A fitting approach?

It is not yet a trend in French courts, but it might become one. We would have to coin a name for it. In Germany, they have bifurcation, so this term is already taken. Maybe “dissociation“?

What I am talking about here is a court ruling on the validity of a patent first, before the question of infringement is addressed by the parties, even though it was raised first.

Why would a court proceed this way, whereas it may seem more time-effective to deal with all claims and counterclaims at the same time, and whereas our good friends in Düsseldorf or Mannheim generally do just the opposite, i.e. get a ruling on infringement first from the Landgericht, typically before getting a decision on validity from the Bundespatentgericht – unless a stay of proceedings is ordered by the Landgericht?

Well, because in some cases, the approach validity first – infringement second is more respectful of the defendant’s rights, and more specifically of its know how and trade secrets.

One harbinger of this possible trend may have been Quadlogic v. Enedis, already commented upon (on the merits) on this blog. On February 24, 2016, Quadlogic initiated infringement proceedings against Enedis. In July 2016, the plaintiff filed a motion with the court to obtain documents from the defendant, so as to be able to fully characterize the relevant technical features of the allegedly infringing Linky meters. Enedis resisted this motion, and the judge in charge of case management issued an order on November 17, 2016, per which “communicating these elements is premature since Enedis pledged to file an affidavit regarding the nature of the memory […] and the validity of the patent is challenged“.

The case proceeded further, with the Paris Tribunal de grande instance (TGI) declaring the patent at stake invalid on November 2, 2017, thus retrospectively confirming that forcing Enedis to produce technical documents beforehand would have been inappropriate.

This approach was taken one step further by a different section of the Paris TGI in Fittingbox v. Acep Trylive.

This case relates to French patent No. FR 2955409, directed to an “augmented reality method of integrating a virtual object into photos or videos in real time“. More particularly, the focus of the technology is the integration of eyeglasses into an image of a face. So that the face’s owner can see whether these glasses fit him or her well.

Yes they do fit perfectly well.

Fittingbox sued Acep Trylive for infringement of this patent on June 9, 2016, after performing an infringement seizure on May 11, 2016, in order to gather evidence of the alleged infringement. The seized evidence was sealed by the bailiff who conducted the seizure, probably upon request of the seized party and/or because the bailiff deemed that the evidence was sensitive in terms of know how and trade secrets.

The plaintiff filed a motion in order to be authorized to access the sealed evidence. More specifically, Fittingbox requested that an expert be appointed in order to study the source code of the software of the allegedly infringing process, and to draft a report.

On March 30, 2017, the judge in charge of case management rejected the motion, holding the following:

The parties at stake are two direct competitors in the very specific field of the virtual fitting of eyeglasses. 

Fittingbox offered to […] purchase the immaterial assets of Total Immersion, which were finally acquired by Acep Trylive, and revealing the source code would enable Fittingbox to access the know-how currently held by Acep Trylive. 

Besides, the validity of the patent relied upon by Fittingbox is seriously challenged both as far as clarity and novelty are concerned. It is put forward that the background section of the application does not mention Total Immersion’s software which was disclosed since 2008 […]. Besides, the INPI, when examining the patent application in 2010, mentioned a lack of inventive step and clarity […] and the EPO mentioned a lack of clarity when examining the European extension of the patent in suit […] and stated that “some of the objections are such that it does not seem possible to remedy them by modifying the application”. 

In view of the present circumstances, Fittingbox’ requests for unsealing should not be granted; first, the court should rule on the merits of the validity of the patent in suit, and then second, if the patent is held valid, the court will address the issue of whether the seals should be lifted in order to prove the existence of the infringement, and optionally also the extent and modalities of such lifting.

Four remarks here:

  • First, the court’s comment on the clarity of the patent will be addressed below.
  • Second, the court was certainly sensitive to the fact that the source code is bound to contain a trove of secret and extra-sensitive information. In principle, appointing an expert to examine the source code is supposed to prevent the defendant’s know how to be directly transmitted to the claimant; but the final report of the expert is then disclosed to said claimant, and it necessarily contains some relevant technical information.
  • Third, the court also took into account the context, namely the fact that both companies are direct competitors, who in particular competed to purchase a third company.
  • Fourth, the patent was perceived as potentially weak.

One problem with this fourth point is that the court relied on negative comments contained in the written opinion or substantive communications issued during prosecution in connection with the French patent itself and its European counterpart. The court was probably excessively impressed by these comments and did not realize that such comments are commonplace at the onset of examination.

Actually, I checked the EPO online register and the decision to grant the European patent was issued a couple of weeks ago, notwithstanding the examiner’s initial position that the application should be thrown out in its entirety. In fact, if I had received 1 euro each time I read a comment from an examiner similar to the one quoted above, although a patent was later granted, I would now be in a position to buy a couple of thick IP law books.

The bottom line is that the case got dissociated, if I may use this term, in order to protect the defendant’s know how and trade secrets, and the counterclaim for invalidity of the patent was examined first.

This leads us to the judgment issued on November 16, 2017. Interestingly, the court did not stick to the case management judge’s initial impression and found the patent valid.

Therefore, Acep Trylive’s counterclaim for nullity was dismissed, and the case was remitted to the judge in charge of case management to deal with the modalities of the unsealing of the seized evidence, and the request for expertise.

Whether this approach is fitting or not, as asked in the title of this post, is a matter for readers to decide.

It is certainly not ideal in terms of timeline, as it is bound to markedly increase the overall duration of the proceedings. But it may be the most fair and balanced way to proceed when the defendant’s know how and trade secrets are at stake. It should also be borne in mind that the rate of patent invalidation in France (just like in other European countries) is high, so that addressing validity first does make sense – timing constraints aside.

While we are at it, we might as well have a look at the merits of the validity ruling. After all, it is not every day that a computer-implemented process is held valid.

For those who are brave enough, here is the main claim at stake:

A method for creating a final real-time photorealistic image of a virtual object, corresponding to a real object, which is arranged on an original photo of a user, in a realistic orientation related to the position of said user, characterized in that it comprises the following steps:

– 510: detecting the presence of a placement area for the object in an original photo,

– 530: determining the position of characteristic points of the placement area for the object in the original photo, step 530 consisting in: 

– determining a similitude β to be applied to an original photo, to obtain a face, similar to a reference face, in size and orientation and

– determining the position of the exact external corner A and of the exact internal point B for each eye in the face of the original photo,

– 540: determining the 3D orientation of the face, i.e. the angles φ and ψ of  the camera having taken the photo relative to the principal plane of the placement area for the object, 

– 550: selecting the texture to be used for the virtual object, in accordance with the angle-of-view, and generating the view of the virtual object in the 3D (φ, ψ) / 2D (θ, s) position in question,

– 560: creating a first rendering by establishing a layered rendering in the correct position consistent with the position of the placement area for the object in the original photo,

– 570: obtaining the photorealistic rendering by adding overlays, referred to as semantic overlays, so as to obtain the final image.

Interestingly, it seems that the defendant did not raise any attack of patent ineligibility. What the defendant did raise is lack of clarity… which is not a ground for nullity.

But the court interpreted this line of argumentation as in fact relating to insufficiency of disclosure – which, of course, is indeed a ground for nullity.

Acep Trylive primarily relied on various objections raised during examination of the French patent in suit and of the corresponding European patent. But the court noted that the objections raised in France were overcome before grant, and that the objections raised at the EPO did not apply to the French claims, which were differently drafted.

Then, Acep Trylive argued that the claims lacked novelty over two videos available online, one from Fittingbox itself, and one from Total Immersion (remember, this is the third company that was eventually acquired by Acep Trylive).

The court held that the dates at which the videos were made available to the public were not indisputable. Too bad there is no detailed discussion of the evidence offered by the defendant regarding these dates, it would have been useful to see what was actually deemed insufficient by the court. One likely reason why the court did not dwell on this aspect is that the contents of the videos were held not to disclose the subject-matter of the main claim anyway. The first video only showed the result of the claimed method without explaining the technical details, while the second video did not show at least two of the steps of the claimed method.

The third and last validity attack was an alleged lack of inventive step. What is of interest in the court’s discussion in this respect is probably the fact that the skilled person would not have combined the various prior art references put forward by the defendant, according to the court.

The skilled person was defined by the court as a “specialist of system software and virtual fitting networks, having average knowledge in augmented reality and 3D modeling“. The technical problem at stake was to “make it possible to integrate a virtual object into photographs or videos in real time as realistically as possible“.

The closest prior art was D1, a prior Fittingbox patent, in the same technical field. Acep Trylive proposed a combination of this D1 with one or more of D2 to D6. But D3 to D6 were discarded by the court as not relating to the technical field of the skilled person. And D2 was commented upon as follows:

However, D2 exclusively deals with 3D modeling of architectural buildings and the skilled person wanting to improve realism of virtual fitting, notably spectacles, by a choice of texture, was taught by D2 away from applying it to another field than architecture. 

In particular, the abstract of D2 mentions: “the modeling system is efficient and robust because it exploits constraints that are characteristic of architectural scenes […]”. And: “a key aspect which distinguishes our approach from a structure of standard computer viewing of the formulation of movement / stereopsis (which makes it possible to circumvent the difficulties discussed in section 2) is that we are dealing with a particular field – architectural scenes – which entails a specific set of features that can be used to make the problem more simple”.

Sounds sound.

As a final note there is one interesting procedural aspect in this case.

According to article L. 614-15 of the Code de la propriété intellectuelle, when an infringement suit is initiated based on a French patent and there is a corresponding European patent or application, a stay of proceedings should be ordered as of right, until the European patent application dies or the European patent is granted and substitutes for the French patent (which takes place at the end of the 9-month opposition period, or, if an opposition is filed, at the end of the opposition proceedings).

In the present case, the grant of the European patent will be published this week, and therefore the application was still pending when the court issued its validity ruling.

Why did the defendant not request any stay of proceedings? Well, because the plaintiff had previously withdrawn the French designation of the European patent in April 2016, according to the EPO register. A possible strategy to keep in mind in situations of this kind.

CASE REFERENCE: Tribunal de grande instance de Paris, 3ème chambre, 4ème section, November 16, 2017, Fittingbox v. Acep Trylive, RG No. 16/09680.

One thought on “A fitting approach?”

Leave a Reply

Your email address will not be published.

This site uses Akismet to reduce spam. Learn how your comment data is processed.