|
How to find images for your Military History articles — and what not to do |
By Adam Cuerden
So, I've been working primarily with images on Wikipedia for, like, a decade now. I have a number of sites I regularly use to find images (and, no doubt, a lot of sites I've used before and forgot about). I know how to figure out what copyright tag to use, I know what good procedure is for documenting an image, and I can give some advice on how to modify an image. And I know how to avoid common pitfalls; to whit, upload the original, unmodified image first, then upload your modified version, and provide links between them, so that A. everyone knows what changes were made, and B. if the site you got the image from goes down, we can still find the original image. But more on that later. So much more.
This started out as a fairly simple article and then kept growing and growing; as such I kind of worry I've ended up with a five-article series that all got published at once. It starts off with a list of sources for images, goes into a description of some techniques to get around attempts to keep you from being able to download images, then explains how to judge if an image is out of copyright, explains how to work out what information to include when uploading an image, then the last two sections were my original plan for the article; a description of common pitfalls that come about when people upload images without documenting changes they've made to them. And then I ended by discussing how to choose images for image restoration. That's a lot. (At least I saved the guide to the actual process of restoring images for later. Sort of.) I'd imagine different parts of this article are going to be useful for different people, so feel free to skip around.
Places to get images |
Before we begin, I should probably note that the archives I use are best for events in Europe and North America after about 1700, Australia and New Zealand after 1800, and then a random selection of earlier periods and the rest of the world depending what's in the archives, though Google Art Project does do something to extend that. But it's just a lot easier to research if you at least know the alphabet in question, so there's going to be some major gaps, since, for example, the best sources for Japanese culture and history are likely to be in Japanese, and I can't write a proper search term in kanji, or even be guaranteed to find the button to get to the search box in Japanese.
Luckily, the legacy of colonialism means that large amounts of treasures of the rest of the world have been dragged off to Europe and America. How convenient! Also luckily, I'm writing for a readership who I think I can expect to understand sarcasm.
Sources in the United States of America
America seems particularly good with releasing content; British sources, for example, tend to be locked down, and Canadian sources are generally unwilling to release more than a thumbnail-sized image. So it's often worth checking these sources first.
There's a lot of sources indexed in DPLA that I actually really like, such as the Minneapolis Institute of Art, but since they are indexed in DPLA, it's probably unnecessary to list them separately.
United Kingdom
The United Kingdom tends to be quite locked down, but there are exceptions. I'm sure there's other good ones, but these are ones I find generally very good:
France
Australia
New Zealand
Norway
The Netherlands
Worldwide
Getting the images when they don't want you to |
There's three main methods you can try:
Copyright |
Image copyright probably deserves its own article, but let's at least cover some common cases. Commons requires that an image be out of copyright both in the United States and its country of origin; English Wikipedia only requires it be out of copyright in the United States. Commons:Copyright rules by territory covers the situation for a lot of countries in a complex, over-detailed way. So here's a very simple guide. If it says something's definitely out of copyright, it should always be right; but there are a number of additional cases where things might be out of copyright that it doesn't include.
For works created outside of the United States, and presuming it's not an anonymous work, you'll need to know the year the photographer or creator died. If you can't find it, see if it comes under {{ PD-old-assumed}}, which basically says that if you can't find the date of death, but the work is more than 120 years old [1] and that if it's more than 190 years old [2] since we know the maximum human lifespan, just treat it as fully out of copyright (Use {{ PD-old-70-expired}}. To find the date of death, first of all, I find libraries often misread names from old cabinet cards if they're written in artistic enough of a font, so check whether there's other readings of the text on them. A Google search for, say, "Bogardus photographer" can sometimes lead you directly to the result; Wikidata also has details for a lot of people, with sources, if you can get enough of their name.
Note that copyright only expires at the end of each calendar year. So, if it says below that a country uses, say, Life + 70 years, then if a creator expired exactly 70 years ago today, Life + 70 will expire 1 January next year.
If the image....
What kind of other cases are there?
There's also a lot of things that are specific to one country that's hard to cover. For example, any photograph from Japan from before 1957 is out of copyright both there and America, because the copyright term for photographs was only 10 years after creation until 1957. Some countries refuse copyright to "simple photographs"... and then you have to dig into case law to find out what that means. But I hope this section at least gives you a guide to understanding the easier cases.
The "wrapper" templates: {{ PD-Art}} and {{ PD-Scan}} |
If the museum or library claims copyright on a two-dimensional image, but going through the list above says it's out of copyright, then we acknowledge that the hosting site says that, then upload it anyway. (If it's a picture of a statue or other significantly three-dimensional artwork, they have a valid claim.) This is done using a wrapper template.
{{ PD-Scan}} is used when it's a mere scan of the image, the kind of thing you'd get by putting a photograph on a scanner, and clicking the "Scan" button, and then maybe giving it some minor levels tweaks. {{ PD-Art}} is a little more general and a little more restrictive, it's for if, say, someone took a photograph of a painting, where they might have had to adjust things slightly more. If you're not sure, just use {{ PD-Art}}. Note that {{ PD-Scan}} is only on Commons, so if you're uploading on English Wikipedia, always use {{ PD-Art}}.
So, let's say you've worked out from about that the copyright tag should be {{ PD-old-auto-expired|deathyear=1903}}. It's a mere scan of a photograph, so we can use the slightly stronger {{ PD-Scan}} template. So, we just kind of add PD-Scan in front of the other bits of the template, so it becomes: {{ PD-Scan|PD-old-auto-expired|deathyear=1903}}.
Documenting an image |
Commons has a certain number of tools to help you do this. I... tend not to use them. So, let's just look at the basic {{ Information}} template, because you can always use the tools, then go back and edit it to add anything you missed.
That's the basic information template. There's also {{ Artwork}} and a few others, but they're much harder to use, and some of them have this weird self-checking code for what you put into them that doesn't always work right, and then... Ech. Just... don't, and use {{ Information}} if you value your sanity.
If the image you're uploading is the original file, add {{ Original}} above the information template. This tells people that this is the original file, you should upload any modified versions as their own file. We'll talk about why that's a good thing more in the next section, but, for now, let's cover everything else, using File:David Livingstone by Thomas Annan - Original.jpg as an example. The image source for it is [1].
"
David Livingstone, 1813 - 1873. Missionary and explorer", a carbon print of a photograph by
Thomas Annan, 36.90 x 30.20 cm. Gift to the National Galleries of Scotland by T & R Annan and Sons, 1930
. Let's go through this, making a rough order that information should appear in.Having done that, add the copyright statement below the information template if you haven't put it into Permission - I still don't know why we have two accepted variants for that, and add a few categories if you can. You can then copy all of this for any other versions of the file, making any changes necessary.
So you want to modify the image? |
I don't want you to be afraid to modify an image. This whole section boils down to "Upload a copy of the original image first, upload your changes as a separate file, and link your modified image to the original version." Because, with a little care, and remembering that subtle changes are often better than huge changes (and pretty much always better than autolevels), you can probably make a better copy of the original image for use on Wikipedia. However, it's misleading to present an altered image as an unaltered original, and it's possible to get an image that genuinely looks better, but, for example, blows out some of the whites, so that information is lost that could have been saved while getting it to look just as nice. If you uploaded the original, this "even better" option is still on the table later, but your copy can be used until that version exists.
So, that said, I'm going to talk about a lot of things that can go wrong. But I don't want to discourage you from trying, just encouraging you to use good image editing behaviour.
First, upload the original to Commons. Mark it with {{ Original}} to tell people not to alter it. Make sure that you upload your altered copies as separate images, and link these images to the original image you just uploaded, stating explicitly that this new copy has been modified.
The documentation is important. Look at these two images of SMS von der Tann:
Despite the obvious differences, these both claim to be the same image, an exact upload of the original source. One isn't. The second one is the actual original file from the Library of Congress. Nothing on the file description page for the first one - at least at time of writing - makes it clear that the first one has been modified in any way.
It's not really a problem to do a bad image adjustment if you've given people the tools to consider how you did by showing people what the image looked like before you did anything.
A good rule of thumb is to look at your image, and look at the original. As long as there's an obvious improvement - and not just at thumbnail: we ideally want people to use our articles as sources to help them find resources as well as information - and you've linked to a copy of the original, everything's great.
And for the record, this "Upload the original image" guide applies to:
Sites change their layout all the time, and many once-free sites now don't allow download of their originals. There's a few cases where someone's mangled an image, but clearly had access to a copy I wish I could work on - but I can't.
Image restoration |
Some images are better for a first-time restorationist than others. Take the following examples, each of which is the original file that a recentish featured restoration was based on.
My rule of thumb is that a first-time restorationist should have a reasonable shot at completing their restoration. If something is too difficult to restore it's just going to result in frustration.
French battleship Justice
First of all, the work in restoring an image is, to some extent, proportional to the image's size. This image is 10,212 × 8,097 pixels in size. Not a great start to things. While one can, in theory, crop it a bit, it's still going to be quite large, and it might result in a poor image composition.
Secondly, while parts of the image aren't too bad, the left edge of this image is appalling. Here's a before-and-after restoration comparison:
Before | After |
---|---|
That level of damage took me around 10-15 hours of work just for that little bit on the left hand side (I'm not sure I'm not underestimating that time) and that's with me being highly efficient as I've been doing this for years. Admittedly, it's on the left edge, so you could crop a bit, but it's not clear if you could crop enough to actually make things reasonable without the ship ending up way too tightly cropped. Reviewing an image beforehand to check for problem areas is a good first step before considering restoration. This also gives you a chance to try to figure out anything that might be mistaken for damage but isn't, such as portholes that show up as dark spots on the image, but aren't damage unlike those other dark spots.
Percy Grainger
It has cracks on the negative, and the pieces are slightly misaligned. One can use a careful hand with the select tool to realign them, then restore over the cracks, but cracks, lines, and other such things are relatively hard: the "healing brush" tool can be thrown by the change in colour leaving clear signs of where they were, and they often intersect detailed parts of the image that need carefully pattern matched.
William Harvey Carney
Had I not already restored this one, this would be a pretty good choice for a novice restorationist. There's a tiny bit of weird damage in the lower left, which is small enough to be fairly easy to fix, or it could be cropped out with little damage to composition. There's some damage a bit higher up on the left intruding in from the left hand side, but it's over a bit of non-detailed background that doesn't really matter, so whatever you do to fix it it's not going to look terrible. The rest of the damage is pretty much isolated spots, which are easily fixed with even a half-careful use of the healing brush.
Of course, selecting images is the first step. I'll cover how to do the restoration in another article, but, for now:
I use GIMP, the GNU Image Manipulation Program, which is free. For basic damage, I like to use the healing brush tool, with a hardness 100 brush, set to a size just slightly bigger than the spots, (generally not bigger than about 12px wide), selecting as the source of the pattern something as close to the damaged area as possible when the patterns are detailed (it matters a lot less with tiny damage spots on relatively flat areas of colour.
The clone stamp is sometimes needed where there's a hole or large damage and the healing stamp would change the colours too much; for that I like the hardness 75 brush as it helps it blend in.
The dodge and burn tools, at generally very low opacities (2-10%) and like, a hardness 25 brush, can be quite useful for dealing with areas that are lighter or darker than they should be - like fading - but they handle colour somewhat badly if you try to take it too far. Size depends on what you're fixing; a lot of the time the lightened area has a certain width, so you want to try to match that. Other times I want it quite large. Just hit undo if it's not right the first time. This is probably the easiest one of these to get wrong, saving only...
Levels and curves: I find it's better to be a lot more subtle than you'd think. Don't use autolevels; it's generally absolutely terrible. You'll often need to nudge saturation down a little bit after a large change in levels, as the colours can end up very over-saturated.
|
|
How to find images for your Military History articles — and what not to do |
By Adam Cuerden
So, I've been working primarily with images on Wikipedia for, like, a decade now. I have a number of sites I regularly use to find images (and, no doubt, a lot of sites I've used before and forgot about). I know how to figure out what copyright tag to use, I know what good procedure is for documenting an image, and I can give some advice on how to modify an image. And I know how to avoid common pitfalls; to whit, upload the original, unmodified image first, then upload your modified version, and provide links between them, so that A. everyone knows what changes were made, and B. if the site you got the image from goes down, we can still find the original image. But more on that later. So much more.
This started out as a fairly simple article and then kept growing and growing; as such I kind of worry I've ended up with a five-article series that all got published at once. It starts off with a list of sources for images, goes into a description of some techniques to get around attempts to keep you from being able to download images, then explains how to judge if an image is out of copyright, explains how to work out what information to include when uploading an image, then the last two sections were my original plan for the article; a description of common pitfalls that come about when people upload images without documenting changes they've made to them. And then I ended by discussing how to choose images for image restoration. That's a lot. (At least I saved the guide to the actual process of restoring images for later. Sort of.) I'd imagine different parts of this article are going to be useful for different people, so feel free to skip around.
Places to get images |
Before we begin, I should probably note that the archives I use are best for events in Europe and North America after about 1700, Australia and New Zealand after 1800, and then a random selection of earlier periods and the rest of the world depending what's in the archives, though Google Art Project does do something to extend that. But it's just a lot easier to research if you at least know the alphabet in question, so there's going to be some major gaps, since, for example, the best sources for Japanese culture and history are likely to be in Japanese, and I can't write a proper search term in kanji, or even be guaranteed to find the button to get to the search box in Japanese.
Luckily, the legacy of colonialism means that large amounts of treasures of the rest of the world have been dragged off to Europe and America. How convenient! Also luckily, I'm writing for a readership who I think I can expect to understand sarcasm.
Sources in the United States of America
America seems particularly good with releasing content; British sources, for example, tend to be locked down, and Canadian sources are generally unwilling to release more than a thumbnail-sized image. So it's often worth checking these sources first.
There's a lot of sources indexed in DPLA that I actually really like, such as the Minneapolis Institute of Art, but since they are indexed in DPLA, it's probably unnecessary to list them separately.
United Kingdom
The United Kingdom tends to be quite locked down, but there are exceptions. I'm sure there's other good ones, but these are ones I find generally very good:
France
Australia
New Zealand
Norway
The Netherlands
Worldwide
Getting the images when they don't want you to |
There's three main methods you can try:
Copyright |
Image copyright probably deserves its own article, but let's at least cover some common cases. Commons requires that an image be out of copyright both in the United States and its country of origin; English Wikipedia only requires it be out of copyright in the United States. Commons:Copyright rules by territory covers the situation for a lot of countries in a complex, over-detailed way. So here's a very simple guide. If it says something's definitely out of copyright, it should always be right; but there are a number of additional cases where things might be out of copyright that it doesn't include.
For works created outside of the United States, and presuming it's not an anonymous work, you'll need to know the year the photographer or creator died. If you can't find it, see if it comes under {{ PD-old-assumed}}, which basically says that if you can't find the date of death, but the work is more than 120 years old [1] and that if it's more than 190 years old [2] since we know the maximum human lifespan, just treat it as fully out of copyright (Use {{ PD-old-70-expired}}. To find the date of death, first of all, I find libraries often misread names from old cabinet cards if they're written in artistic enough of a font, so check whether there's other readings of the text on them. A Google search for, say, "Bogardus photographer" can sometimes lead you directly to the result; Wikidata also has details for a lot of people, with sources, if you can get enough of their name.
Note that copyright only expires at the end of each calendar year. So, if it says below that a country uses, say, Life + 70 years, then if a creator expired exactly 70 years ago today, Life + 70 will expire 1 January next year.
If the image....
What kind of other cases are there?
There's also a lot of things that are specific to one country that's hard to cover. For example, any photograph from Japan from before 1957 is out of copyright both there and America, because the copyright term for photographs was only 10 years after creation until 1957. Some countries refuse copyright to "simple photographs"... and then you have to dig into case law to find out what that means. But I hope this section at least gives you a guide to understanding the easier cases.
The "wrapper" templates: {{ PD-Art}} and {{ PD-Scan}} |
If the museum or library claims copyright on a two-dimensional image, but going through the list above says it's out of copyright, then we acknowledge that the hosting site says that, then upload it anyway. (If it's a picture of a statue or other significantly three-dimensional artwork, they have a valid claim.) This is done using a wrapper template.
{{ PD-Scan}} is used when it's a mere scan of the image, the kind of thing you'd get by putting a photograph on a scanner, and clicking the "Scan" button, and then maybe giving it some minor levels tweaks. {{ PD-Art}} is a little more general and a little more restrictive, it's for if, say, someone took a photograph of a painting, where they might have had to adjust things slightly more. If you're not sure, just use {{ PD-Art}}. Note that {{ PD-Scan}} is only on Commons, so if you're uploading on English Wikipedia, always use {{ PD-Art}}.
So, let's say you've worked out from about that the copyright tag should be {{ PD-old-auto-expired|deathyear=1903}}. It's a mere scan of a photograph, so we can use the slightly stronger {{ PD-Scan}} template. So, we just kind of add PD-Scan in front of the other bits of the template, so it becomes: {{ PD-Scan|PD-old-auto-expired|deathyear=1903}}.
Documenting an image |
Commons has a certain number of tools to help you do this. I... tend not to use them. So, let's just look at the basic {{ Information}} template, because you can always use the tools, then go back and edit it to add anything you missed.
That's the basic information template. There's also {{ Artwork}} and a few others, but they're much harder to use, and some of them have this weird self-checking code for what you put into them that doesn't always work right, and then... Ech. Just... don't, and use {{ Information}} if you value your sanity.
If the image you're uploading is the original file, add {{ Original}} above the information template. This tells people that this is the original file, you should upload any modified versions as their own file. We'll talk about why that's a good thing more in the next section, but, for now, let's cover everything else, using File:David Livingstone by Thomas Annan - Original.jpg as an example. The image source for it is [1].
"
David Livingstone, 1813 - 1873. Missionary and explorer", a carbon print of a photograph by
Thomas Annan, 36.90 x 30.20 cm. Gift to the National Galleries of Scotland by T & R Annan and Sons, 1930
. Let's go through this, making a rough order that information should appear in.Having done that, add the copyright statement below the information template if you haven't put it into Permission - I still don't know why we have two accepted variants for that, and add a few categories if you can. You can then copy all of this for any other versions of the file, making any changes necessary.
So you want to modify the image? |
I don't want you to be afraid to modify an image. This whole section boils down to "Upload a copy of the original image first, upload your changes as a separate file, and link your modified image to the original version." Because, with a little care, and remembering that subtle changes are often better than huge changes (and pretty much always better than autolevels), you can probably make a better copy of the original image for use on Wikipedia. However, it's misleading to present an altered image as an unaltered original, and it's possible to get an image that genuinely looks better, but, for example, blows out some of the whites, so that information is lost that could have been saved while getting it to look just as nice. If you uploaded the original, this "even better" option is still on the table later, but your copy can be used until that version exists.
So, that said, I'm going to talk about a lot of things that can go wrong. But I don't want to discourage you from trying, just encouraging you to use good image editing behaviour.
First, upload the original to Commons. Mark it with {{ Original}} to tell people not to alter it. Make sure that you upload your altered copies as separate images, and link these images to the original image you just uploaded, stating explicitly that this new copy has been modified.
The documentation is important. Look at these two images of SMS von der Tann:
Despite the obvious differences, these both claim to be the same image, an exact upload of the original source. One isn't. The second one is the actual original file from the Library of Congress. Nothing on the file description page for the first one - at least at time of writing - makes it clear that the first one has been modified in any way.
It's not really a problem to do a bad image adjustment if you've given people the tools to consider how you did by showing people what the image looked like before you did anything.
A good rule of thumb is to look at your image, and look at the original. As long as there's an obvious improvement - and not just at thumbnail: we ideally want people to use our articles as sources to help them find resources as well as information - and you've linked to a copy of the original, everything's great.
And for the record, this "Upload the original image" guide applies to:
Sites change their layout all the time, and many once-free sites now don't allow download of their originals. There's a few cases where someone's mangled an image, but clearly had access to a copy I wish I could work on - but I can't.
Image restoration |
Some images are better for a first-time restorationist than others. Take the following examples, each of which is the original file that a recentish featured restoration was based on.
My rule of thumb is that a first-time restorationist should have a reasonable shot at completing their restoration. If something is too difficult to restore it's just going to result in frustration.
French battleship Justice
First of all, the work in restoring an image is, to some extent, proportional to the image's size. This image is 10,212 × 8,097 pixels in size. Not a great start to things. While one can, in theory, crop it a bit, it's still going to be quite large, and it might result in a poor image composition.
Secondly, while parts of the image aren't too bad, the left edge of this image is appalling. Here's a before-and-after restoration comparison:
Before | After |
---|---|
That level of damage took me around 10-15 hours of work just for that little bit on the left hand side (I'm not sure I'm not underestimating that time) and that's with me being highly efficient as I've been doing this for years. Admittedly, it's on the left edge, so you could crop a bit, but it's not clear if you could crop enough to actually make things reasonable without the ship ending up way too tightly cropped. Reviewing an image beforehand to check for problem areas is a good first step before considering restoration. This also gives you a chance to try to figure out anything that might be mistaken for damage but isn't, such as portholes that show up as dark spots on the image, but aren't damage unlike those other dark spots.
Percy Grainger
It has cracks on the negative, and the pieces are slightly misaligned. One can use a careful hand with the select tool to realign them, then restore over the cracks, but cracks, lines, and other such things are relatively hard: the "healing brush" tool can be thrown by the change in colour leaving clear signs of where they were, and they often intersect detailed parts of the image that need carefully pattern matched.
William Harvey Carney
Had I not already restored this one, this would be a pretty good choice for a novice restorationist. There's a tiny bit of weird damage in the lower left, which is small enough to be fairly easy to fix, or it could be cropped out with little damage to composition. There's some damage a bit higher up on the left intruding in from the left hand side, but it's over a bit of non-detailed background that doesn't really matter, so whatever you do to fix it it's not going to look terrible. The rest of the damage is pretty much isolated spots, which are easily fixed with even a half-careful use of the healing brush.
Of course, selecting images is the first step. I'll cover how to do the restoration in another article, but, for now:
I use GIMP, the GNU Image Manipulation Program, which is free. For basic damage, I like to use the healing brush tool, with a hardness 100 brush, set to a size just slightly bigger than the spots, (generally not bigger than about 12px wide), selecting as the source of the pattern something as close to the damaged area as possible when the patterns are detailed (it matters a lot less with tiny damage spots on relatively flat areas of colour.
The clone stamp is sometimes needed where there's a hole or large damage and the healing stamp would change the colours too much; for that I like the hardness 75 brush as it helps it blend in.
The dodge and burn tools, at generally very low opacities (2-10%) and like, a hardness 25 brush, can be quite useful for dealing with areas that are lighter or darker than they should be - like fading - but they handle colour somewhat badly if you try to take it too far. Size depends on what you're fixing; a lot of the time the lightened area has a certain width, so you want to try to match that. Other times I want it quite large. Just hit undo if it's not right the first time. This is probably the easiest one of these to get wrong, saving only...
Levels and curves: I find it's better to be a lot more subtle than you'd think. Don't use autolevels; it's generally absolutely terrible. You'll often need to nudge saturation down a little bit after a large change in levels, as the colours can end up very over-saturated.
|