Slashdot Mirror
Microsoft Patents "Cartoon Face Generation"
theodp writes "The latest round of patents granted by the USPTO included one for Cartoon Face Generation, an invention which Microsoft explains 'generates an attractive cartoon face or graphic of a user's facial image'. Microsoft adds, 'The style of cartoon face achieved resembles the likeness of the user more than cartoons generated by conventional vector-based cartooning techniques. The cartoon faces thus achieved provide an attractive facial appearance and thus have wide applicability in art, gaming, and messaging applications in which a pleasing degree of realism is desirable without exaggerated comedy or caricature.' A Microsoft Research Face SDK Beta is available. Hey, too bad Microsoft didn't have this technology when they generated Bob from Ralphie!"
I don't have the link for MS tech but if you surf long enough you'll find some others already offering to make your avatar photo cartoonish with a pretty animated demo.
---- MISSING MISCELLANEOUS DATA SEGMENT --- [sigdash] trolololol
Has the definition of prior art changed? otherwise how are you claiming these completely different techniques are prior art? from my search of definition it is still "constitutes all information that has been made available to the public in any form before a given date that might be relevant to a patent's claims of originality.".
Well, $SUBJECT says it all, really.
Sigh. When will those plagues disappear.
Before you waste your time downloading the SDK note that it is for windows phone 7 only. I know they did sell a few, but I've never met anyone who bought one.
I am sure I have seen those "photographic" type booths in service stations that do cartoony caricatures of people - are these not prior art? and what about avatars for forums - does Microsoft now own the patent on these? What about services like befunky? http://www.befunky.com are they now going to have to pay MS a license fee for continuing to provide services?
Microsoft patented the technique. Now, clam down...
This actually seems like a worthwhile invention.
Sadly, like most recent Microsoft "innovations", it's a copy of other people's work being patented in an effort to block competition (likely including the original developers) and/or extort money from people with a real product.
http://www.hongkiat.com/blog/11-sites-to-create-cartoon-characters-of-yourself/
How about just using a photograph instead?
Microsoft Song Smith is useful to obtain amusing Youtube videos. See for yourself: http://it.bing.com/videos/search?q=songsmith+vs&qs=n&form=QBVR&pq=songsmith+vs&sc=0-15&sp=-1&sk=
Burger King did this via a website with the Simpson's years ago. It was called SimpsonizeMe or something like that. Also, the Wii U does it now to create Mii's.
There is a war going on for your mind.
Presumably it's so that the next-generation Kinect will be able to assemble user avatar faces automatically using its camera. Trying to assemble a good face using sliders and decals, as in current implementations on the Wii and Xbox, is a pain in the arse, so having it knock together a reasonable starting point using 3D scanning and face recognition would be nice..
Doesn't the 3DS already do this, in fact?
Obligatory: this gives MS a patent on an approach to cartoon face generation, not the entire idea. I've not read the claims so they could all be obvious, previously existing and unpatentable waffle but it history suggests we'll have a raft of comments about "blah blah did this already" without reading the patent.
No kidding!!! What do you say at this point?
It doesn't even have to be more or less realistic, so long as it uses a different approach. The text about realism is to establish what their invention is supposed to accomplish, which is important in establishing that something is worth a patent in the first place but not part of their legal power.
No kidding!!! What do you say at this point?
I haven't learned how to draw in 40 years and I certainly do not enjoy drawing.
I guess you didn't actually read the reference pages you berating. From the Rotoscoping wikipedia page:
In the mid-1990s, Bob Sabiston, an animator and computer scientist veteran of the MIT Media Lab, developed a computer-assisted "interpolated rotoscoping" process which he used to make his award-winning short film "Snack and Drink". Director Richard Linklater subsequently employed Sabiston's artistry and his proprietary Rotoshop software in the full-length feature films Waking Life (2001) and A Scanner Darkly (2006).[7] Linklater licensed the same proprietary rotoscoping process for the look of both films. Linklater is the first director to use digital rotoscoping to create an entire feature film. Additionally, a 2005–08 advertising campaign by Charles Schwab uses Sabiston's rotoscoping work for a series of television spots, under the tagline "Talk to Chuck".
So, even though Rotoscoping was first a manual technique done in 1915 and patented in 1917 (tracing live action frames that are projected onto the back of a frosted glass panel), that process moved to using a computer, and then was automated within the Rotoshop software.
There's even a patent from 1994 mentioned: US Patent 6,061,462 http://www.google.com/patents?vid=6061462 for a digital rotoscoping process (that's a separate work from the Rotoshop software).
These may all be listed in Microsoft's patent - I haven't read it - but they certainly seem related, if not prior art.
It's only prior art if they did things the same way. God help us if the USPTO's standard of an invention becomes "does the same thing as something else".
No kidding!!! What do you say at this point?
Already do this. The device takes a picture through it's front facing camera and automatically generates a "Mii" with similar features. I'm sure somebody else can find similar software that did this even before.
Yeah, just like there's a thin line between crazy and genius. Doesn't make every crazy person a genius though, just like there is very little beauty involved with these Microsoft Research projects. I'm sure they do plenty of awesome stuff, but I don't think the notion of automating creative processes can lead to anything good.
Pretty good is actually pretty bad.
You will of course let me know when they patent the "Face Palm"
Build a Man a Fire, and He'll Be Warm for a Day. Set a Man on Fire, and He'll Be Warm for the Rest of His Life.
Isn't Song Smith an extensions of the kind of DirectMusic research that gave us videogame soundtracks that respond to the current action? That's been pretty useful.
Just because you see a cute consumer-facing endpoint that you don't find useful doesn't mean the entire project was "automate composing music".
No kidding!!! What do you say at this point?
Unfortunately, that's what most software patents seem to be like. Look at the ridiculous 'rubber-band' effect for indicating you've reached the end of scrolling in iOS. The patent office approved it and I believe ruled against Samsung in using it. I can almost guarantee the code used to implement it was not the same. That is patenting the idea, not the implementation. The same goes for most other examples of software patents. You can't work around them by doing something differently, you must actually do something different, or so it would appear.
Indeed, it's disgraceful which software patents have been approved. That is an issue. If MS have generated a new algorithm in this case then that should, in principle, be patentable. And if the algorithm is described previously elsewhere, it shouldn't be.
No kidding!!! What do you say at this point?
Adaptive music in video games has absolutely nothing to do with what Song Smith attempts to do. This is Song Smith's algorithm: it analyses a melody (recorded by voice), then does a bad guess at the suggested tonal scale and harmonies (which is basically impossible to get right, because the data just isn't there) and then uses these guesses to accompany your singing with chords, played by a cheap sounding midi rhythm machine. The result can barely be called music at all and is completely unrelated to anything that happens during the actual process of creating music.
Now I'm somewhat of a fundamentalist on this subject, as I believe that any from of generative music is a mistake. In my opinion music, like any art form, is about communicating ideas. They have to be from one human to the next, by definition. Just like animals or plants, a computer doesn't have ideas it can communicate, therefore it is unable to ever create music. It might, at best, automate the task of creating variations of human compositions. That might be somewhat useful as long as a human is in control. But as the human factor decreases, the result will move away from music and toward random sound. And even if you like listening to Yoko Ono, sound does not equal music.
Pretty good is actually pretty bad.
where Microsoft wanted to automate the "tedious chore of composing music"
[citation needed], I can't find this line anywhere.
I was also under the impression specific algorithms were not actually patentable.
The patent isn't on automatic generation of the caricatures, it's on their particular algorithm for doing so.
-jcr
The only title of honor that a tyrant can grant is "Enemy of the State."
I remember people doing something like this back in about 1990 using very similar techniques - locating the nostrils, the eyes, the curve of the mouth to generate a real-time animation. Back then, this allowed people to transmit a cartoon over the existing phone network, allowing the deaf to lip-read. Back then, this was a clever idea.
'generates an attractive cartoon face OR graphic of a user's facial image'
So it can either be attractive OR look like you.
Now, I admit I can't be bothered to actually read the patent, but does it describe the algorithms for the method? Or is it just like the linked picture? (Photograph->"magic box"->cartoon picture)
I surely hope not, because then the patent system is useless. But it also would mean that I could patent any input->magic box->output method I can dream of.
I would argue that whoever creates the music generation system is the one doing the expression in this instance, inasmuch as the design of the algorithm is determined by the author's sense of what is musical. Is this really any less expressive in principle than any other piece of algorithmic art?
I don't think we have to worry about algorithmic music taking over creative expression at any rate. By necessity, if the prevalence of such a system caused an objectionable reduction in expression in music, non-generative music genres would become more popular amoungst musicians and listeners.
No kidding!!! What do you say at this point?
So you shouldn't be allowed to patent, say, a car that does one trillion miles to the gallon, or is made out of four pounds of plastic, because at the end of the day it does the same thing?
No kidding!!! What do you say at this point?
I mean algorithm in the more general sense here of a process by which something is done. That was a poor choice on my part. (Yes, I'm the kind of nerd who would describe his process for sorting socks as an algorithm.)
No kidding!!! What do you say at this point?
Obligatory: this gives MS a patent on an approach to cartoon face generation, not the entire idea. I've not read the claims so they could all be obvious, previously existing and unpatentable waffle but it history suggests we'll have a raft of comments about "blah blah did this already" without reading the patent.
In theory you're right. In reality, patents are so broadly written that they actually do cover the entire idea.
Slashdot social media options: AIM, ICQ, Yahoo, Jabber and Mobile Text. Why no MySpace?
It will probably look like this! http://en.wikipedia.org/wiki/File:MsComicChat.png
Good ol' chat.
no comment
Songsmith is just a cheap ($30) little program obviously aimed at the casual or party game market to allow people to experience what it is like to be a singer with a backing group. It is not intended for serious use or to replace real musicians. If that had been the intention then they would have priced it much higher to compete with Band-in-a-box.
Similarly, this cartoon drawing system is not part of some plot to put real artists out of work. It is designed to incorporate stylised caricatures within games or for generating avatars.
The whole f**king world has gone stark staring mad.
No, you shouldn't, because that's not an invention to be patented. The car already exists as prior art. You would patent the engine design used to make it so efficient. You would patent the material composition or internal cellular structure used to make it so light yet sufficiently strong. You would patent any special, novel manufacturing techniques used to produce either of those. The car manufactured using those patents is covered under copyright, or potentially trademark.
First of all the bottom line - patents suck shit. I'll try to explain, although there's so much to tell.
The only positive thing about patents was the original idea that inventors need a tool to protect themselves from copycats. A pretty good idea, but then came other people and turned the idea into a steaming pile of shit, as we often do with everything.
There's a number of complexities that people do not take into account when they file a patent:
1. It's freaking expensive because at some point you have to work with patent attorneys. You can submit provisional yourself, but you must hire patent attorney in PCT or National phases and it's unlikely that you can use proper language that the patent attorney will use, so you risk not having your provisional be considered a priority document (which sets the priority date, when your invention is considered to be created).
2. You'll spend a lot of time on it. Time that could better be spent on building and improving your product.
3. You'd have to reveal all details of your invention to everyone. What if it's hard or even impossible to prove if anyone infringes it later? Patent attorneys will never tell you not to patent it (see #2 above).
4. There's a very good chance that there's something like you invented already, so you won't get a patent despite all the efforts. In other words, unless your invention is abso-fucking-lutely genius, you're doomed to fail on "novelty" or "inventive step" requirements. Almost no doubt about it. Patent examiners are not the sharpest pencils in the box. Read this if you don't believe it: http://www.cbsnews.com/8301-505124_162-57581797/the-letter-to-the-patent-office-you-have-to-read/
5. Let say you got your patent granted. You think you just skipped to the "profit" bullet? Ha! In order to benefit from your patent you need someone who infringes it and benefits enough from it to pay you damages. Not a small startup, they won't be able to pay you damages so no sane lawyer will take your case. If the infringing party is rich enough they still need to benefit enough from the infringing product or again no sane jury will give you the damages. Take into account that the bigger the infringer is, the stronger they'll fight you. They will bully and starve you.
6. IP cases take a very long time and cost millions of dollars, and their lawyers will drag you all the way through Inter Party Review at USPTO (where they'll reexamine your patent, another year and $500K) and it's quite possible that they will find some freaking conference presentation in Chinese that's close enough to your invention. Nobody will give a shit that you don't understand Chinese and never been to that conference.
7. In the US you could go to ITC (International Trade Commission) instead of a district court, but that only works when the infringing party is a foreign company and they import the infringing products into the US. ITC also doesn't give you damages, they can only stop the import of infringing products.
All the above taken from my very bitter experience in managing about 40 patent apps in different country.
JUST DON'T DO IT
They can't cover the idea because the idea has already been implemented multiple times before through other mechanisms. There is not only prior art, there is patented prior art. All they can do is patent their implementation of that concept, in however broad of language they can.
Perhaps you should start to learn. You might enjoy it.
So instead of drawing a cartoon face yourself, which is something everyone can do...
Not everyone can draw a plausible caricature of themselves. Try it sometime.
Long time back, Microsoft Research China released an IM client for low bandwidth video chat at dial-up speeds. (less than a 15K modem)
It's just like the horror that is Microsoft Song Smith, where Microsoft wanted to automate the "tedious chore of composing music" by letting a computer generate tunes instead.
Not stupid at all.
The most obvious practical application would be in video games, where synching art, animation, music, audio and visual effects to the player's actions without odd incongruities and obvious repetition is a major problem and expense.
I find I get better performance appraisals if I don't go to work with one of my 'Wicked Witch of the West' socks and one of my sparkly gold ones.
(1) Not everyone can draw a cartoon version of themselves that would be aesthetically pleasing.
(2) They animate the face to match captured motion.
Now, granted, for (2) they could instead analyze a cartoon face drawn by a human and figure out how to animate it...but I'm guessing you would think THAT was a waste of time as well.
I used to work in creating educational software. Having this technology would have been a boon for some of the stuff we were planning.
You sure are down on what, I think, are interesting research projects. I don't think anyone could have ever been under the impression that Song Smith would or was intended to replace actual composers.
Owing to laziness and slow turnover I own about fifteen brands of socks, all of which have different thicknesses. If I don't match them up, I walk lop-sided.
No kidding!!! What do you say at this point?
Nobody said it did.
No kidding!!! What do you say at this point?
It take to patentee to file the patent but then the USPTO to grant the patent and those involved in granting this patent need to be fired for their incompetence. Characachers have been around for a long time and I have seen booths at arcades and Dave & Busters that do exactly this since the late 90s. This is beyond obvious prior art.
So, now, mathematics are patentable.
Actually, that isn't quite right. The patent itself would cover the entire business case to be made for such an invention, that is the whole car, or any possible way in which the engine might be used.
The INDEPENDENT CLAIMS in the patent would include all of the things you listed. A solid, enforceable patent will have independent claims specific enough to exclude as much prior art as possible, but that doesn't mean any prior art would invalidate the patent. Your super efficient engine design, for example, could include components like a fuel injector that people have been putting into engine designs for decades, but that doesn't mean it isn't patentable. Your patent should at least reference the prior art, however.
Here is where lawyers like to get general. A super efficient engine is novel, but it is a useless novelty, unless you put it into a machine, such as a car, that performs a useful function. That is where DEPENDENT CLAIMS come into play. You want your dependent claims to be big, broad, sweeping and general to cover as many business cases as possible. This process invariably invokes big giant buckets of prior art. All of which you need to reference, of course.
I think it is high time for /. to stop cherry-picking lawyer-speak from patent applications. If your invention builds off of previous inventions, like all inventions do, and the patent application is written correctly, then there is always going to be some set of broad, sweeping claims which, when quoted, will send the comment section into a rantfest.
I'm genuinely interested: how could these sort of research projects be of any help at all with anything? Isn't it this simple: if you want a drawing but can't draw one, hire an artist. If you want music but can't make music, get a musician. It's not expensive nor complicated to get some help from an expert on these things.
Software as a tool to empower people who create is great, but software that actually generates creative content itself is completely useless in my opinion.
Pretty good is actually pretty bad.
Can I patent my algorithm for automatically generating unique patents?
The G
Q: So, girly, you like roller skatin'?
|
| --> A: Yes --> Yeah, everybody loves roller skatin'..
|
| --> A: No --> Yeah, everybody loves roller skatin'...
.
Prisencolinensinainciusol. Ol Rait!
Sftware and business method patents simply should not exist. Period.
Inneficiency is the enemy of laziness. It doesn't take very long at all to make sure new purchases match your existing sock stockpile, you probably use more time every week seeking out matching socks than it would have taken to just buy same-brand socks. Your application for a Laziness merit badge is hearby DENIED.
--- Most topics have many sides worth arguing, allow me to take one opposite you.
NO. No, no, no. Ideas are *explicitly* barred from being patentable in the US constitution, along with mathematics, in either case the chilling effect on innovation was deemed far greater than the benefits it might provide. Remember patents and copyright explicitly exist to benefit society, the benefit to the creator is simply the carrot dangled to incentivize them to create.
It's the specific design that can be patented - in physical objects that corresponds roughly to blueprints+description, the implementation (metal, plastic, etc) is irrelevant unless stated. If we were being consistent then in software it would be something like pseudocode instead of blueprints, and the patent would extend to any language-specific implementation of that design. Of course two things would then become immediately obvious in almost all software patents:
(1) the proces is mathematical, and hence ineligible for patent protection
(2) Due to the flexibility of software development it's trivially easy to design around almost any specific design to yield the same results. (plenty of physical-device inventors get shafted by the same problem, but there are generally far fewer physical variations possible so an inventor has at least a hope of listing all the major ones in his patent.)
--- Most topics have many sides worth arguing, allow me to take one opposite you.
Correction, you would patent the PROCESS of how to make the material composition or internal cellular structure. Cant patent the actual material.
Good-bye
get cancelled?
To ensure perfect aim, shoot first and call whatever you hit the target
The patent isn't on automatic generation of the caricatures, it's on their particular algorithm for doing so.
-jcr
To find out that the mo-cap I've done with standard web cams to create game sprites and do digital lip-syncing since the mid 90s isn't infringing just because I can use a Kinect for an input will cost millions...
Ohh yeah, isn't it nice to be in a country that is moving away from this insanity?
http://www.iitp.org.nz/newsletter/article/430
http://no.softwarepatents.org.nz/
Even 20 years ago I would probably use the term "smart" rather than "clever". Using the location of features to represent a face is pretty intuitive, and although there are other concepts covered in Microsoft's patent, none of it looks especially innovative. Of course, the issue here goes beyond Microsoft and is more about the patent office.
I understand that the patent office has limited resources and a lot of requests, but they need to do a better job with the resources they have. I suggest something like the following, which forces the patent submitter to better describe why they deserve the patent ...
The submitter is required to provide an ordered list of the innovative aspects of the patent. They must describe each such "assertion" in less than 500 characters. An assertion can reference details in the patent body, but the core argument must be brief and clear. The patent office (with input from the public) starts at the top of the list to review the asserted innovations. Once two assertions are struck down, every assertion after the second rejection is implicitly rejected. This gives the submitter motivation to put the most innovative assertions at the top of the list, and it limits the amount of mundane filler a patent reviewer must sift through. All rejected assertions become like prior art for use in evaluating future patent applications.
Ann C.
Table-ized A.I.
Engines already exist, though. So do materials. It's just another method of doing the same thing, after all.
No kidding!!! What do you say at this point?
>Then explain the patent for the Hilbert-Huang Transform
Easy. Ignorant and/or corrupt patent examiners. Wouldn't even need to be bribery, the US patent office is funded in large part by the fees charged to grant a patent, thus incentivising them to grant as many patents as posible.
Honestly I'm still on the fence about the software=math argument. I'm a programmer, so certainly I know full well that all a computer can do is arithmatic operations (well, and conditional jumps - I'm not sure those technically count as math, for all that they're often incorporated into sophisticated mathematical operations), ergo all software is built out of mathematical components, but by a similar line of reasoning all physical inventions are built out of unpatentable materials and laws of physics. Nevertheless the whole is greater than the sum of it's parts, and it's the whole that is potentially patentable. I'd just as soon see software patents disappear, but I'd be satisfied if we just got rid of the 99.9% of software patents that are issued for trivial, self-evident, overbroad, nonspecific, etc,etc,etc "inventions". Those should never have been granted regardless of your position on whther software patents should exist at all.
--- Most topics have many sides worth arguing, allow me to take one opposite you.