Systeme D

June 7, 2009

Aerial photography, cock fighting and vodka bottles

Big Important Warning. This post argues that it is legal to trace from Google-provided satellite imagery and upload the results to OpenStreetMap. Do not do this. “Legal” is one thing; “accepted by the OSM community” another; and “liable to expose a not-for-profit Foundation to a lawsuit from one of the world’s biggest companies” something else entirely. Seriously, don’t. This post is here for interest and to start a discussion.

Since the year dot, it’s been accepted within OpenStreetMap that you don’t trace from anywhere without express permission (Yahoo!) or a really clear public domain waiver (US Government stuff). And you definitely, definitely don’t trace from Google satellite images.

That said, since about the year dot+1, we’ve had this little sentence which I put in the FAQ back on 6 December 2005:

“It’s not yet clear whether it’s ok to create a derived work from aerial photography: some readings of UK law suggest that you can do this without ‘inheriting’ the copyright in the photography. A definitive ruling on this could open up new avenues for Openstreetmap and similar projects, but in the absence of such a ruling, we’re continuing with the approach of sourcing our own, 100% independent data.”

So three and a half years on, let’s get a bit more definitive.

Bauman vs Fussell

I’m indebted to The Law of Photography and Digital Images by Christina Michalos for several crucial points, and to the Squire Law Library in Cambridge where I was able to consult it. Simon Stokes’ Digital Copyright has also proved a useful primer in related issues.

The question is: does tracing features from an aerial/satellite photograph (I’ll refer to them as aerial throughout) infringe any copyright in the aerial photograph?

The clearest case is Bauman v Fussell (UK, 1953). In this case, a painter created a painting of two cocks fighting, based on a photograph. The only element derived from the photograph was the position of the birds. Michalos summarises the judgement thus:

“1953: Bauman v Fussell (UK) held that the positions of birds not under the photographers control in a photograph not a substantial part of the photograph. A painting copying those positions was not infringing.”

Three judges ruled in this case: Lord Justice Somervell, Lord Justice Birkett, and Lord Justice Romer. The key issue, according to Michalos, was “whether the defendant’s painting reproduced a substantial part of the photograph”.

Somervell and Birkett believed it didn’t, thereby carrying the majority verdict. They concluded that the positions were not a “substantial” part of the artist’s copyrightable work (OSM legal eagles will recognise the word) and therefore could be copied. By analogy, the positions of roads, buildings and other features on an aerial photograph are not a substantial part of the copyright.

Romer, disagreeing with Somervell and Birkett, said the positions of the birds were indeed a substantial part:

“The appellant did not arrange the position of the birds, but no doubt he waited for the moment to take the photograph when the birds were in the position he wanted them to be and his photograph produces the position.”

But, in fact, even Romer’s viewpoint supports the theory that one can copy from aerial photography. Romer is saying here that the photographer’s skill in capturing the moment produced the original, copyrightable work which was infringed. This cannot apply to aerial photography. There is no originality in an orthophoto: by definition all orthophotos are essentially the same. As Michalos says:

“It is submitted that the decision in when to take the photograph and the angle to take it from, particularly in a fast moving event such as a cock fight, would require sufficient skill as to establish originality. This can be distinguished from a photograph where the relative positions of object do not form part of the protectable originality, such as buildings in relation to each other or the permanent setting of a monument.” (my emphasis)

So all three judges in Bauman v Fussell establish that there is no infringement in an activity such as tracing from aerial imagery. v Rodney Fitch & Company

Fast-forward to the 21st century and we have another very relevant case – and one which is reasonably well documented on the Internet. It’s a particularly interesting case.

Fitch had designed the branding and website for For this he needed photographs of antiques; so he scanned them from Miller’s antiques encyclopaedia. At the same time, he also produced vectorised tracings of various of these antiques, for use in logos and business cards. claimed that Fitch’s work was an infringement of Miller’s copyright and therefore not usable, so sued for its money back.

The photographs themselves, of these 3D objects, were judged to be copyrightable and therefore Fitch had infringed. Michalos again:

“It was held in respect of photographs of antiques designed to bring out particular features of the objects, that the positioning of the object, the angle, the lighting, focus were all matters of aesthetic or commercial judgment sufficient for copyright to subsist.”

So the photographs themselves are copyrighted. This is no great surprise. But the judgement goes on to say that tracing from these photographs, as Fitch did to create logos, does not infringe this copyright.

“I do not consider that it can seriously be contended that these logos represented an infringement of the copyright in the photographs in the Encyclopaedia. The simple photograph of an object, however carefully lit and angled or focused the photograph may be, and however skilful and careful the photographer may have been, is not really carried through, to any significant extent whatever, into these logos.”

Or as Michalos summarises it, “Where photographs of antiques were copied by tracing the outline of the object and simplifying the object to an outline, there was no infringement as what was original in the original photographs (such as lighting, angle and focussing) were not reproduced.”

So this second case backs up Bauman v Fussell: tracing from a photograph, where that photograph is a faithful reproduction of a 3D object, does not infringe the copyright in the photograph.

Ets-Hokin v Skyy Spirits

This concept is given a helpful name, “thin copyright”, in the US case of Ets-Hokin v Skyy Spirits.

Skyy commissioned photographer Joshua Ets-Hokin to photograph their vodka bottle. They then chose not to use his photos, but rather, went with similar images taken later by other photographers. There was no suggestion that Skyy had physically reproduced Ets-Hokin’s actual photographs. Nonetheless Ets-Hokin sued, claiming that the new photographs were so similar that they amounted to an infringement of copyright. The Court disagreed:

“Though the Ets-Hokin and Skyy photographs are indeed similar, their similarity is inevitable, given the shared concept, or idea, of photographing a Skyy bottle.  When we apply the limiting doctrines, subtracting the unoriginal elements, Ets-Hokin is left with only a “thin” copyright, which protects only against virtual identical copying.”

So let us apply this statement to tracing from aerial photography. If we subtract the unoriginal elements – i.e. the actual lie of the land which is being photographed – then the “thin copyright” applies solely to the image itself. This is protected against “virtual identical copying”, but there is no restriction on tracing the unoriginal elements.

Michalos summarises:

“It is submitted that for a non-original photograph of a mundane 3-D object taken “straight-on” which is intended to be an accurate copy of that object, any copyright that subsists is a “thin” copyright which is only infringed by actual reproduction of that image…

“It is the work labour and skill in creating that image which is protected… The ‘thin’ copyright conferred on photographs intended to reproduce 3-D objects should protect only the work labour and skill and not the photograph’s subject.”

Weetman v Baldwin

A fourth case, heard in British Columbia (Canada) in 2001, is not a direct analogy but is nonetheless relevant as the only one to deal explicitly with aerial photography.

Weetman produced a map principally from tracing over aerial photography, but also by adding his own knowledge and using his own specially-written software. Baldwin copied this map – but didn’t realise that Weetman had included an Easter egg (a misspelt lake name).

When Weetman sued, Baldwin claimed in defence that the information was factual and therefore not subject to copyright. Weetman won, with Judge Romilly specifically citing:

“trails, roads and other features detailed with a precision and an accuracy not previously attained by other mapmakers of the region in question, which was facilitated by a particular process pioneered by a mapmaker”

Weetman’s actual tracing work from the aerial photographs, which Michalos calls “interpretation of those sources”, is therefore deemed copyright-worthy  – i.e. the act of deciding “that’s a railway” and “that’s a tree”. (Comparable cases agree, such as Meshwerks v Toyota.) But there is, again, no suggestion that any copyright in the aerial photographs is involved.

(See also this overview.)

Case law – summary

Using two examples from Britain, one from Canada and one from the US, we can see clearly that there is no infringement of copyright in tracing features from a copyrighted aerial photograph. We can close our consideration of case law with this, from the oft-quoted House of Lords decision on Ladbroke v William Hill in 1964:

“Whether a part is substantial must be decided by its quality rather than its quantity. The reproduction of a part which by itself has no originality will not normally be a substantial part of the copyright and therefore will not be protected. For that which would not attract copyright except by reason of its collocation will, when robbed of that collocation, not be a substantial part of the copyright and therefore the courts will not hold its reproduction to be an infringement. It is this, I think, which is meant by one or two judicial observations that ‘there is no copyright’ in some unoriginal part of a whole that is copyright.”

So if an OpenStreetMap contributor were to purchase a printed aerial photograph, trace over it, and then upload the result to OSM as their copyrighted work (under the standard OSM licence), it would not be an infringement of copyright.

But what if, for example, the Google Maps API were used? Yahoo!, of course, has for a long time given permission for OSM users to trace over its imagery. Google’s imagery is better for large parts of the world, and there can be few OSM contributors who haven’t wished that it was available in a similar way.

It is clear from the four cases examined above that there would be no infringement of copyright. Copyright, though, is not everything. Use of Google’s aerial imagery is only by permission of their Terms of Service. Do these terms permit use?

The Google Maps API Terms of Service

The primary document is the Terms of Service for the Google Maps API, the service which would be used to provide the aerial imagery for tracing. (The Google Maps Terms also apply, but 1.4 in the API terms explains that “the Maps APIs Terms will take precedence”.)

In 7.1, Google provides two crucial definitions:

“(b) ‘Content’ means any content provided through the Service (whether created by Google or its third party licensors), including map and terrain data, photographic imagery, traffic data, or any other content.
“(d) ‘Your Content’ means any content that you provide in your Maps API Implementation, including data, images, video, or software. Your Content does not include the Content.”

We must read these definitions in the light of the case law above.

It tells us the ‘Content’ cannot include the lie of the land as depicted in the aerial photography. ‘Content’ is defined as “created by Google or its third party licensors” – i.e. the aerial photography providers. The four cases are unambiguous that the material which we are tracing is not created by Google or its third party licensors; otherwise a copyright would exist in it.

(Compare with the street name and geometry data in Google Maps’ own map tiles, created from TeleAtlas data. Mason v Montgomery Data explains that, even in the US, such data deserves copyright protection for the “mapmaker’s selection of sources, interpretation of those sources, discretion in reconciling inconsistencies among the sources”. Therefore this is indeed “Content… created by Google or its third party licensors”.)

Therefore a section like 7.3b, which explains that “Certain Content is provided under license from third parties… and is subject to copyright and other intellectual property rights”, does not apply to tracing from aerial photography, since this is not an act restricted by copyright.

Similarly, 10.2 forbids you to “copy, translate, modify, create a derivative work of, or publicly display any Content or any part thereof (for example, the following are prohibited: (a) creating server-side modification of map tiles; and (b) stitching multiple static map images together to display a map that is larger than permitted in the Maps APIs Documentation)”. Again, we are not creating a derivative work, modifying, translating or copying the Content by tracing features; we are creating an independent work. The two examples chosen by Google, however, are clearly copyrighted derivatives.

Nor does 10.6 apply. It says that you may not “use the Service in a manner that gives you or any other person access to mass downloads or bulk feeds of any Content, including but not limited to numerical latitude or longitude coordinates, imagery, and visible map data”. Again, this simply isn’t an issue, since tracings are not feeds or downloads of Google-licensed Content.

Other documents

The Terms & Conditions for Google Maps (i.e. not exclusively the API) define Content similarly (section 4): “a variety of content including but not limited to photographic imagery, map and terrain data, business listings, reviews, traffic, and other related information provided by Google, its licensors, and its users”.

The “provided by” here is slightly more all-encompassing than the “created by” in the API ToS. Nonetheless, the case law we have considered makes it clear that it still does not restrict tracing. We do, of course, also know that the API terms would take precedence in the case of a website using the API to help you trace over imagery.

Other restrictions, such as 2b (”you must not copy, translate, modify, or make derivative works of the Content or any part thereof”) and 2e (”mass downloads or bulk feeds of any Content “) are the same as what we have seen in the API terms with no significant changes of wording.

Google also provides a FAQ for the Google Maps API. In it they explain sites on which they don’t permit the API to be used:

“There are some uses of the Google Maps APIs that we just don’t want to see: maps that identify the places to buy illegal drugs in a city, for instance, or any other illegal activity. We also respect people’s privacy, so the Google Maps APIs shouldn’t be used to identify private information about individuals.”

Again, none of these examples could in any way be construed as forbidding tracing from the aerial photography or any similar activity.


We can conclude:

- Case law firmly establishes that tracing from aerial photography is not an infringement of any copyright in that imagery.
- Read in this context, the Terms of Service for the most popular aerial photography API, that provided by Google, do not prohibit such tracing.

With this in mind, we can see why Google appears happy for the high-profile Wikimapia to continue to exist – even though Google is hardly shy in sending take-down notices to sites that they believe contravene their terms of service.

So, should you start tracing from Google aerial imagery, and upload the results to OpenStreetMap? You could – but you shouldn’t. OSM is by its nature very cautious about sources: the point of the whole project is to provide a “clean room” map of the world unencumbered by traditional copyright. You should not endanger such a terrific project just on the say-so of one chap with a blog.

You should, however, take this say-so, and investigate further. Are there other sites or publications, like Wikimapia, which encourage tracing or other means of obtaining factual information from imagery, and which have also been left alone? (I can think of one superb printed map, for example, which acknowledges that “Google Earth and Panoramio are not only invaluable sources, they have practically revolutionised the work of map publishers”.) And can Google or other aerial imagery providers confirm this reading – or otherwise?

It matters. The ability to trace from high-quality satellite imagery will not hurt the providers of this imagery. But it will greatly aid many, many mapping projects, and the applications – from hobbyist to humanitarian – that benefit from the data they create.

23 Responses to “Aerial photography, cock fighting and vodka bottles”

  1. Thomas Wood says:

    You’ve ignored the issue of rectification – can it be deemed a creative work by matching and interpreting the photo to control points on the ground?

    If it was the case, it would most likely be possible to re-rectify the images ourselves to overcome this issue.

  2. ict student says:

    As a Nokia lawyer put it: “Even if you know you’re not breaking any patents, you will be sued for patent infringement by a competitor once your global market share exceeds 5 %. We weren’t prepared and only after millions of dollars and years of proceedings, the case was withdrawn.”

  3. Duncan MacGregor says:

    I think I agree with Thomas on this, there are two bits of data associated with each photograph—it’s precise coverage and its relation to others in the dataset—which could well be deemed a creative work, and I think you’d need to be careful to discard such information and recreate it yourself as part of the tracing process.

  4. Andrew Denny says:

    There was a true case recently that two identical twins were acquitted of murder because while it was acknowledged that one of them committed the murder, the other one was definitely innocent – and the prosecution could not prove which one did it.

    Seems to me that mappers could argue the same. One mapping service or another *must* have provided the data, but because the data is factual and could have been provided by any mapping service, how can the prosecution prove which map it came from? All the tracer has to do is exercise his right to stay silent!

  5. I think Thomas Wood has a good point. The original uncorrected image, being taken pretty much automatically and with no creative input, is probably not protected by copyright under English law. But the rectified version, having had an additional skilled input, very likely is. However, to make use of this would probably require getting hold of the original raw imagery rather than just trying to ‘un-transform’ it. (Yes, logically the result is the same but legally you’d be making a transformative work from a copyright image.)

    Andrew: Um, I don’t think you’d get very far with that! For starters, this is why mapping services include dummy features. Also, the privilege against self-incrimination relates to criminal offences, not civil wrongs, and furthermore it specifically does not apply to intellectual property claims (via section 72 of the Supreme Court Act 1981).

  6. Richard says:

    I didn’t think rectification was particularly significant, but if you insist…!

    Certainly rectification doesn’t merit protection through “originality” (the crux of Bauman v Fussell) or creativity. It may perhaps merit some protection through “sweat of the brow”.

    In the US, though, I think it’s pretty cast-iron that it doesn’t, because of the idea-expression merger doctrine (summary). This is a well-established principle in US law that says if there’s a very limited number of ways of expressing a given idea, it’s not copyrightable. This is certainly true of rectification. There is only one possible result of rectification: a direct correlation between ground features and the imagery, to the given scale and projection. (You could of course misrectify things but I’m pretty sure that sporadic, accidental incompetence doesn’t merit copyright protection!)

    Michalos agrees:

    “A standard aerial photograph will have no uniform scale and will distort the image due to perspective and the topography of the land. An orthophoto corrects these distortions and produces an image with a uniform scale from which distances can be measured as they could be from a scale map. An orthophoto is intended to reproduce as accurately as possible various features of its subject topography.

    “In some ways therefore orthophotos entirely lack originality as they are mere copies of geographical features… Protection does not extend to the geographical features represented but only to the orthophotographer’s selection of geographical data, presentation, use of symbols and annotation and any other artistic additions.”

    I would consider that “presentation” would include colour balance/matching and potentially sharpening, so the images themselves do still deserve copyright protection. But these factors don’t carry through in any significant way to the tracings.

    In the UK, as ever, the law is less clear-cut than in the US. It might even be possible to argue that the EU database directive has some relevance to a database of control points, and that these do not meet the required standard for protection (”no criterion other than originality in the sense of the author’s intellectual creation”). But you could go insane trying to figure out the database directive.

    (I should, of course, declare a slight interest in this in that my OS New Popular Edition scans, which I spent many evenings rectifying, are offered for unrestricted tracing. I don’t see how anything that I’ve done has added anything to the map data expressed in the scans, although it has significantly improved the quality of the scans themselves.)

  7. blackadder says:

    An outstanding piece of work Richard. Well argued and compelling. Lets hope it pushes forward the informed debate.

  8. An interesting idea might be to set up a new project, most likely using some of the technology from OSM. The aim of this project would be to create and maintain an open map as an “alternative” to OSM. The project would then systematically map one city or perhaps even country that Google has good imagery for but, perhaps, OSM hasn’t really covered yet. The project would then publicise their data and the source of it and try to get people to do interesting things with it. The hope then would be that you got sued. I guess if you didn’t get sued there would be no testing of the legal issues but, still, might be interesting to try out.

    For the record, I have no intention of doing it myself ;-)

  9. Rather than the full alternative project I was going to suggest that this legal information could be used to allow you to happily mash OSM and google-tracings for your own work (but the google-tracings) never hit OSM’s servers.
    You would only have to credit OSM (as partial data provider), but you get a map where OSM coverage isn’t full.

  10. Gustav F says:

    A very interesting read. I hope this can be one small step to move OpenStreetMap from a overly cautious approach to sources to a cautious approach.

    Could there be any database rights in the images?

  11. Mikel Maron says:

    Well done Richard.

    In order to test this, in a safe way, something like John’s idea seems the way forward. An independent, clean experiment designed to receive a take down notice, or get sued (or perhaps not). Would want to see this carefully constructed so as not to draw too much energy from OSM, the city/country restriction would help. Who’s up for this?

  12. Ben says:

    A really interesting read and very well thought out.

    However, there’s one thing bugging me. Google have database rights under UK law regarding access to their databases and servers. If they don’t like what you’re doing, they simply change their Terms of Service to include something as simple as “You must not access our servers or databases for the purpose of tracing provided data” and it’s game over.

    Regardless of the data held and who owns the copyright and derived works (that’s another argument), they simply are the gatekeeper – because it’s their servers and they’ve paid for them – and can therefore state who can access the data and what they can do with it at will.

  13. Richard says:

    Ben – you’re quite right, of course; they could do that at any time.

    But if Google did want to change their terms to prevent this, wouldn’t they have done it already to stymie Wikimapia? And I guess there’d be a lot of collateral damage – plenty of people have created “innocent” mashup sites including some tracing, even if it’s just identifying, say, their favourite pubs by reference to intersections.

  14. Gustav F says:

    It is also worth noting that Google Earth has a tracing tool included. This, at the very least, indicates that some sort of tracing is allowed, and even ecouraged. Nowhere in the terms, does it say that Google have any rights whatever to these traces.

  15. Eugene says:

    Excellent piece! While reading through the article I ticked down mental notes about the Google TOS and Wikimapia. I was pretty impressed that you mentioned those two points later on in the article and was no longer left with unanswered questions!

  16. Roy says:

    Ben – yes, Google could change the ToS in the future. But the question remains as to whether we should be legally tracing in the meantime…

    Surely a future change in ToS could not be applied retrospectively..?

  17. John says:

    ToS, unlike laws, can’t be applied retroactively because you would have no way to disagree and opt out and there is already case law against click through agreements deeming them less than legal and in this case you don’t even have to agree to ToS to use their services.

  18. [...] by saying that tracing from their imagery is ok - just like Yahoo have done. Several posts ago, I looked into the legalities of this and concluded there’s nothing in law stopping them from doing so. It’s entirely their [...]

  19. Hermann says:

    There’s one further issue not mentioned until now I think: competition law! If you traced Google’s aerial images to create maps, you would do it as a direct competitor in the market of map and map data providers. Google’s maps are placed right next (or even on top) of the aerial images. I don’t think you could claim to use “only” the images and then make a competing product of it and offer it for a lower price – completely Free as opposed to an ad-model.

  20. P says:

    Firstly, I’ll admit that I’ve not read this *very* closely, and that this comment may therefore be a waste of everyone’s time. However…

    Your reading of the legalities of tracing and rectification seem reasonable (IANAL). The query that came to mind, though, concerns the part in the ToS (10.6) that forbids ‘mass downloads of coordinates’. While the location of geographical features is a fact, and you make a case that tracing the _relative_ position of features is permissible, could they argue that since the images *are* positioned accurately, when tracing from them each tracing point is a coordinate download? (I feel that I may not have expressed myself very well here, but don’t have much time to rephrase – sorry.)

    I would hope that they wouldn’t, and as you show, in some cases they haven’t – but wikimapia, for example, retains the Google logo on the imagery, whereas OSM wouldn’t (and as pointed out above, OSM is a ‘competitor’).

    It would be really nice, though, to get a definitive statement of their views. It’s good to see that your Data Liberation suggestion is far and away the most popular.

  21. Robert says:

    Another case which touched on originality was Bridgeman Art Library Ltd v Corel Corp 25 Fed Supp 421 (1999). Applying UK law, the judge found that as photographs of public-domain works of art claimed to be as accurate as possible copies of those works, they lacked originality under UK law and so could not be protected.

    The Fitch case showed that lighting of a scene could generate copyright in a photographic image. Where this leaves aerial photographs is unclear. Given the effort in taking them and rectifying them, they will certainly be copyright in the US.

    There may also be a separate database right in aerial photographs in respect of the coordinates of the points represented.

    Fitch confirmed that human extraction and interpretation of information from a photograph does not infringe copyright. The greater the transformation wrought by the human, the less open the process is to challenge. Validating a derived map by walking the streets and interpreting features creates a new artistic work with its own copyright.

    I would be dubious about automating the transformation process as this could infringe the database right.

  22. I would like to find other people who are interested and start to trace the geoeye sat data into its own dedicated server. Basically we would use osm software to create a separate layer for the traced data. this would not be compatible with osm and the license would restrict its combining with any other layer. it could only be used as a comparison layer against osm.

    basically I would like to trace features and contours and polygons from the sat imagery tag them and then compare that against osm. we could then find missing streets and areas that are shifted over.


  23. [...] have copyright?  If so, what are the risks? Many different perspectives tangle with one another. One comprehensive blog posts posits that it may be alright under copyright, but not socially acceptable in open mapping circles [...]

Leave a Reply