnecessary white space below the image? Belbury (talk) 19:43, 24 March 2026 (UTC)

There could be a difference between our devices, but when I view a Wikipedia article on my computer, there is often a chunk of white space underneath left-aligned images, and I've always just assumed that it's because the text has a harder time wrapping precisely around those images. It's interesting that you've never experienced this. If you're able, you might try viewing WP on a different device or using your browser's zoom function to see how the formatting changes when you zoom in and out. For example, on my computer, zooming makes the problem better with some images and worse with others, as the display adapts to the zoom level. Right-aligned images are usually not affected. OrdinaryOtter (talk) 21:30, 24 March 2026 (UTC)

Text flowing around a left-aligned image

Do you mean the situation shown to the right, where there isn't enough space below the thumbnail to take "Abramson" to the left margin, so it leaves a gap and takes the next line instead? Belbury (talk) 21:47, 24 March 2026 (UTC)

Yes. For whatever reason, sometimes that gap is very small or barely noticeable, and sometimes it can be huge and awkward. Your example is kind of in the middle. OrdinaryOtter (talk) 21:50, 24 March 2026 (UTC)

I can see how that would vary depending on the thumbnail height. Wouldn't the whitespace be the same if the image was on the right, though, in any given paragraph where it was happening? I suppose it might be more noticeable on the left because it's a rectangular block, where it would be merging into the varying lengths of each text line, on the right. Belbury (talk) 21:55, 24 March 2026 (UTC)

It's possible that either (A) it doesn't happen as much (or as egregiously) with right-aligned images for some reason, or (B) I personally notice it more (and am bothered by it more) with left-aligned images.

If the discrepancy turns out to be about my own perception and/or aesthetic preferences, then we can remove the mention of the extra white space issue from the article. I would like to see what others think, so we'll know if this is just a "me" issue. 😊 OrdinaryOtter (talk) 22:06, 24 March 2026 (UTC)

It's less noticeable with right-aligned images because of our ragged right text alignment. In this screenshot, look at the shape of the whitespace. It's a bit like a P lying down, or a ౨ character. The left-aligned paragraph draws a line straight down, showing you the exact corner where the text resumes. The ragged right photo caption leaves a gap at the end of the lines, trapping white space between it at the paragraph of text. Some people will notice it more than others. (Those people might consider a career in book design and page layout work.) WhatamIdoing (talk) 22:35, 24 March 2026 (UTC)

Thanks for the reply. Did you mean to insert a link to a screenshot? OrdinaryOtter (talk) 23:41, 24 March 2026 (UTC)

I think WhatamIdoing is talking about the earlier "Abramson" screenshot.

A right-aligned image with whitespace beneath

The whitespace will tend to be statistically less common on right-aligned images, but is still prominent where it appears. Aesthetically I'd agree it wasn't perfect, but would consider the issue too minor (and too difficult to explain!) to single out in the MOS as a drawback of left-aligned images. Belbury (talk) 15:05, 25 March 2026 (UTC)

Yes, I'm talking about the "Abramson" screenshot.

The vertical whitespace is a function of image width vs font size. If the image is shorter (vertically) than the paragraph (as displayed on that reader's screen), a gap underneath is exactly as likely, no matter which way it is displayed.

If the bottom of the image happens to fall just above the last line of the paragraph (in this second screenshot, imagine if "the invention of the..." were a sentence containing only a few more words), then it could actually end up with an even larger vertical gap.

The difference is that ragged right makes the predicted whitespace slightly less defined, so it's not as easily noticed by most people. WhatamIdoing (talk) 15:15, 25 March 2026 (UTC)

Here is a good example of what you are talking about, from Norway.

Another issue with this is that it produces this strange and undesirable floating text effect when the following line is short and the paragraph ends, as here. This is of course precluded by placing images on the right. Wh1pla5h99 (talk) 22:05, 5 April 2026 (UTC)

All floated objects - not just images - have a bottom margin which creates a visual gap. For thumbnail images, this is 1.3em. Similarly, paragraphs of text have a top margin of 0.5em. This means that there must be a minimum of 1.8em between the bottom image border and the top of a text line directly below. This is not measured from the top of the tallest letter, but from the top of the text line - an invisible box enclosing all text which creates the visual gap between two lines of line-wrapped text. --Redrose64 🌹 (talk) 18:51, 25 March 2026 (UTC)

This is all quite illuminating, thank you. Do you think my addition to the MOS about extra white space is warranted, or should we take it out? I am okay with either route. OrdinaryOtter (talk) 17:42, 25 March 2026 (UTC)

I'm inclined to take it out. WhatamIdoing (talk) 19:14, 25 March 2026 (UTC)

 Done OrdinaryOtter (talk) 22:12, 25 March 2026 (UTC)

Arbor-treeish break

Sorry to be late to the party, but I strongly object to avoid left-aligned images whenever possible. That's way too strong. Saying that left-aligned should not be used near bullets etc. is fine, but "avoid whenever possible" is going to give the green light to some mindless gnome to go around moving everything to the right, unless someone else can "prove" that right-alignment is "impossible" in a given situation. Furthermore, A consistent left margin creates a stable anchor for tracking through lines of text, making it more readable or accessible is just so much unsupported bullshit.

The old text was better. EEng 22:46, 25 March 2026 (UTC)

If you read my post which began this thread, I was asking for clarify about why the MOS says most images should be placed on the right. I had not been aware of this policy, and was a little surprised by it. After some discussion, I felt like I understood that guideline. The process I undertook with a couple other editors to change the text of the MOS was not meant to introduce a new policy, but rather to provide clarification of the reasons for the existing policy. If we failed to accomplish that, and introduced text that is too strong, that is a valid complaint to be discussed. I am not opposed to changing the text back to how it was.

After I felt like I understood the reasons for the guideline, I began moving left-aligned images to the right when I encountered them, as long as a right-alignment did not create any problems. I was doing this in good faith—attempting to edit Wikipedia in accordance with the MOS. To have my well-intentioned work be characterized as that of a "mindless gnome" feels really hurtful and mean. To be honest, comments like this ruin my day and make me want to quit Wikipedia. OrdinaryOtter (talk) 23:41, 25 March 2026 (UTC)

What you're describing -- making changes here and there in situations you happen to encounter -- are not what I'm calling mindless gnoming, even if it's based on a guideline which shouldn't read as it's recently been changed to read. Mindless gnoming would be someone going systematically from article to article on a search-and-destroy mission to eliminate left-aligned images willy-nilly. And -- trust me -- if the guideline stays the way it is, someone sooner or later will start doing that. EEng 00:07, 26 March 2026 (UTC)

Thank you for the clarification. Unfortunately, I don't think I have the patience for another discussion of the relevant section. I'll ping ᛗᛁᛟᛚᚾᛁᚱPants and @WhatamIdoing, in case they haven't been following the thread recently, as they were involved in the initial changes we made. I trust that whoever is involved with this stage of the discussion will arrive at a sensible conclusion, and like I said earlier, I also don't mind if the text gets reverted to its original state. Please feel free to reach out to me with any specific questions or concerns. OrdinaryOtter (talk) 01:20, 26 March 2026 (UTC)

@EEng, maybe you should try cleaning up that section of the guideline. WhatamIdoing (talk) 03:54, 26 March 2026 (UTC)

Saying to avoid left-aligned images whenever possible (and I'm finding it hard to imagine a situation where it wouldn't be possible) does also go against the later MOS recommendation that For people, animals etc. it is preferable to place images so that they face towards the text. Belbury (talk) 09:42, 26 March 2026 (UTC)

Think the meaning has been lost over the years. - "Left-aligned text with a consistent left margin is widely recognized as a cornerstone of digital accessibility, enhancing readability for all users, particularly those with dyslexia, low vision, or cognitive impairments. Placing images on the right can be helpful strategy to support text flow, but the primary accessibility requirement is a steady left margin, include meaningful alt text,

Despite what I have written ...this is an example of something on the left that causes accessibility problems ...as without any spacing this is a readable nightmare for those with dyslexia (me) and other visual impairments

Rank	Country	Number	Percentage
1	China	33,908	13.1
2	India	30,576	11.99

Placing images on the right side of text is not inherently better for accessibility ..the main accessibility consideration for image placement is ensuring they are in line with the prose so that screen readers can interpret images related to the prose. Something like India#Ancient India is a screen reading nightmare as the images are at the end thus an afterthought layout wise conveying info/data for screen readers way beyond the related information in prose. Images should be used sparingly for text, as they hinder readability. Moxy🍁 19:33, 26 March 2026 (UTC)

For people who have trouble reading, having a generous number of images or other features is also helpful. An image can help visually help people keep their place in the text. WhatamIdoing (talk) 20:14, 26 March 2026 (UTC)

Agree 100%..... the images just need to be near the prose so that it all makes sense for those using screen readers and for those with visual impairments I would imagine. Moxy🍁 20:18, 26 March 2026 (UTC)

Saying that text should be aligned on the left means it should look like this:

Lorem ipsum dolor sit amet, consectetur adipiscing elit, sed do eiusmod tempor incididunt ut labore et dolore magna aliqua. Ut enim ad minim veniam, quis nostrud exercitation ullamco laboris nisi ut aliquip ex ea commodo consequat. Duis aute irure dolor in reprehenderit in voluptate velit esse cillum dolore eu fugiat nulla pariatur. Excepteur sint occaecat cupidatat non proident, sunt in culpa qui officia deserunt mollit anim id est laborum."

not this:

Lorem ipsum dolor sit amet, consectetur adipiscing elit, sed do eiusmod tempor incididunt ut labore et dolore magna aliqua. Ut enim ad minim veniam, quis nostrud exercitation ullamco laboris nisi ut aliquip ex ea commodo consequat. Duis aute irure dolor in reprehenderit in voluptate velit esse cillum dolore eu fugiat nulla pariatur. Excepteur sint occaecat cupidatat non proident, sunt in culpa qui officia deserunt mollit anim id est laborum.

It doesn't mean you can't have floating elements on the left. Even the quote above -- Placing images on the right can be helpful strategy, but the primary accessibility requirement is a steady left margin -- is hardly an injunction against images on the left, and in fact makes clear that "steady left margin" is not anything about no-left-images. Instead, it's telling you not to have a ragged left margin like I exemplified above.

The parade of horribles may distort the layout of an article and make it harder to navigate. They can make lines of text jump to the left in the middle of a paragraph adds nothing: I don't know what "distort the layout" means or how left-aligned images make the article "harder to navigate"; and "can make lines of text jump to the left" is just another way of saying "text flows around a floated image", which is exactly what's supposed to happen.

I vote we go back to the text from before: Most images should be on the right side of the page, which is the default placement.[nb 4] Left-aligned images may disturb the layout of bulleted lists and similar structures that depend on visual uniformity, e.g. by pushing some items on such lists further inward. Hence, avoid left-aligned images near such structures. EEng 02:55, 27 March 2026 (UTC)

I wonder if some screenshots of impeded bullet lists and a MOS:SANDWICH would be helpful. (Whether the images in MOS:SANDWICH are problematic depends on your screen size and skin choice. On mobile, the two images stack nicely one above the other, and there is no difficulty at all.) WhatamIdoing (talk) 19:24, 31 March 2026 (UTC)

I'd support going back to the earlier wording. I don't see the "jump to the left" aspect as being an issue, when there are no bullet lists, indents or headings involved. Belbury (talk) 08:18, 17 April 2026 (UTC)

Image markup embedded in paragraphs

Recently I've been seeing a lot of images with the markup embedded inside a paragraph, like this:

Westminster Palace is somewhere in England, I think.[[File:Westminstpalace.jpg|thumb|Figure 1.1]]But I'm not really sure about that.

This doesn't actually embed the image in the paragraph, because images (almost) always produce a paragraph break. So I've been routinely breaking these up in the markup, putting a blank line before the image and a line break after it. I do this to make the paragraph breaks easier to find in the markup. The rendered text is the same both ways. But maybe I shouldn't be doing this? Some editors get touchy about their markup, and what appears to me to be a mistake might be legitimate to someone else. GA-RT-22 (talk) 19:16, 16 March 2026 (UTC)

I think we want image markup just ahead of the paragraph the text it supports, if floating. Makes it far cleaner to edit and otherwise has no net effect on display. Masem (t) 19:22, 16 March 2026 (UTC)

+1 ᛗᛁᛟᛚᚾᛁᚱPants Tell me all about it. 20:08, 16 March 2026 (UTC)

I think that it's WP:VE that causes it. I add newlines, sometimes moving the image to before the content that it relates to, as here. I've been doing it for years. I recall a discussion somewhere several years ago that indicated that absence of newlines could make editing more difficult for screen reader users. --Redrose64 🌹 (talk) 21:17, 16 March 2026 (UTC)

Yes, that's exactly what I'm talking about, and that's the same fix I usually apply. Would it be appropriate to add something to the MOS? I'd like to have a guideline I can point to if someone challenges these changes. GA-RT-22 (talk) 00:09, 17 March 2026 (UTC)

Nobody is going to challenge those.

Also, I'm pretty sure that there are already about 1.5 lines for this in the rules somewhere. One is probably in WP:ACCESS and says that the image goes at the start of (i.e., not in the middle of) the relevant content. The other says that we should leave a blank line before a ==Section== heading to make it easier for editors using a wikitext editing environment to find the section breaks. This is just an extension of that. WhatamIdoing (talk) 19:51, 24 March 2026 (UTC)

A new dark mode section

We address dark mode issues with respect to text at Wikipedia:Manual of Style/Text formatting, where section § Color, point 6 says regarding text color issues: "Content should also be easy to read in dark mode". I would say a very similar point applies to images in dark mode with respect to background color, and we ought to say something about it on this page. For a concrete example, see WT:WikiProject Infoboxes#Monochrome company logos with light/dark mode radio buttons. For general background, see mw:Recommendations for night mode compatibility on Wikimedia wikis, such as § Apply filters to dark images with transparent background. Thanks, Mathglot (talk) 21:46, 6 April 2026 (UTC)

Criteria for a "wide image"

I have started a discussion at Template talk:Wide image#Criteria for a "wide image" regarding the size dimensions of a "wide image". Your input is welcome. Magnolia677 (talk) 11:11, 8 April 2026 (UTC)

mos:peoplegallery

At Wikipedia's image use policy, WP:GALLERY reads: "There is consensus not to use a gallery of group members as the lead image for articles about large groups of people such as ethnicities" (emphasis added), while the manual of style for images reads, at MOS:PEOPLEGALLERY: "Articles about ethnic groups or similarly large human populations should not be illustrated by a photomontage or gallery of images of group members" (emphasis added), and refers the reader to this discussion ("Articles about ethnic groups or similarly large human populations should not be illustrated by a gallery of images of group members") and this conclusion. That said, I would very much appreciate some help on

• whether the discussions cited in MOS:PEOPLEGALLERY are, in fact, the source of the most recent consensus,
• whether the text at WP:GALLERY should extend the policy to the whole article (instead of just the lead), which seems to have been the conclusion aforementioned (though, admittedly, I haven't read the whole discussion, so I might have misinterpreted),
• whether MOS:PEOPLEGALLERY's mention of "photomontage" (whatever that means) can also be attributed to consensus (not necessarily from the RfC linked above), and, perhaps most importantly,
• whether it's relevant to raise these issues here, in addition to or instead of a case-by-case approach (i.e. directly in the articles' talk pages).

For a concrete example (which also happens to be what brought me here), see Brazilians § History and overview. Myselfthethird (talk) 21:04, 10 May 2026 (UTC)

Historical images

The MOS doesn't currently provide guidance on how images depicting historical events/figures should be selected, which has led some articles to include images of lesser encyclopedic value when that is outweighed by editors' personal preferences (An example of this: much later painting v. an almost contemporaneous image of the same historical event) Since this has been a recurring issue in several articles, I propose that WP:PORTRAIT, currently an essay, be incorporated into this guideline. GumballNine1Nine (talk) 06:54, 20 May 2026 (UTC)

Template talk:Infobox military conflict#RfC to deprecate collages in conflict infoboxes

Please see subject discussion. Cinderella157 (talk) 22:35, 28 May 2026 (UTC)

Extremely large image on a page

Hi, can I ask for a third option regarding the inclusion of an extremely large image on this page in this discussion (Talk:1911 World Artistic Gymnastics Championships)? Thank you. FromCzech (talk) 04:15, 11 June 2026 (UTC)

@FromCzech I fixed it EvergreenFir (talk) 04:57, 11 June 2026 (UTC)

Thanks! Do you have also an opinion on the length of the caption (see discussion)? FromCzech (talk) 05:17, 11 June 2026 (UTC)