Some Analysis of Physically Based Rendering’s Bibliography (1-3ed)
While I’ve continued failing to deliver on my promises to blog about debugging renderers, I have at least been productive on other fronts. The fourth edition of Physically Based Rendering is coming along well, and I’ve recently been spending time working on the “Further Reading” sections that end each chapter.
It’s been fun work, diving through all kinds of graphics research and especially in getting caught up in the areas where I haven’t been paying close-enough attention. As the end of that task started to come into sight, I started to wonder: who is the most cited person in the bibliography? It’s a short step from that to wonder who that was in previous editions and how that has changed over the editions. After a little bit of shell scripting, the TeX files yielded the goods. Today, we have the results for the first three editions. It should be possible to say something about the bibliography of the fourth edition in a week or two, once things get a little more first-draft complete in the bibliography department.
Before we get to the results, some words about methodology. I counted named authors on each and every publication in the bibliography, be it a blog post or a SIGGRAPH paper, all of it weighted equally. Editing a book or a proceedings didn’t count but otherwise everything was fair game. No points for position in author list, one way or the other.
In terms of interpreting the results, there are many caveats. First and foremost, the citations are naturally affected by the topics that the book does and does not focus on. There’s plenty of graphics not covered in the book, and so all of the corresponding work generally isn’t cited at all. Also, the specific topics covered in each edition shift around and that also affects the citations; we’ll see the effect of that shortly. Finally, although we try to be evenhanded, there is surely some unevenness in the choices: for example, I am pretty sure that papers written by the authors of Physically Based Rendering benefit from some unconscious bias as far as whether they make the bibliography or not. In short, the metric that this ranking is respect to is hard to define precisely and even if one could define it, it probably wouldn’t mean much of anything. (Alternatively, I think that the h-index is safe for now.)
With that said, here are the top-cited people for the first three editions. The numbers after names are the number of citations the corresponding person had in the book’s bibliography.
First Edition (2004) | Second Edition (2010) | Third Edition (2016) |
---|---|---|
Greenberg (26) | Jensen (31) | Jensen (33) |
Shirley (25) | Shirley (29) | Shirley (31) |
Hanrahan (22) | Greenberg, Hanrahan (27) |
Keller (27) |
Jensen (16) | Ramamoorthi (23) | Wald (25) |
Arvo (14) | Wald (18) | Greenberg, Hanrahan (24) |
D. Mitchell (13) | Keller (17) | Slusallek (21) |
Keller (12) | Arvo, Seidel, Slusallek (16) |
Marschner, Ramamoorthi (19) |
Heckbert, Torrance (10) | D. Mitchell (13) | Seidel, Arvo (17) |
Cook, Kajiya, Levoy, Pattanaik, Ward (8) |
Pattanaik, Torrance, Walter (12) |
Jarosz (16) |
There’s already plenty to see in the tally. For starters, with Eric Veach just missing the list for the first edition with 6 citations and with he and Jim Kajiya not being on it at all after the first edition, this list is evidently not a perfect measure of impact.
Moving on and accepting it for what it is, there’s Pete Shirley consistently maintaining his number two spot for three editions running, adding to his count at enough of rate to stay in place. Henrik Wann Jensen jumped up to first place for both the second and third editions; I was surprised that he only netted two more citations between those two editions given how much good stuff he published in the interim. Upon review, we seem to have cut a number of his second edition citations for the third due to shifting focus and cases where there came to be work from others to point to instead. (Sorry for the headwind, Henrik.)
Changes to the topics discussed in the book have had other impacts. Don Greenberg lost his number one spot in large part because the first edition had more to stay about tone reproduction and perceptual issues than later ones have. Thus, the first edition cited a lot of the work from Cornell on that front from the 1990s, but when that topic was cut in the second edition, the corresponding citations went away as well. Don is, needless to say, still doing fine in the end.
Related, Ravi Ramamoorthi rocketed up from two citations in the first edition to 23 in the second, a flood of great research being further boosted in the bibliographic count by a chapter on precomputed light transport (and thus, plenty in the book about spherical harmonics). Dropping that chapter and those references in the third edition didn’t help his placement in the list, in spite of citations added from lots more neat work in the meantime. (That change also hurt Pat Hanrahan between the second to third editions, though his Turing award level productivity mostly made up for it with new citations.)
Last but not least, Wojciech Jarosz has the distinction of being the only person to go from being not cited at all in the first edition to being on the list in the third.
As for the fourth edition, it is already clear that the list will be shaken up with a number of new names. Beyond that, the specifics are not yet settled. Will Henrik maintain the top spot once again? Will it be another round of “always a bridesmaid” for Pete Shirley? The answer is not yet clear, but stay tuned for early results once drafts of the rest of the “Further Reading” sections are finished in the next few weeks.