Trials@uspto.gov
`571-272-7822
`
`
`
`
`
`Paper 27
`Date: May 6, 2024
`
`
`
`UNITED STATES PATENT AND TRADEMARK OFFICE
`____________
`
`BEFORE THE PATENT TRIAL AND APPEAL BOARD
`____________
`
`APPLE INC.,
`Petitioner,
`
`v.
`
`IMMERVISION, INC.,
`Patent Owner
`____________
`
`IPR2023-00471
`Patent 6,844,990 B2
`____________
`
`Record of Oral Hearing
`Held: April 11, 2024
`____________
`
`
`Before JOHN D. HAMANN, STEVEN M. AMUNDSON, and STEPHEN
`E. BELISLE, Administrative Patent Judges.
`
`
`
`
`
`
`
`

`

`IPR2023-00471
`Patent 6,844,990 B2
`
`
`APPEARANCES:
`
`ON BEHALF OF THE PETITIONER:
`
`
`W. KARL RENNER, ESQUIRE
`Fish & Richardson P.C
`1000 Maine Avenue, SW
`Suite 1000
`Washington, D.C. 20024
`axf-ptab@fr.com
`(202) 626-6447
`
`DAVID HOLT, ESQUIRE
`holt2@fr.com
`(202) 626-7783
`
`
`ON BEHALF OF THE PATENT OWNER:
`
`
`STEPHEN MURRAY, ESQUIRE
`Panitch Schwarze
`Two Commerce Square 2001
`Market Street Suite 2800
`Philadelphia, PA 19103-7004
`smurray@panitchlaw.com
`(215) 965-1331
`
`
`
`The above-entitled matter came on for hearing on April 11,
`2024, commencing at 1:00 p.m., via video teleconference.
`
`
`
`
`
`
`
`
`2
`
`

`

`IPR2023-00471
`Patent 6,844,990 B2
`
`
`
`
`P R O C E E D I N G S
`- - - - -
`JUDGE HAMANN: This is a hearing in IPR2023-00471, Apple
`Inc. v. Immervision, Inc. I'm Judge Hamann. Also on the panel are Judges
`Amundson and Belisle. I'd like to start with an introduction of the parties.
`And so, who is here on behalf of Petitioner, please?
`MR. HOLT: I'm David Holt, Your Honor, and I'm joined by my
`colleagues Karl Renner and Karan Jhurani, on behalf of Petitioner.
`JUDGE HAMANN: Thank you. And for Patent Owner, who is
`appearing on its behalf?
`MR. MURRAY: Good afternoon, Your Honor. On behalf of
`Patent Owner, Stephen Murray. And with me is Dennis Butler, as well as
`my colleague.
`JUDGE HAMANN: Welcome to you all. I also want to point out
`that we also have a public line today. So, the public may be joining,
`listening in, or potentially watching it. And so I remind the parties not to
`convey any confidential information. I don't believe we have protective
`order in this proceeding, but I provide that reminder, nonetheless. Now, per
`our hearing order, each side is going to have one hour to present. We're
`going to begin with Petitioner as it bears the burden as to unpatentability,
`followed by Patent Owner's response, followed by any time reserved for
`rebuttal and sur-rebuttal.
`I'd like to also remind the parties of a few things. Obviously, this
`is a virtual hearing. So, when you're speaking, make certain you unmute
`yourself, please. And then when you're also done presenting please make
`3
`
`1
`2
`3
`4
`5
`6
`7
`8
`9
`10
`11
`12
`13
`14
`15
`16
`17
`18
`19
`20
`21
`22
`23
`24
`25
`
`

`

`IPR2023-00471
`Patent 6,844,990 B2
`
`certain to mute yourself, so we don't get any background noise that's not
`needed. If during the course of the hearing you have any technical
`problems, we expect you can let us know immediately. There's still some
`level of connection during the hearing please do that. To the extent that
`you've lost connection entirely or can't let us know otherwise, the
`information that was provided in setting up for you all's connection to this
`hearing, you should reach out to those folks.
`We have a copy of all the relevant papers and exhibits here, so we
`ask that, for if we're following along, as well as provide for a clearer record,
`that you refer to a slide number or exhibit number or page number or
`whatever is relevant in presenting your arguments as you present them.
`Lastly, as I said, each side will have an hour to present their
`arguments. I will try to give you time warnings towards the end, but you
`may find it helpful to also track your own time. Therefore, you can better
`pace the arguments you want to present. With that, before we turn to the
`Petitioner and to be getting into the arguments, I just want to ask one quick
`question of Patent Owner's counsel, Mr. Murray. Am I correct that Patent
`Owner did not file or is not relying on demonstratives for today's hearing?
`MR. MURRAY: That's correct, Your Honor.
`JUDGE HAMANN: Okay. Thank you. With that, I turn to hear
`Mr. Holt, and if you'd let me know how much time you would like to reserve
`for rebuttal.
`MR. HOLT: Thank you, Your Honor. We'd like to reserve 20
`minutes, please.
`JUDGE HAMANN: Okay. Thank you. You may begin when
`you're ready.
`
`4
`
`1
`2
`3
`4
`5
`6
`7
`8
`9
`10
`11
`12
`13
`14
`15
`16
`17
`18
`19
`20
`21
`22
`23
`24
`25
`26
`
`

`

`IPR2023-00471
`Patent 6,844,990 B2
`
`
`MR. HOLT: May it please the board? Mr. Renner and I will be
`discussing the case with Your Honors today. I will be starting with a brief
`discussion of the relevant claim language and the overall technology, which
`should take about 12 to 15 minutes. Thereafter, Mr. Renner will be focusing
`on how the combination of Baker and Shiota renders the claims obvious.
`If we turn to slide 4, we see Independent Claim 27, which is the
`focus of this proceeding. Claim 27 has two main limitations. The first
`relates to the capture of a panoramic image via a specific type of objective
`lens. The lens is one of the primary purposes of Immervision's purported
`invention. It's a panoramic lens with a nonlinear distribution function that
`includes at least one excluded zone, and at least one compressed zone. We'll
`talk a bit more about the nonlinear distribution function and the zones in a
`moment.
`The second limitation relates to displaying a corrected version of
`the image obtained through the panoramic lens. In essence, the nonlinear
`zones in the lens cause a type of distortion to the obtained image and this
`step includes correcting for that distortion based on two things. First, the
`nonlinear distribution function of the lens and second, the size L of the
`obtained image. There are a fair number of words to these two limitations,
`but you'll likely have noticed that this proceeding is focused in on only the
`last eight words. There is no argument in this proceeding that the
`combination of Baker and Shiota teaches everything that doesn't teach
`everything recited other than those final eight words. Nor is there any
`argument that a POSITA would not have been motivated to combine Baker
`and Shiota, or whether they would have had a reasonable expectation of
`success in doing so. With this in mind, I'd like to take a brief look through
`5
`
`1
`2
`3
`4
`5
`6
`7
`8
`9
`10
`11
`12
`13
`14
`15
`16
`17
`18
`19
`20
`21
`22
`23
`24
`25
`26
`
`

`

`IPR2023-00471
`Patent 6,844,990 B2
`
`the '990 Patent and its file history to understand the eight words that are the
`focus of the proceeding and why we focused on them.
`On slide --
`JUDGE HAMANN: Mr. Holt?
`MR. HOLT: Yes, sir, Your Honor.
`JUDGE HAMANN: This is Judge Hamann. Just as we -- to
`follow up on what's not in dispute, am I correct that there's nothing that we
`need to construe in this proceeding?
`MR. HOLT: That's correct, Your Honor. The parties have not
`focused on any of the claim constructions. There was some claim
`construction that happened in prior proceedings and in parallel, but none of
`these arguments today focus on any of this claim construction.
`JUDGE HAMANN: So, we can apply the plain and ordinary
`meaning to the claim terms?
`MR. HOLT: Yes, Your Honor.
`JUDGE HAMANN: Thank you.
`MR. HOLT: So, on slide 5, we see that the '990 Patent's
`description of its invention generally aligns with the first limitation of Claim
`27. We see a panoramic lens with a nonlinear distribution function that
`includes zones of expansion and compression. Here, the '990 Patent
`explains that the expanded zones of the image cover a higher number of
`pixels of the image sensor to provide higher definition in those zones.
`Notably, there's no mention of utilizing the size of the obtained image and
`whether that's a discovery made by these inventors or is otherwise a key
`focus of the purported invention. In fact, as we see on slide 6, the only
`discussion of using the size L of the obtained image as part of the correction
`6
`
`1
`2
`3
`4
`5
`6
`7
`8
`9
`10
`11
`12
`13
`14
`15
`16
`17
`18
`19
`20
`21
`22
`23
`24
`25
`26
`
`

`

`IPR2023-00471
`Patent 6,844,990 B2
`
`process is two lines, 18 and 19, at the end of the second algorithm for
`correction. You can see the variable L boxed in red at the lower right of the
`slide. As we'll see in just a bit, this was confirmed by Immervision in its re-
`exam when it added Claim 27. In doing so, Immervision was required to
`identify the Specification’s support for this limitation, and these lines of the
`algorithm are the only support that they cited in the re-exam. Thus, there's
`no prose in the '990 Patent Specification that explains why using size L
`obtained of the image is important or why it needs to be done. Instead, the
`algorithm simply presents L's use. Frankly, that's likely because the drafters
`recognize that a POSITA would understand the usage of size L of the
`obtained image, and that its use is a necessity of the process. But let's talk
`about why that is.
`If you'll jump with me to slide 10, we see a standard linear
`distribution function. Figure 4A on the upper left is the image disk projected
`by the lens onto an image sensor. Light in the real world enters the lens at
`one angle of incidence at the surface and then lands on this image disk at a
`related position. The graph in Figure 4B is the distribution function and
`effectively maps where on the radius of the image disk light that enters the
`lens at a specific degree of field angle is projected onto the image disk. The
`field angle at which the light enters the lens is on the X-axis and moves from
`0 degrees to 90 degrees. The corresponding location on the radius of the
`image disk that the light enters the lens at a given angle is projected is on the
`Y-axis. Here the distribution function is linear such that as you move from a
`field angle of 0 degrees to a field angle of 90 degrees, every degree of the
`field corresponds to an equal amount of the image disk. What does this
`mean practically? If, for example, light enters straight into the lens from the
`7
`
`1
`2
`3
`4
`5
`6
`7
`8
`9
`10
`11
`12
`13
`14
`15
`16
`17
`18
`19
`20
`21
`22
`23
`24
`25
`26
`
`

`

`IPR2023-00471
`Patent 6,844,990 B2
`
`center of the field, this graph of the distribution function shows that it will
`land at the center of the image disk. And if that image disk gets centered on
`an image sensor as it is assumed to be, the light will land at the center of the
`image sensor. On the other hand, if light enters the lens from a field angle of
`45 degrees, about halfway between the center of the field of view and the
`outer edge of the field of view, the graph of the distribution function shows
`that it will land at 0.5 of the radius of the image disk, or halfway between the
`center and the edge of the image disk. Importantly noticed here is the image
`disk radius is between 0 and 1 for the description of this distribution
`function, and we'll return to the importance of this in a moment.
`On slide 11, we see a nonlinear distribution function. Here, some
`angular positions of the field of view entering the lens are allotted more real
`estate on the image disk and thus on the image sensor that captures the
`image disk. We see an initially steep slope of the distribution function in
`Figure 7B from about 0 to 20 degrees of the field of view, followed by a
`shallower slope than a standard linear distribution function from about 20
`degrees to the edge of the field at 90 degrees. The steep slope means that
`light entering from the center of the field of view from 0 to 20 degrees is
`allotted more of the image disk creating a zone in which this portion of the
`real world image is expanded and ultimately allotted more pixels on the
`image sensor. On the other hand, a shallower slope creates a zone of
`compression in which the light from 20 to 90 degrees of the field is allotted
`less of the image disk and therefore fewer pixels than it would otherwise get
`in a linear distribution function. This means that you get more detail
`captured at the center and less detail captured out past 20 degrees of the field
`of view. It also means that if you were to view the real world image disk as
`8
`
`1
`2
`3
`4
`5
`6
`7
`8
`9
`10
`11
`12
`13
`14
`15
`16
`17
`18
`19
`20
`21
`22
`23
`24
`25
`26
`
`

`

`IPR2023-00471
`Patent 6,844,990 B2
`
`its projected onto the image center, it would appear distorted from the real
`world, with the image looking particularly expanded at the center and
`compressed at the edge. Why might you do this? Well, column 3, lines 43
`through 61 of the '990 Patent give several examples of use cases where a
`zone of expansion may be useful, including video conferencing like we're
`doing today.
`As you'll notice in today's Webex here, each person's face is in the
`center of the image captured by your camera, and your background is
`towards the edge of the image. By using the nonlinear lens, we just
`discussed, more pixels and therefore detail and definition could be devoted
`to the user's face and fewer pixels dedicated to the background, which is of
`less interest. In fact, the lens designer would be trading detail at the edges of
`the image for details at the center. Any questions about that, Your Honors?
`JUDGE HAMANN: Yeah, and this is Judge Hamann. I just had
`one question. Could you explain the meaning you're providing as to the
`terms field angle and field of view in the context that you just used them, so
`you have it here in the record?
`MR. HOLT: Certainly, Your Honor. So, the field of view is the
`amount of the real world that the lens effectively is able to capture of light
`and then project back onto the imaging plane. And so, for instance, in a
`panoramic lens like the one we're talking about here, it has a 180 degree
`field of view, which means that if you pointed the lens forward from you, it
`would be able to see from that optical axis kind of straight out and then 90
`degrees to either side of you. So, if you had the kind of camera as your eyes,
`you'd be able to see straight out, and the field would be 180 degrees field of
`view from that center axis all the way out to kind of your right and your left
`9
`
`1
`2
`3
`4
`5
`6
`7
`8
`9
`10
`11
`12
`13
`14
`15
`16
`17
`18
`19
`20
`21
`22
`23
`24
`25
`26
`
`

`

`IPR2023-00471
`Patent 6,844,990 B2
`
`directly to your side. And then the field angle is the angle at which light
`would enter into any part of the lens from that position in the field. So, we
`consider the example of light entering the center of the lens, the very kind of
`middle of the lens, light from all objects around that 180 degree field will be
`coming at all parts of the lens or at least most parts of the lens, depending on
`the curvature. But all light can hit the center, for example, and as it hits the
`center of the lens, it's going to be coming into the lens at that point of the
`lens from the various angles related to where that object exists out in the
`field of view. And so that angle is called the field angle. Does that answer
`your question, Your Honor?
`JUDGE HAMANN: Yes. Thank you.
`MR. HOLT: Excellent.
`JUDGE AMUNDSON: This is Judge Amundson. I had just a
`question. I'm making sure I'm understanding this. So, with regard to Figure
`7 of the '990 Patent, as I understand the unpatentability theory, Baker, you
`rely on Baker for something analogous to Figure 7, where you've got a
`nonlinear function of the light coming in and striking the disk. Is that right?
`MR. HOLT: Yes, Your Honor, that's correct.
`JUDGE AMUNDSON: Okay. So, now when Baker goes to
`display something, did Baker correct for that nonlinearity?
`MR. HOLT: Baker describes a correction for that nonlinearity,
`yes, Your Honor. But it does not provide details of the process by which
`you would actually, from a mathematical, conceptual perspective, do that
`translation. And Shiota is a reference that provides some of those more
`granular details, Your Honor.
`JUDGE AMUNDSON: Okay. I got it. All right. Thank you.
`10
`
`1
`2
`3
`4
`5
`6
`7
`8
`9
`10
`11
`12
`13
`14
`15
`16
`17
`18
`19
`20
`21
`22
`23
`24
`25
`26
`
`

`

`IPR2023-00471
`Patent 6,844,990 B2
`
`
`MR. HOLT: So, before I move on from this slide, which is slide
`11 we were looking at, I just want to note again with respect to figure 7B of
`the '990 Patent that the image disk radius is between 0 and 1 for the
`description of this distribution function. And so, you might be asking
`yourself, why does the '990 Patent describe the radius of the image disk as
`being between 0 and 1? And on slide 12, we see that Dr. Kessler, Apple's
`expert, explained that the '990 Patent's description of its image disk is being
`between zero and one, and as a relative distance, means that the image
`distribution function is a normalized function. In other words, the '990
`Patent's image transformation function that gets utilized to transform the
`image that's captured assumes a relative or normalized size of the image
`pickup device, and that normalization effectively is dimensionless. And this
`ultimately we'll learn later when Mr. Renner is talking to Your Honors about
`Baker and Shiota simplifies the math. In effect it represents a radial location
`on the image expressed at the percentage of the distance between the center
`and the edge of the image disk that's created by the lens and picked up by the
`sensor. Mathematically we see that represented in the equation in red on the
`bottom right of this slide. But practically, what all this means is that if the
`normalized or relative distance on the Y-axis is 0.3, for example, it is 30
`percent of the distance between the center of the real world image disk and
`its outer edge. If the relative distance on the Y-axis is 0.67, it's 67 percent of
`the distance between the center and the edge.
`On slide 13, we see that Mr. Munro, Immervision's expert, agrees.
`Mr. Munro explains that relative distances and normalized distances are the
`same thing, and they both are just percentages of the radial location on the
`image plane. As we see, Mr. Munro provides an example where the relative
`11
`
`1
`2
`3
`4
`5
`6
`7
`8
`9
`10
`11
`12
`13
`14
`15
`16
`17
`18
`19
`20
`21
`22
`23
`24
`25
`26
`
`

`

`IPR2023-00471
`Patent 6,844,990 B2
`
`or normalized distance is 0.5, which would mean that it's halfway or 50
`percent of the radial distance between the center of the image disk that's
`created by the lens and projected onto the image pickup device, and the edge
`of that image disk. On slide 14, we see that Mr. Munro also confirms that
`the size L of the obtained image that is recited in the last eight words of
`Claim 27 is the same distance between the optical axis of the image
`produced by the lens and the outer edge of that image, which he also calls
`the extent of the image. So, Mr. Munro was saying that here to measure the
`size L of the obtained image, you would measure the radius of the circular
`image projected by the lens onto the image plane where the image pickup
`device is generally positioned. To be clear, Mr. Munro is acknowledging the
`link between the size of the image projected by the lens and the claimed size
`L of the obtained image from the image sensor.
`Turning to slide 15, we see Mr. Munro explaining that because the
`output of the image transformation function is relative or normalized, as it is
`in the '990 Patent, you would need to scale that output based on the actual
`size of the image. This stands to reason and, frankly, is perfectly
`straightforward and logical. If you have a normalized and unitless quantity
`in order to utilize it in the real world, you need to apply that normalized
`value or percentage to some quantity. And you apply it by simply
`multiplying the percentage by the quantity to which the percentage applies,
`which in this case is the size of the image disk. That's what we see in lines
`18 and 19 of the second image transformation algorithm that we looked at
`earlier. Given how straightforward this concept is, it's unsurprising why the
`'990 Patent simply includes this operation at the end of the code and
`otherwise spends no time focusing on its importance or explaining the
`12
`
`1
`2
`3
`4
`5
`6
`7
`8
`9
`10
`11
`12
`13
`14
`15
`16
`17
`18
`19
`20
`21
`22
`23
`24
`25
`26
`
`

`

`IPR2023-00471
`Patent 6,844,990 B2
`
`operation to the reader of the Patent. In each a POSITA would have
`understood that in order to utilize this normalized value, this is what you
`must do. Nonetheless, if we go back to slide 7, we see that this
`straightforward feature that gets very little attention in the '990 Patent itself
`becomes the focus of Immervision's arguments about why new Independent
`Claim 27 distinguishes the Baker and Shiota combination. In the clip at the
`top left, we see Immervision acknowledging that Baker and Shiota were in
`the same field of endeavor and that it would have been obvious for a
`POSITA to consult Shiota when implementing Baker's system. In the clip at
`the center, we see that Immervision explained that the final eight words of
`the new Claim 27 were supported by the last two lines of code from the
`second algorithm that we were looking at earlier. And in the last clip on the
`lower right, Immervision focused on those last eight words of Claim 27, and
`thus those two lines of code to try to differentiate their claim from the
`combination of Baker and Shiota. They distinguished the combination by
`specifically saying that the image size of the image disk is not utilized to
`correct the image but did so here without citing to or otherwise explaining a
`key teaching in paragraph 23 of Shiota that a POSITA clearly would have
`read as teaching or at least rendering obvious, what we just discussed as the
`straightforward way to account for normalized output of an image
`transformation function.
`And with that, Mr. Renner will now walk you through the key
`paragraphs and the other relevant teachings of Shiota, unless Your Honors
`have any further questions for me about the general technology.
`JUDGE HAMANN: Mr. Holt, this is Judge Hamann. I just had a
`couple of questions looking at slide 15 in particular, for example.
`13
`
`1
`2
`3
`4
`5
`6
`7
`8
`9
`10
`11
`12
`13
`14
`15
`16
`17
`18
`19
`20
`21
`22
`23
`24
`25
`26
`
`

`

`IPR2023-00471
`Patent 6,844,990 B2
`
`
`MR. HOLT: Yes, Your Honor.
`JUDGE HAMANN: And it looks like here, pages 37 and 39 of
`Mr. Munro's deposition, Exhibit 1025 I believe it is, are cited. I assume that
`these portions are cited in Petitioner's reply, because I did not see the
`reference to where on this slide.
`MR. HOLT: Yes, Your Honor. Yes, Your Honor. I believe that
`they are. I will find that. I don't have that directly in front of me at this
`moment, but I can find that for you, Your Honor. It is cited though.
`JUDGE HAMANN: And similar as to the arguments I think you
`were making; was it slide 17 you were referring to when you were talking
`about the re-examination?
`MR. HOLT: Yes, Your Honor. There was a discussion in the
`Petition itself actually about the re-examination history in the prosecution
`history section of the Petition.
`JUDGE HAMANN: Thank you.
`JUDGE AMUNDSON: This is Judge Amundson. I think just to
`make sure that the record is clear, I think he referred to slide 7, not 17, for
`the re-exam. Is that right?
`MR. HOLT: Yes, Your Honor. You're asking about the question
`or my reference to the slide?
`JUDGE AMUNDSON: It may have been something Judge
`Hamann said, okay, that's all. I think he may have -- that's all.
`MR. HOLT: Yes, Your Honor, though, the re-examination is
`discussed in our slide 7. You're correct, sir.
`JUDGE AMUNDSON: Okay. I got it. Thanks.
`MR. HOLT: Okay. Thank you, Your Honors.
`14
`
`1
`2
`3
`4
`5
`6
`7
`8
`9
`10
`11
`12
`13
`14
`15
`16
`17
`18
`19
`20
`21
`22
`23
`24
`25
`26
`
`

`

`IPR2023-00471
`Patent 6,844,990 B2
`
`
`JUDGE HAMANN: Mr. Renner?
`MR. RENNER: Yes, Your Honor.
`JUDGE HAMANN: This is Judge Hamann. You have about 20
`minutes left if I'm doing my math correctly in your initial allotment.
`Obviously, you could bleed over if you prefer, and we can just do the math.
`MR. RENNER: Thank you, Your Honor, appreciate that.
`Hopefully, that will do. We'll get moving and see what happens. So, if we
`look at slide 16, please, Your Honors. This is Karl Renner on behalf of
`Apple, Petitioner. I'll be addressing, as Mr. Holt had indicated, the
`combination and the applicability of the combination of Baker and Shiota in
`particular, and how it feels, in particular, given the dispute that the parties
`have had with the features correcting the non-linearity of the initial image, as
`well as in doing sort of the correction using the size L of the obtained image.
`And to the point, let's look at slide 17, if we could. You'll recall in the
`Petition the Petition looks to Shiota, really, with respect to establishing a
`POSITA's knowledge of non-linear image distribution functions to
`compensate for distortions in a captured image. And on this slide, we can
`see evidence speaks to this issue.
`The evidence here in the upper left would be paragraph 36 of
`Shiota, as well as on the upper right, paragraph 201 from Dr. Kessler's first
`declaration. We see this really is not a disputed point. And we can see that
`when we look even at the Patent Owner's acknowledgement of the use of
`nonlinear functions and '990 re-examination as it relates to their
`conversation of this reference, the Shiota reference. So, I just wanted to
`make the point, since it's a fundamental one, that Shiota is bringing forth the
`details, Judge Amundson, on the nonlinear correction.
`15
`
`1
`2
`3
`4
`5
`6
`7
`8
`9
`10
`11
`12
`13
`14
`15
`16
`17
`18
`19
`20
`21
`22
`23
`24
`25
`26
`
`

`

`IPR2023-00471
`Patent 6,844,990 B2
`
`
`If we look at slide 18, we also note that from Shiota we have a
`reference to normalization another generally understood concept we'll get
`into as it relates to image processing and a POSITA's knowledge of
`normalization as an image transform operation. On this slide, upper left,
`we've got paragraph 23. It's a paragraph that's gotten a lot of attention. On
`the upper left in that paragraph, we know that a fisheye image projected in
`circle of radius one is described in the yellow text here, a fisheye image
`projected in a circle of radius one. Not unlike what you see in the '990
`Patent because it's just normalization really. On the right you've got
`testimony from Dr. Kessler that speaks in greater detail and analyzes this
`section of paragraph 23 and that is found in his original dec. at 205.
` Also, he provides further testimony on this is his supplemental
`dec. at 55 for Your Honors' reference. Basically, what he's saying as it
`relates to each, is it normalizing to a size of one an image processing is a
`well-known concept, and there are various reasons to do it. But it's well-
`known. It's common in optics to do it. Staying with that for maybe a couple
`slides for it to give you a rounded out since that the record on this slide 19
`shows you what is cited to by Dr. Kessler providing his -- his testimony on
`the point. That is the Fisher reference up top. And that's just corroborating
`this concept that normalization because you can see the scale of 0 to 1 on the
`left. It's something you could find regularly in optics. We'll talk a little
`more about that when we talk about some testimony from the Patent Owner's
`expert, but this is confirmed as we see on the right as we begin that
`conversation by the testimony from the Patent Owner's expert. And if we
`look at slide 20 we see it a larger excerpt from that testimony which we --
`we definitely commend to Your Honors. Here, we had a conversation with
`16
`
`1
`2
`3
`4
`5
`6
`7
`8
`9
`10
`11
`12
`13
`14
`15
`16
`17
`18
`19
`20
`21
`22
`23
`24
`25
`26
`
`

`

`IPR2023-00471
`Patent 6,844,990 B2
`
`their expert about normalization in particular and we talked about the
`perspectives that a POSITA has a normalization in the field, as well as
`motivations that they bring to the table while they're thinking about
`normalization. And we hear from him things we expect to hear, consistent
`with what Dr. Kessler had indicated, that normalizing is commonly done he
`says throughout the field of optics and lens design. And the reasons for that
`are several.
`One, for instance when you normalize from a 0 to 1 scale, the math
`is just simply made more generic. It actually simplifies while they're being
`more generic, it becomes more applicable to a wider range of lens
`configurations. And it becomes applicable to them without regard for actual
`image size. So, the computation becomes easier to do regardless of what
`scale is applied and then you can apply the scale after and correct and make
`it applicable. And in this sense he says it's possible to genericize or
`generalize across different types of lens configurations when you use
`normalization. So, we don't see any dispute from the Patent Owner that
`these benefits and these motivations would have not been applicable when
`thinking about Shiota or the Baker-Shiota combination.
`But that returns as we go to slide 21 and bringing this back into
`Shiota itself. Paragraph 23 we show you here. We wanted to explore it a
`little bit. The blue text here, we notice a focus on a term, this image circle
`diameter. You can see it's the upper part of the blue text highlighted. That's
`really in size. It's diameter. When I looked at it a couple times it took me,
`but the words image circle diameter you can see it's really an image size.
`And that size is said by Shiota in this blue text to differ to vary and to vary
`on what? Continuing in the blue text to vary on sensor size. It says the size
`17
`
`1
`2
`3
`4
`5
`6
`7
`8
`9
`10
`11
`12
`13
`14
`15
`16
`17
`18
`19
`20
`21
`22
`23
`24
`25
`26
`
`

`

`IPR2023-00471
`Patent 6,844,990 B2
`
`of the image pickup device. So, it's Shiota that's making a suggestion for
`relating these concepts. They don't have to be the same necessarily, they
`just have to be relating to one another. And we know with one we see the
`other, and we see variances between them called out.
`So, variations in fact in physical dimensions with that of the image
`sensor, per Shiota's teaching itself, are said to lead the variations in its size.
`You know when you continue on in what Shiota was telling you in this
`paragraph in the yellow text, again it tells you the fisheye image is projected
`in a circle of radius one but notice that just before that it begins with the
`word consequently. It's pretty important word because it's telling you
`consequently. What are you looking? You're looking at what just happened
`to inform what's about to happen, right? So, what's just happened in blue
`text do you remember, was we talked about the diameter of the image size
`varying based on the sensor size.
` Consequently, you have a projection in a normalized sense. In
`other words right here you have Shiota itself telling a POSITA who can read
`it, that because you have the dimensions varying, is what you'd contemplate
`as a Shiota writer, that you'd want to normalize to in fact make it easier to do
`your computations, make them more generally applicable and make it so that
`when you have a variation of the type they contemplate the math does not
`have to be completely redone. You just have to skip. This as a word choice
`is pretty important we can see because normalizing here is pretty purposeful,
`it consequently tells us that. That image size and the center are tied together
`as well.
`
`Slide 23 if I may. So, a point of suggestions by Shiota, what we
`see is that a person of ordinary skill would find it obvious to apply the
`18
`
`1
`2
`3
`4
`5
`6
`7
`8
`9
`10
`11
`12
`13
`14
`15
`16
`17
`18
`19
`20
`21
`22
`23
`24
`25
`26
`
`

`

`IPR2023-00471
`Patent 6,844,990 B2
`
`nonlinear distribution functions and to apply normalization together. And
`more to point, as suggested by Shiota, a POSITA would find it obvious in
`the context of the Baker-Shiota combination that the image distribution
`function that would result would apply these two points drawn from the
`captured image. To the left, we can see a discussion that relates to that in
`paragraphs 36, 37, and 41 of Shiota. Those paragraphs discuss an output of
`an applied in its distribution function, and importantly, they speak to t