Wow Gold Store Welcome you! Look here to Buy wow gold, Cheap wow gold,world of warcraft gold

c893t

Defrag regularly - The faster your hard drive does its work - less demand you are going to put on the hard drive and your battery. Make your hard drive as efficient as possible by defragging it regularly. (but not while it’s on battery of course!) Mac OSX is better built to handle fragmentation so it may not be very applicable for Apple systems.Such as apple laptop battery,apple a1022 battery,apple a1079 battery,apple a1078 battery.

caelco

 
Hi, I'm new too. But here is a recipe given to me from www.holyfoodimports.com - If you order their food products they send recipes!

Honey-Carob Brownies

1/2 cup butter, melted
1 1/4 cup honey & carob spread (available at www.holyfoodimports.com)
2 large eggs
1 teaspoon vanilla extract
1 cup whole wheat flour
1 teaspoon baking powder
1/4 teaspoon salt
1/2 cup walnuts, chopped

1.Preheat oven to 350*F
2.Blend together melted butter, honey & carob spread in a bowl. Add eggs and vanilla; mix well.
3.Add a sifted mixture of whole wheat flour, baking powder, and salt; mix well. Stir in chopped walnuts. Pour into a greased 8-inch baking dish.
4.Bake for 40 minutes. Cool on a wire rack before cutting into squares.
Makes 15 bars.
laptop battery
laptop batteries

when they cheat us on our counts what they are doing is punishing hard work, I used to bring into our office 10 times the amount of postage sales as the other three carriers combined, but when I finally got it thru my head that I was not going to get compensated any more than the other carriers I quit trying so hard. Its hard to stay enthused when you know you're going to get cheated out of your hard work. Usually when a business is having the problems the PO is its because of POOR mgmt. With very few exceptions PO mgmt stinks from top to bottom. I heard today the none of the top managers have ever worked in any of our crafts. Anyone who has ever carried a route would have known that the LLV's should have been built with front wheel drive, and a window in the rear.laptop battery
laptop batteries

Convert to MP4

Easy Way to Convert Video Files to MP4 with Cucusoft Ultimate Video Converter

Need new Rip DVD to AVI ?
Rip DVD to AVI
Have a nice surfing！

How do you feel about Siggraph Asia?

3D scanning for hairstyles (critiques as off-topic for SIGGRAPH below) sounds on-topic to me. Computer vision in general, certainly! It's "Computer Graphics and Interactive Techniques", not "Rendering". SIGGRAPH will certainly have many rendering papers this year as always, but it's a large multidisciplinary conference and IMHO that's the beauty of it.

For those of you who have reviewed papers for SIGGRAPH 2008, I would like to take an unofficial poll about what you think of this year's review process, in particular, the new part asking reviewers (including all primary, secondary and tertiary) to try to reach consensus before the committee meeting.

Among the 8 papers that I reviewed, 6 of them reached unanimous consensus so that's all cool. In one of the remaining 2, I had a different opinion from other reviewers and in the end the senior review essentially twisted my arm to agree with them. In the other one, the paper received a bimodal distribution of scores (2 >= 4.0 and 3 <= 3.0) but interestingly, only 3 reviewers chimed in during the discussion process. In the end, no consensus is reached (or even attempted).

I think I like the transparency of this added phase of asking all reviewers to reach consensus, and I would say at least in the 8 papers that I reviewed, I have not seen any abuse of the review process. Even for the 2 papers that did not reach full consensus, they both got accepted so at least they are erred on the positive side; I always think it is a bigger crime to reject a good paper than to accept a bad one; for the former, I lose a chance to read a really good paper but in the latter, I could always choose not to read it or in the worst waste a little bit time to figure out the paper is bad.

This also brings us to the earlier posts regarding topics of accepted papers. I believe as long as the papers have gone through a fair and organic review process, I have no problem seeing non-traditional stuff, e.g. computer vision or computational photography. In all likelihood they could broaden people's research scope, and in the worst case you could always choose not to read them.

To #274.

Yes, the topics in the SIGGRAPH this year will be very diverse. So far I see those non-graphics topics (according to your definition):

1. Computational photography
2. Perception and color science
3. GPGPU
4. Deblurring

In my opinion, they are all very valuable to the computer graphics community. Many computational photography papers relies on the light transport theory, and their results may benefit the traditional problems in the future. For example, there is a paper that uses flash/non-flash image pairs to reconstruct the depth map. Also, if you can use those devices to achieve the special effects in acquisition, why you need to rendering them in the post-processing stage?

All perception related topics are of fundamental importance, especially when the physics-based rendering algorithms are very mature (if we don't consider the speed issue) and the expressive images mostly rely on the post-processing at the pixel shaders. Finally, when GPU becomes so powerful, porting some complex algorithm onto GPU is a great idea. As for the deblurring...I have no comment. But at least the results shown in the papers are really amazing.

Most reviewers and committee members have published numerous pure-graphics SIGGRAPH papers and still decide to make the SIGGRAPH like this. Maybe you should see the quality of the rejected papers to make the fair judgement. Also, the topics you like may kill the graphics in other's point of view. For example, I can blame measured BRDF data that kill the analytic BRDF functions. We should ask more talented people to design new BRDF functions to kill the measured BRDF data :P.

Ok, so the semi-official SIGGRAPH 2008 paper list is being compiled, and it seems it's true what they say. Graphics is dead. We now have a conference called SIGGRAPH, but most papers appear to be about yet another project where somebody used a number of digital cameras and duct tape to build yet another specialized 3D scanner of some sort. Shouldn't there be some IEEE Optics Conference or something like this? Almost all of this is computer vision at best, if not just homebrew hardware engineering. You certainly would hardly have called any of this computer graphics 10 years ago. It seems the remaining people doing graphics research can now be found in venues such as IEEE Visualization and the like.

Seriously, there is a paper on a 3-D scanner for acquiring HAIRSTYLES this year. You've got to be kidding me!

Hello all,

I am definitely out of it--I had no idea this forum was going on. But then, I more or less dropped out of CG a bit before Michael (but less ceremoniously) for mostly the same reasons, and I haven't been to SIGGRAPH in a few years. It is very encouraging however, to see so many others agreeing that there are some serious problems with SIGGRAPH as well as the peer-review process in graphics. In fact, it has revived my interest in the field. Thanks!

It looks like this forum is essentially dead and I missed my chance to post, but in case anyone is still reading it there a couple of things I'd like to add. First, I have come to learn that many of the problems described here also show up in other fields. To my mind, the "peer review" process has two basic problems: we are never formally trained in how to do reviews, and we are seldom (if ever) held accountable for our reviews. That is not a good combination. My second comment has to do with another down side of the emphasis placed on SIGGRAPH papers. If work published in SIGGRAPH becomes a "sacred cow", what happens if research comes along later that contradicts it? In my experience, it is pretty hard to publish that kind of research. I agree that graphics is getting "overcrowded" and it would be nice to have other top-tier venues.

Michael, thanks for starting this forum, and I am flattered that you included me in your "A-list."

Edited by author 11-03-2007 11:53 PM
 In response to post by 'clipi': I am still "around" and, as you see, even clean up this by now dead forum once a (few) month(s), as promised. Currently I do not closely follow events in what is now "alive, well and very vibrant" CG community and although certainly "hear things", do not consider it appropriate for me to comment on CURRENT state of affairs any longer. Of course, you will be better off not asking me for any advice in the first place given my, well, rather equivocal reputation in CG community. But who knows - maybe someone else still reads this forum and can give you some useful "insights and suggestions" if you post your questions/concerns here and maybe other people find such discussion useful as well.

 Thanks for wishing me good luck - we all need it ...
 M.A.

First of all I just want to say that I am sorry tha Michael is leaving
the field. He had made such great contributions to the CG world that
it certainly will miss his expertise. I hope at least that he would
be around to give insigths or suggestions for other people and share
his knowledge. Is there anyway other way I could personaly contact him?
Thanks so much and good luck in your new field.

I think the computer graphics community is alive, well and very vibrant. However, I am a little pissed off about grad school in general. Because funding can be difficult to come by for any research program, this is not CG specific. But I hate fundraising when it gets in the way of actual research.

 Wow - two people whose work I deeply respect (both on my "A-list" of farewell acknowledgements, so I am not just saying this now) got siggraph awards!!!

 I hope that whoever still reads this forum will join me in congratulating Greg Ward and Nelson Max for getting the well deserved (and, I might add, long overdue) formal recognition. Most impressively to me personally is the fact that neither had a huge count of siggraph papers recently. Is this a sign of real change ? Surely hope so ...

 As for the forum, well, some people seem to have expressed interest in keeping this forum alive and so be it. For one, it would be interesting to hear people's impressions from upcoming siggraph as related to this forum's main topic.

 Since nobody volunteered to take it over, I guess I will keep cleaning this place from time to time (no, there is no automatic tools to do this). Apologies if I can not do this often enough and as a result you see too much junk posts here.

 M.A.

It turned out that Michael has another cult following. See
http://hardware.slashdot.org/article.pl?sid=06/08/29/0443210

I have been on the SIGGRAPH paper committee for the last 2 years. In general, I consider myself a nice guy and far from being a paper killer. However, just like many other committee members there are certain papers that are considered good by others but for various reasons I would like to reject. Here I would like to offer a glimpse on how I exploited the current system to kill these papers. Since I won’t be in the committee next year, I hope some of these exploits will be fixed so that they won’t come around to haunt myself.

1. Try to learn as much as possible of the submitted papers and identify those that you would like to reject. In some situations, you might get to know some papers even before the SIGGRAPH review process (e.g. faculty applicants listing ongoing work on their research statements).
2. Try NOT to become the senior review for that particular paper. Even though it is unknown if the paper will be assigned to you, at least the best you could do is not to request the paper. The advantage of not being a senior reviewer is that your opinions won’t be seen by the authors during the rebuttal period. Then, during the committee meeting, you could simply jump out and request to take a look at a paper already listed as accepted (this was actually advocated by the Paper Chair himself this year). Since during the committee meetings the authors’ opinions are not consulted, you can easily bring on new issues without receiving much challenge, just make sure that you possess sufficient knowledge to deal with the senior reviewers in charge of that particular paper. (In case you wonder how a paper could fall into the hands of people without the best expertise, remember that the paper assignment, done by only a few people (Paper Chair and Advisory Boards in previous years and Area Coordinators this year), could hardly be perfect.)
3. If you are one of the senior reviewers for a popular paper you would like to reject, there are multiple tricks. One option is to have really nice language in your review (accompanying a lukewarm score like 3.2) but with a fatal comment that seems innocuous, such as “I really like the main ideas presented in this paper and the results look all very impressive. However, I felt that a certain component of the presented algorithm has been done before in [a previous work that is not cited or adequately discussed in this paper].” Or something like “This is a key part of the algorithm; however, I don’t think the exposition is clear enough for reproducibility.” The key point is to obscure the severity of this statement from the authors so that they might neglect to fully address this fatal point during rebuttal.

There are certainly other tricks that I know of and/or have seen deployed by other committee members, but it will be too long a post. Here I have simply presented a few tricks that could be done mostly alone. You might note some of these exploits are benefited from this year’s particular innovations, giving the committee members much more power than in previous years.

Hi Michael

Why don't you submit the D-BRDF paper to a journal like TOG or TVCG instead. At least a published D-BRDF paper would be much better for the CG community than a repeatedly rejected siggraph sumbission! I believe there is more than one person genuinely interested in using the D-BRDF framework for BRDFs. Hence I would suggest that it should be submitted to an alternative venue.

regards
Abhijeet

I am not sure if that is a question of fairness and I have no big concerns about fairness at this point. I have not seen a final program, but if Marc really has a paper at Siggraph this year I would be disappointed. He did make a big announcement.

Regarding post #181 by Anon:

As Marc's former student, I can personally vouch on his fairness because, despite "a spirit of generosity prevailed this year", *all* my SIGGRAPH submissions this year are rejected, an standard-deviation-10 achievement given the good reviews I have received.

Even though what happened to my papers might provide some interesting case studies (I am still scratching my head in disbelief), however, I have decided not to post them up for now because it will sound too much like whining. Instead, I will try to share what happened to me this year after next year's SIGGRAPH announcement if this board is still alive.

Meanwhile, I hope my case will cheer everybody up. ;-)

PS: my posting appeared multiple times due to network issue.

Regarding post # 181 by Anon:

As Marc's former student, I can personally vouch on his fairness because, despite "a spirit of generosity prevailed this year", *all* my SIGGRAPH submissions this year are rejected. (And if you are one of the committee members who have seen the scores of my papers, you would probably agree with me on the astronomically unlikely probability of this achievement.)

John,

These are interesting hypothetical arguements of what *might* happen if the TOG eidtors were non-anonymous. But given that TVCG already has non-anonymous editors, do you have any examples of how it has hurt their careers? In your TOG capacity, have you spoken with other journals who have non-anonymous editors and found that they agree with the flaws you mention? I wonder how each journal internally makes this decision... etc...

All 5 journals that I am an editor for have me as non-anonymous. I do not fear reprisal, but rather must act fairly. In particular, I would *never* write anything even close to what I typically recieve from Siggraph! Even if a paper of mine gets accepted, the person I correspond with tends to be way more harsh than I suspect that they would be if their name were attached to the correspondance.

Ron

PS. You're overworked. That's way too many papers for you too have to read all the reviews for. The community owes you a lot for all this effort.

Glad to see this list is still active, though I'm sorry to see that the same problems still exist that prompted the list's creation. I see we also have some nice deals on purchasing computers available to us.

/m180 suggests that TOG change its policy from anonymous editors to non-anonymous editors. That's an interesting idea but there are additional concerns.

The desire to hold someone responsible for a paper's treatment in review is driving this request, and the community seems to be upset with SIGGRAPH's treatment of papers, not TOG's. The primary concern with TOG has been keeping the process moving quickly so papers can be published as rapidly as possible, which is more challenging with TOG's revolving schedule than with SIGGRAPH's hard deadlines. The primary concern with SIGGRAPH, on the other hand, is the quality of the decisions, and the primary contributing factor to this is likely SIGGRAPH's hard deadlines for decisions (and the fact that SIGGRAPH resubmission is annual).

TOG receives on the order of 100 submissions each year, and these submissions are spaced fairly evenly throughout the year, though with some peaks and some troughs. A big difference between the SIGGRAPH Papers Chair and the TOG Editor-in-Chief is that the Papers Chair oversees the decision of about 450 SIGGRAPH submissions in one weekend, whereas the Editor-in-Chief oversees the decision of 100 TOG submissions over the entire year. As such, the TOG EiC in fact is responsible for ultimately making the accept/reject decision on TOG papers. These decisions usually (but not always) follow the recommendation of the (anonymous) AE in charge of the paper. Thus for TOG, the EiC can be held accountable for the decision and should be the responsible party if something goes wrong.

If the AE made the ultimate decision based only on reviewer input, and the AE were not anonymous, it could lead to other complicating and unfair issues. For example an AE could fear reprisal from the author of a rejected paper, or could expect quid pro quo reward from an accepted paper's author. The current process discourages that as I tend to follow AE recommendations and if I disagree with the AE I have to present a strong and convincing argument why.

There is some discussion to try to diffuse SIGGRAPH's once-a-year hard deadline into several deadlines throughout the year, which would accelerate the resubmission of revised papers, allow authors to submit work closer to when it is truly ready, and reduce the workload on reviewers/deciders to a level where they can be more careful. But to work, that solution would need a lot of support from the community to drive through some pretty dramatic changes to the current system.

In related news, in message 22 of this thread (07-25-2006 04:54 PM ET), Marc Levoy wrote "Stanford will probably have fewer papers next year (not more), partly because I plan not to submit any papers with my name on them. No rule requires this, but I am committed to avoiding the appearance of bias." He also made a *big* deal out of this fact at last year's "How to fix SIGGRAPH" meeting that was held during the conference.

I like how Marc's commitment to avoiding the appearance of bias extended all the way to his submission of "Veiling Glare in High Dynamic Range Imaging"...a (surprise!) paper with his name on it that was accepted for publication in SIGGRAPH this year.

I would not have had any problem with him submitting papers as previous papers chairs have done, but this was fundamentally dishonest and in my opinion he should be taken to the woodshed for it.

Edited by author 04-04-2007 02:42 AM
Michael's last post points out that things are no better this year. In some ways they are worse. This year I got my 15th and 16th Siggraph paper, so overall I shouldn't complain. But on the other hand I just received the most uneven treatment *ever* on one of my submissions.

We had a submission with better reviews than the two that were accepted, and it had no real criticisms, and the lowest score wasn't even that bad. So we had very little to say in the rebuttal, but alas the paper was shockingly rejected. I've published about 100 papers across various fields in conferences and journals and never seen one with such good reviews and only superficial concerns rejected before.

The main reason given was that we violated anonymity in the rebuttal. Since a reviewer pointed out that there was prior art to our work, we had to cite our previous sketch to rebut this, but aha, catch 22, you either violate anonymity or get dinged on novelty. Yes, I bet the people who rejected this paper are very pleased with themselves - at the clever method they found to reject the paper. Congrats geniuses!

One of my students (first author on this paper) has 6 Siggraph papers at the ripe old age of 25, and has decided to quit academics and go to industry. He's quite unhappy with the intentional bias and stupidity that he has seen in the Siggraph reviewing process. Again, these reviewers are probably quite pleased with themselves, as they removed even more competition. The second author on that paper had wanted to dedicate his career to teaching even over research - although we could use more people like that in academia, he too is shaken by what he has seen in the Siggraph process.

To top this all off, let me comment on anonymity. In two previous years I had papers to referee that had the authors names plastered across the front, and was told to ignore it by the conference chairs and proceed anyway. Another year I found the paper on the authors website, and again I was told to ignore it. And yet a fourth time, I came across a submission that had been published almost verbatim in a Siggraph course the prior year, and again was instructed to ignore this. And... this year, I heard from the conference chair (Levoy) that he explicitly instructed the committee to ignore any breaches of anonymity.

Yet in our posting, the #1 reason for rejecting our paper was given as our breach of anonymity. Again, I can only marvel in awe at how clever these people are. Not only did they directly ignore what the papers chair said, but they wrote about it in our BBS posting. Of course, you say, we could complain, especially since they wrote about it in the posting. Well they might as just as well wrote "ha ha, we got you" as the main reason. Because there is no recourse, the decision cannot be reverted, and to save face you can bet that we will hear back... "well... there were some other issues too... sorry, try again next year".

These reviewers also know that there is no editor assigned to the paper as there is in almost every other respectible area of research. Typically an editor is supposed to be *non*-anonymous, make sure the rules are followed by the authors and reviewers, and make a final impartial decision on the paper. Well in Siggraph one or two of the reviewers get to play in the shadows pretending to do this - with zero accountability. Sure, you say that there is a papers chair. Well the reviewers also know that one papers chair can not be held accountable for hundreds of papers, and thus it is these secret reviewers with no accountability that get to pull the trigger however they see fit.

Siggraph needs to assign a name to each paper. Someone that can ensure that the reviews are fair, without bias, well informed, and geez that the reviewer at least reads the paper. Actually, I'd settle for looking at the pictures and reading the captions - some of our reviewers on other papers didn't even do this.

The primary should *not* review the paper, but instead but be assigned in a non-anonymous way to the paper as a judge, to make the final decision, and tell the authors exactly why he agrees with the reviewers. And be accountable. "Ha ha we got you" should not cut it. I am an editor for a number of journals, and have to send email accepting and rejecting papers all the time, and it is my reputation, so I am fair, and careful, *and* correct myself when I goof.

And ACM TOG has the same problem. Secret shadow editors. John Hart, you are in charge of that journal. FIX IT! IEEE TVCG does not operate that way. Maybe if you fix TOG, Siggraph will follow suit - maybe they will *have* to follow suit to stay an issue of TOG. Or maybe the powers that be will drop out of TOG then, making it obvious just who the secret system benefits - those illustrious committee members. Yes, of all the students and young people in the area, let's make sure to protect that committee.

Michael,

Please leave this board up. This year's round of reviews is coming out soon, and I'm sure people will be congregating here. I respect your courage and decision to leave this world behind, but maybe you can watch from the sidelines to let those of us still bound to it try to make some progress from the inside (well, inner periphery).

(Do you have the ability to easily filter "lenovo" or "nokia", etc.?)

Hello,

I'm a young researcher in computer graphics without any SIGGRAPH publications; thus, you probably want to stop reading at this point. Thanks for your attention, have a nice day.
 
Also thanks for this very interesting discussion. My personal suggestion to improve the SIGGRAPH reviewing process would be to establish some specific rules (as opposed to "guidelines") for writing reviews. This would require the community to actually state precisely what the duties of a reviewer are. It would also give authors a chance to mention violations of these rules in the rebuttal process and might thus help to improve the quality of reviews in the long term by teaching reviewers (and their students) how to write reviews. For the short term one might also think about a right of authors to reject particular reviews in the rebuttal process if there are clear violations of these rules. (The primary reviewer has to make the final decision and therefore should not be anonymous because he or she is responsible for this decision.)

Secondly, my suggestion for frustrated authors would be: if a video ... oops ... if a manuscript is rejected at SIGGRAPH, don't wait a year but submit the manuscript to a journal that gives better opportunities to discuss shortcomings of the manuscript and requires reviewers to defend their statements. Yes, it will take longer to publish the manuscript this way but maybe not as long as waiting for next year's SIGGRAPH. Also, journal publications are valued higher than conference papers (even if the proceedings have journal status) by many people outside of computer graphics and might therefore be even more important careerwise, at least sometimes.

Thirdly: don't expect life to be fair. As Sheryl Crow put it:

   No one said it would be easy
   But no one said it'd be this hard
   No one said it would be easy
   No one thought we'd come this far


Best regards

Martin Kraus

 Dear all,
 This place is clearly dead and is just filling up with junk advertisement posts. I guess the proper action for me is to close it down very soon assuming there is no new posts...

 Before I do so, one last story: one of CG researchers who got familiar with my ill-fated BRDF paper thought that there is value in it and asked me if it is worth to resubmit it for siggraph 2007. I replied that I am burnt out by whole siggraph system, have no time to do anything for the paper, everyone who makes decisions already saw it and made up his/her mind (so there is zero chance of success), but he is free to do whatever he feels like. So he made an editing pass, improved some images and submitted it. He forwarded me the reviews a few days ago and of course I could not help looking at them.

 It's just so sad how ABSOLUTELY NOTHING has changed in this past year. I am very tempted to post at least two out of five reviews here (reviewers #1 and #5 in the system, if these people read this) as almost perfect illustrations to everything which I was talking about more than half a year ago. I am not doing this only because this would be a violation of unwritten rules and although it would be quite difficult to screw up my CG career more than I already done on my own, this might impact the researcher who submitted the paper. Let me just say that "explain your rating" section, i.e. the actual review, contained 61 word (4 lines of text) in review #1. Prolific writing of reviewer #5 had no less than 96 words, ~6 lines of text. This is on a 10-page paper which I still think is a bit more than total junk (I do make an outrageously bold claim of knowing something about BRDFs). And somehow I do not think I am just unlucky here ...

 Finally, I just stumbled upon Lee Smolin's "Trouble with Physics" book, significant part of which is devoted to exploring systematic issues leading to what author considers a lack of real progress in fundamental theoretical physics over the past 30 years. CG is a much younger field and is really not a science, but it is scary how much of what he is writing about can be directly (or almost directly) applied to this area and echoes (in a much more eloquent way, of course) some of the things I mentioned last year. CG is a fast evolving area indeed - we managed to reach the state of stagnation about 10 times faster than theoretical physics people.

 M.A.

I think there is another bad thing also possible because of SIGGRAPH: almost all the good conferences select their deadlines close to SIGGRAPH. Look at next year's conferences:
SIGGRAPH: deadline Jan 24, Notice Mar 19
Eurographics: deadline Feb 8
Graphics Interface: deadline Dec 18, Notice Feb 19
Computer Graphics International: deadline Jan 5, Notice Feb 20
CVPR: deadline Dec 3, Notice Mar 31
...
If you want to submit your paper to any of them, and do not want to submit one paper to multiple conferences in a same time, and if unfortunately your paper is rejected, there will be few good conferences you can submit it again. :(

 Dear all,

 It has been a while since I had a chance look at this forum (new job is a new job) and the most amazing result to me is that people still post here. Although I remain pessimistic about the final outcome, I start to think that maybe there is some value in keeping this around. If nothing else, people might vent their anger and feel better (if this survives till March when siggraph results come back, I am sure there will a great need for that).

 Unfortunately, as such things go, this froum seem to be quite vulnerable to obvious spam. I just removed several recent invitations to buy junk from here (I hope people do not mind me using administrative permissions in this case) but I am sure it will come back in volumes. Since I can not clean up this place on regular basis, I was wondering if anybody has enough time to take on the task? I would probably have to give my QuickTopic login to this person, so I hope you understand that I would prefer this to be someone people know and everyone can trust. If interested, write to my old but still working account: ash AT cs.sunysb.edu (which is, sorry, not being checked very often either)

 Alternative would be moving this discussion to another place which has some kind of anti-span tools or more restricted posting access.

 Finally, this place can just be kept as it is now. From time to time, whenever I feel particularly nostalgic, I can come here and clean this place in (probably unrealistic) hopes that in the mean time span does not grow to swallow all the rest.

 Any other suggestions ? Thanks,
 M.A.

 P.S. I would like to again thank everyone who had kind works for me (posted here or send via e-mail). I only hope you will be as kind if I ever try to return to graphics.

Some informal descriptions of research challenges in the view of various individuals who have been involved in computer graphics for some time can be found at:
http://www7.nationalacademies.org/CSTB/pro...raphics_papers.html

I am not a leading researcher, but I recall a challenge of "rendering/simulating spaghetti" in an earlier SIGGRAPH proceedings introduction. A little searching reveals that it is part of Jim Blinn's 1998 keynote, which contains such a top ten list:

http://old.siggraph.org/s98/conference/keynote/blinn.html

It seems that some of these have been solved, and some are too abstract to have permanent, satisfying solutions. But it's interesting to see where the priorities were.

Thanks to Michael for starting this discussion thread. It is refreshing to see researchers introspect and question the system which is so very heavily biased towards publishing at a glamourous conference such as SIGGRAPH.

I have a simple question addressed to Dr. Levoy or other leading academic researchers in graphics. Every self-respecting active "hot" scientific discipline should have a set of top unresolved problems that people are trying to solve.

Can you give me a list of the top 10 GRAND CHALLENGE problems that the field of graphics is trying to solve at this point in time ?

Dear Veteran, thank you for your answer,
anyway, I've got good reasons to remain optimistic,
for instance, it is still possible for a perfect outsider to have
a paper accepted at Siggraph (I'm aware of several stories like
that, including my own story), this means that everything is
possible even if you are not a member of a "clique".

However, I agree that the way the PC works is not perfect. For instance,
two friends of mine co-authored a paper that got an average score above
4 and that was finally rejected, was it cornered by a paper-killer ???.
I also once reviewed a paper that was full of errors (recommanded rejection), but that had a well-designed video, and that finally got accepted. That's life, no matter how many reviewers you use,
no PC can be perfect.

Anyway, if the only way to get an academic position is to publish at
Siggraph, is this Siggraph that should be blamed ? I've seen in a post
that Peter Schroeder said that Siggraph needs to be "demistified", I think he exactly gots the point: commitees, students etc... seem to
think that Siggraph is magic. For instance, one of my Ph.D. students
asked a question to a student that did co-author a paper this year
(like: "does your method also work in some specific case"), and the
other folks just answered "Hay, you are at Siggraph !". I think this
is this perception of the conference that we need to change in the
CG community rather than the way the conference works.

All the anonymous posts in this thread give also more strength to the "mystification" (if I put my name here, a white lightning will
disintegrate me, or even worse, will disintigrate all my future
submissions to Siggraph ??)

  Why is Siggraph seen like that ? OK, it has high-quality papers, there
is a good impact factor etc..., but this does not explain everything.
I think this is also due to the fact that it is so big, maybe separating
the tradeshow from the conference would partly restore a more normal perception of this conference.

To end my post, as Veteran said, CG is not the only case, see
this thread about FOCS (Fundamentals of Computer Science)
  http://weblog.fortnow.com/2006/06/focs-accepts-and-movie.html
In particular, have a look at the posts of a folk named Mihai

And at least, since Siggraph is not as prestigious as Fields medal,
we do not have journalists, have a look at that story:
http://science.slashdot.org/article.pl?sid...21/1442239&from=rss
http://www.doctoryau.com/

Edited by author 09-17-2006 12:33 PM
I know it won't happen, but please let's not even consider having a winter SIGGRAPH!!!

We need to realize that there's life beyond SIGGRAPH. We have journals, I3D, EG... I know a lot of people in rendering that, had they just one shot at attending a conference, would rather go to EGSR than SIGGRAPH, and that's pretty healthy (publishing a paper is a different issue, though...). The winter SIGGRAPH idea will just reinforce the impression that it's SIGGRAPH or nothing. All borderline summer rejections would of course become winter submissions. All winter rejections would immediatly become summer submissions again. Tenure committees will scan CV's looking for SIGGRAPH papers only, even much more than now (hey, you now have twice the chances, how come you don't have 2 or 3 SIGGRAPH papers per year?). We either demistify SIGGRAPH or, as Peter Schröder said during the Town Hall Meeting, we'll be "eating our own children". Great intervention, Peter, BTW.

A very good idea I found a few posts below is the restrictions on video. The level of refinement expected, profesional voice-over and all, is reaching more unchartered heights of absurdity by the year. Having a committee member say "the video looks so good we have to find a way to get this paper in" is a pretty serious indication that we're treading the wrong path. Force the video to add useful information beyond what is in the paper. Don't let them be flashy showoffs.

As for reviews, I have nothing to say. My SIGGRAPH reviews have been fair so far, both for courses and papers, even though I of course didn't always agree.

Bruno states:
"Do you think that simply expressing your views in this forum can
have a bad impact on your careers ? Do you really think that
the situation is *that bad* ?"

Yes Bruno, that is exactly what I think. I hope your optimism turns out to be valid, but I doubt it. Not that graphics is different in this regard than any other field...

Edited by author 09-07-2006 08:10 AM
I'm also a new researcher, and I would also like to add a few thoughts to the existing list of excellent observations and suggestions.

First of all: On a paper-by-paper basis, the individual acceptance decisions at Siggraph probably are not always fair. This is as true for Siggraph as for any other scientific conference. Such is the nature of peer reviewing; the evaluation of the novelty and usefulness of a submission is in the eye of the beholder. Does this mean that the paper selection process at Siggraph is worse than in other places? Personally, I do not think so. The reviews I got for my Siggraph submissions so far have always been detailed, and it was clear that the reviewers spent a lot of time reading the paper, pondering its contribution, and writing the review. Although I did not always agree with the conclusions the reviewers had come to, the diligence with which the reviews had been prepared definitely matched or even surpassed the level of quality I have seen in many other places (conferences and even scientific journals).

Still, there are some aspects of Siggraph which I also find annoying, and although these have been mentioned before, I would like to put an emphasis on them:

1.) Computer vision / image processing papers. Here, I disagree with post #75. I do not think that pure image processing and computer vision papers should be presented at a computer graphics conference. The number of "paper slots" at an event like Siggraph is small enough already. You have to keep in mind that every image processing paper which is accepted at Siggraph possibly leaves another researcher in core computer graphics, who may have done very good work in his field, frustrated and disappointed. I do agree that many of the CV/IP-oriented papers are interesting and useful for computer graphics. The same, however, is true at least to a certain degree for developments in a.i./machine learning, hardware design, software design and programming methodology, networking, sensor development and so on. I am pretty certain that nobody here would support a pure "software engineering" session at Siggraph. So why should we have one on image processing? Computer graphics is the process of converting abstract scene representations into images, as well as anything which is more or less closely related to this. This is what you are taught in your Introduction to C.G. course, and I think we should stick to this definition.

As a random example, "Fast Median and Bilateral Filtering" presented a technique which converts one image into another. Is this a good paper? Yes. Is it useful? Yes. Will it probably be helpful for future computer graphics research? Yes. Is it a computer graphics paper in its own right? Definitely no. Not even in the least.

2.) The video mania. The custom of having to send a video demonstration along with your paper submission is really getting out of control. The official Siggraph submission guideline urges researchers to send a video demonstration, and most submissions are accompanied by one. These videos are getting more and more elaborate, almost to the point of being ridiculous. These days, a Siggraph paper video has to have professional-quality voice-over, must be several minutes in length, and should have professional-grade special effects and editing. This leads to a number of undesirable consequences:

- It is not rare that researchers have to spend as much or even more time on the "presentation" aspect of their Siggraph submission as on the research itself. I have experienced this myself and heard similar stories from several colleagues. This means that highly qualified and motivated researchers in our field are forced to spend a considerable amount of time each year on this task, which is certainly not helpful for the overall progress.
- While the video demonstration is only meant to accompany a paper, its actual significance is continually increasing. As a reviewer, most people have a look at the video first because watching a video is easier and more entertaining than reading a multi-page scientific text. This, however, means that the reviewers have a strong opinion about the submission before they even read the actual paper. Given that a reviewer evaluates several papers, those with comparatively polished video demonstrations stand out, regardless of their actual scientific value. This is damaging to the scientific quality of the entire paper program! We should select the best research achievements, not those submissions with incredibly elaborate video demonstrations.
- Most people in English-speaking countries are not aware that the emphasis on the accompanying videos represents an additional unfair disadvantage for non-native English speakers. While it is relatively easy for English-speaking groups to produce a good voice-over, which is now an absolute necessity, this is very difficult for other researchers. Any verbal explanation recorded by an author or other lab member who is not a native English speaker will typically have some sort of accent. This makes the resulting video demonstration appear unprofessional and second-grade compared to the tv-commercial-style videos submitted from English-speaking countries. This represents an additional obstacle, considering that producing a high quality scientific text in a foreign language already is challenging enough! As a matter of fact, I have more than once had the impression that certain research groups team up with colleagues from English-speaking countries mostly to get "language support" for presentation aspects like these.

While it would be silly to demand a ban on accompanying video demonstations, I think Siggraph should impose certain limitations on the submitted videos. These would significantly reduce the amount of time required to produce them and increase the fairness of the process for foreign researchers:

- Videos should not contain an explanation of the entire algorithm or method! This is what the paper itself is there for.
- Only results should be presented, in the sense of an animated version of the images in the Results section of the paper.
- NO AUDIO TRACK AT ALL! No music, no sound effects, and no voice-overs.
- Videos should be short, maybe 1 or 2 minutes at most.
- Maybe it would even be possible to somehow supply some sort of a "video styleguide" with defined title screens, fonts, video resolution etc. This way, the presentation of all videos would look the same. People would not have to waste time on editing and special effects. This is the reason why there is a styleguide for the actual paper, so why should we not apply the same principle to videos?

I'm a new researcher in the field (I've been working for a tenth of
years, published a couple of papers at Siggraph and tertiary-reviewed
approx. 5 Siggraph papers per year since 98).

First, there is something I want to say:
There is something that strikes me, most people posting in
this thread remain anonymous. What you guys are afraid of ? Do
you think that simply expressing your views in this forum can
have a bad impact on your careers ? Do you really think that
the situation is *that bad* ? I do not think so!

I've got an experience of Siggraph from both the paper author and tertiary
reviewer point of views. As a paper author, most of the reviews I
received were well detailed and fair. The only thing I regreat is that
the only way of getting a paper accepted is to present a completely
finished "product" (including the "TV commercial") rather than a new
idea. This sometimes closes the door to fresh ideas, but fresh ideas
can also be submitted to other more specialized conferences / journals.

As a reviewer, I'm always trying to do my best. It's a hard work,
in most cases you know where the paper comes from, and you need
to avoid getting influenced by the fact that you know who the authors
are (it's especially true when they are well renowed people).
I think that the anonymous submission is only effective for newcomers
(but this is not specific to Siggraph)
Then, I think that it may a good idea to involve the tertiary reviewers
in the rebuttal process, as suggested in a previous post. When you write
your review, you do not see the other reviews, and sometimes discussing
with the other reviewers and authors may help better explaining / detailing
the reviews and improve the quality of the paper selection process.

I see some people argued that papers about image processing/computer vision should not show up in Siggraph; on the contrary I think we should thank Siggraph to accept these papers. I try to explain the situations of these two communities, so people can understand why so many papers about IP/CV submitting to Siggraph. I think on this topic, people should blame conferences of IP/CV instead of Siggraph. There are many things in other communities worse than Siggraph in graphics community.

Let's take a look at image processing community. To be honest, the quality of most conferences directly related to IP (ICIP, ICME...) is much lower than that of Siggraph. The acceptance rate is higher (so is the acceptance number), the review process is worse (author name is visible to reviewer), and usually you can only have 4 pages to describe your work. Most people will try to find a better place to submit their papers if they really do a nice job.

For computer vision, it is a different story. The acceptance rates of top conferences (like ICCV, ECCV, and CVPR) are amazingly low (see the website in post 74). Few people will check the paper list of CV conferences, and it means few people can see your work (Here I mean normal people and investors). So here is the problem: Siggraph has higher acceptance rate, Siggraph has more attendee. I can easily put the term “image-based xxxx” in my paper, and even more, I use OpenGL in my demo program! Why can’t I submit my paper to Siggraph?

Ok this is too ironic, but I really think there are too many "image-based" now. Most papers are great and they can be accepted even without that term. As long as the paper is good and related to human perception, I don’t think Siggraph should exclude any specific area.

It is not just in Siggraph but in almost all computer graphics conferences the acceptance rate is not very high. I review for many conferences, I see many papers with good ideas rejected. I think that this is very healthy for the computer graphics field.

I discovered the following page that can give you an idea about acceptance rates of many graphics conferences.

http://vrlab.epfl.ch/~ulicny/statistics/

As you can see, acceptance rates for almost all conferences are lower than %40. This list does not include many graphics related conferences such as Solid & Physical Modeling. But, if you even look at those conferences you will see that acceptance rates are also at most %40.

I think this is an important information for the tenure track faculty or the PhD students who consider academic careers. Siggraph is not only venue for publication of your computer graphics research. There exists many conferences you can publish. It also mean that if you send your paper only one conference it will be rejected with at least %60 chance. You should not take it personal. Read reviwers' comments and improve the paper and send it to another conference.

One good thing about high rejection rates, reviewers are friendlier (seems controversial but that is my experience). If the reviewers know the field, they usually try to be helpful. After carefully respond the reviewers' comments and sending the paper relevant conferences, if your paper is still rejected (say more than two conferences), you can safely assume that the paper is in the lower 10%. Then, you can retire the paper. But, this still does not mean that the ideas in the paper is not good. There may still be a good ideas in the paper but the paper is probably not a good fit for existing graphics research conferences.

The main problem with retiring the paper is that you may lose ownership. Even if the paper is not accepted, you may still want to say that you are the first to find these ideas. The simplest solution is that as soon as you finish the paper, make a technical report and put pdf file to your homepage. Don't be afraid of people who can steal your idea. It is very hard to reproduce the results in a short time.

If a paper is rejected only from Siggraph, it is not a good idea to retire the paper. Submit it at least one more conference. It is better not to make visuals Siggraph related. So, you do not have to create new visuals for new submissions.

An important concern for academicians is the effect of publishing in Siggraph on tenure & promotion. For tenure & promotion, outside reviewers are tenured graphics faculty. They have a good understanding of the field. I do not think it will matter much where you publish. They will mainly look at the quality of your publications. I have been in Tenure & Promotion committees for several years, I can say that for P&T committees and university administration, Siggraph does not mean much. As far as outside reviewers like your publication record, you are fine in terms of publication.

I think the real problem is not really tenure & promotion or publishing. The real problem is that perception of the Siggraph among graphics practitioners and graduate students. Without a Siggraph exposure, many interesting ideas cannot find market easily. Better students choose to work with faculty who frequently publish in Siggraph. This is really a problem for ambitious researchers who wants to make a quick impact. But, I believe that strong ideas will eventually be known wherever they are published.

Siggraph cannot find any real solution to this perception problem. It is hard to change our perception. It may even be better if Siggraph does not function well. Then, the other conferences will start to be perceived more important. In fact, since it is possible to meet with more people with similar interests, smaller conferences with high rejection rate will be much better than Siggraph for a tenure track faculty.

So, to me nothing is broken in the system. The real problem is our perception of Siggraph.

This is a reply to "insider". I do not think the "few really good papers get in". My own stats are that I have around the same number of my papers in SIGGRAPH and really good papers submitted to SIGGRAPH. The overlap of those groups is at most one. SIGGRAPH is simply a mess. Where the weights are on incompetence and corruption is debatable. I do agree that people put a huge amount of honest effort in. I also think that effort is worse than wasted. SIGGRAPH is a cancer on the community. It is like the Iraq war. The sooner we take an honest look in the mirror, the better.

Edited by author 08-29-2006 08:03 PM
Computer Graphics research has, sad to say, been much less exploratory, less cutting-edge, than it could be. There are relatively few radically new ideas. The law of diminishing returns places an absolute limit on what can be obtained by reification of prevailing theory or refinement of prevailing techniques.

Normally, in any arena, you can expect a totally revolutionary concept every few months. Of those, a few may emerge into the mainstream with perhaps one a year gaining dominance in some field. Of those that succeed, a few will become so dominant that they become synonymous with that field for a while.

You should also expect an idea to mature in roughly ten year cycles, starting from pure theory and evolving to become a polished commodity item that is simply a part of how things are. It took about six such cycles to move from written theory on what a computer was, along with prototype proof of concepts, to machines that were almost ubiquitous.

How does the field of Computer Graphics live up to this? Very well, for the most part. It is relatively easy to track the history of image rendering, display technology, scene description, graphics languages, graphics acceleration, etc. There are hiccups, where an idea doesn't survive for any one of a number of reasons - including political one, but in general progress has followed the path one might expect of any technology.

"Most part"? Well, over recent years, some areas are not progressing at the rate that might be expected, and new areas are not emerging at the rate one would anticipate for a field not short of intelligent and creative people. Based on both the typical and observed rates of progress in computer graphics, this is what the state of computer graphics should be:

• Research "raytracers" could be expected to use the wave theory of light, support interference and treat frequencies as relatively continuous and shiftable. This allows for the correct handling of edges (where light need not travel in a straight line), diffraction, the impact of the difference in path lengths, the ability to visualize in different ways*, etc.
• *I'll add this as a seperate item, it is so important. Not everyone has three colour cones. Some people have only two, some people have four (so-called tetrachromats). Existing rendering systems are largely incapable of optimizing for such vision without a lot of hard-coding. Nor is there any display technology optimised for such users.
• Talking of display technology, it seems to be either/or. Either you have a higher than normal resolution display (eg: the double-width systems fashionable in CAD), High-Dynamic Range (eg: 32-bit floating-point colours), or an extremely high refresh rate. Sometimes the problem is refresh rate, sometimes it's mechanical, but it's always avoidable.
• Commercial "raytracers" would be expected to replace both rays/cones and radiosity with a statistical function showing percentage reflected for a specific channel (red, green, blue), providing diffuse and direct reflection in a single operation. Discussed in SIGGRAPH in the 1980s, I seem to recall.
• Full "CAVE"-type virtual reality should - by now - be buildable in kit form in the garage, in much the same way early radios, televisions and computers were in their respective developmental eras.
• "Line of sight" calculations (for clipping, etc) are correct for engineering purposes but are incorrect for photorealism, because a plan view is not the same as an eye's view.
  

Modern computer graphics are really not significantly further on than they were 5-10 years ago. Progress has occured, but at a snail's pace. What is now research should have become a standard feature. What is merely imagined should have become a hot-ticket research item. Oh, sure, there may be counter-examples, but my point is that it's not happening on nearly the scale one would expect or at the pace one would expect. The past 5-10 years has been essentially stagnant, and that is NOT good.

Edited by author 08-29-2006 06:03 PM
I am an artist in the Movie industry and have a group of outcast talents forming a company. We were outcast due to our effectiveness and the hurt egos that our success engendered on an upcoming project. We haven't been to Siggraph in the last few years because we believe that E3 has a better look at what's Happening for the future, That said, As I am starting a company I may have to present or be present at a future Siggraph so my real name is not posted here. In any event we are trying to do breakthrough work in Mental Ray on Human Skin using what artistic skills we have, with very little in the way of math to back us up. I wonder if The anger that drives us to "Show these big companies a thing or two" also would drive you in a similar way to bypass the academic route and instead band with us to make history in spite of them.

Contact me to establish a dialog. MentulRaymond@radiofree.tv

Having served on the SIGGRAPH papers committee multiple times and having been its chair in the past let me provide my own "inside view" of what goes on. First, the chair volunteers a good portion of 2 years assembling and chairing a committee of 50 peers. The committee voluteers most of 2 months of their lives to reviewing and soliciting reviews and then ultimately deciding the fate of 500 papers submitted. Tertiary reviewers contribute untold hours of their time. Of the papers chairs I have witnessed work, every one of them has, to a first approximation, done everything in their power to create and run a fair process. Of the 50 committee members, I would say on average 49 out of the 50 are equally committed to performing in an honorable and fair way. They do the best they can given their often diverging interests, knowledge, and personalities. They should all be applauded (except that 50th member perhaps...).

That said, the process is very noisy and often inconsistent. Many good ideas have been expressed below by others on how to make it a better process, but none of the ideas will really "solve" a lot. In general, the few really good papers get in. The really bad papers get rejected. But as in most distributions, the bulk of the papers are somewhere in the middle. There is no way a group of 50 humans can, in any truly objective manner, separate the pretty good, from the good, from the almost good. But those 50 committee members and the hundreds of tertiary reviewers do their best. If you could be a fly on the wall at the committee meeting, I am certain (almost all) of you would be impressed with the honesty and integrity and diligence with which the work is done.

Keep the good ideas coming, but please rest assured that those in charge of the process are both listening to you and trying to do their best.

Edited by author 08-29-2006 05:04 PM
I think a large part of the problems seen here with SIGGRAPH is due to its cross pollination with the arts community. To put it in perspective, one should take note of the artistic pretention required to get something into an art gallery or museum, regardless of whether or not it is computer related.

One specific example I saw a year or so ago at the Fowler Museum in LA. The museum was doing a themed set of exhibits on rice and culture. There was an interactive exhibit that used videoprojectors and cameras, that would track the motion of people in front of a large wall, record some of their movements and then correlate the pattern created against some of the patterns in rice DNA. My first thought was WTF? What does the movement of a visitor in front of this wall have to do with rice DNA? But then I read the exhibitors description and realized the significant-sounding mystic-feelgood-psychobabble description had really been tuned to a "t" and by an expert (wish I could make up BS like that!)-- from just reading the one page description on it I could see how the concept of the piece would have won many reviewers over with an "oh wow-- sounds like it has important observations, and is right on topic" response. My guess is the interactive construct was already set up for some other purpose, and all that was done was to re-explain it so that it seemed integrated with the theme of the exhibit, and made "unique observations about the human condition and culture" (or some such, that's about the best I can do with that stuff-- my heart is never in it) in order to improve the chance for acceptance...

I think this sort of thing is not atypical in art settings-- overall concept and presentation are often far more important, certainly than aesthetic or technical prowess. Art is about *selling* something as art, and often without the sideshow pitch and gallery setting it wouldn't be recognized as art at all. SIGGRAPH some time ago chose to follow that example, so there's no real surprise here IMHO...

Art is about aesthetics, which is largely subjective-- and where subjectivity rules, objectivity is left behind...

SIGGRAPH is like other major conferences in that they receive a very large number of papers for review, and can accept only a relatively small number of them. The program committee acts as a gatekeeper for the themes and areas, so it is natural that certain areas will be favored because they fit the themes. The result is that even very good papers do not always appear in the conference. Even if people all acted in the best way (which they don't), the conference would still have to reject good papers. Then the essentially stochastic nature of review can explain why not all papers that appear in the conference are the best ones. I say all this not to defend the practice, nor even to apologize for it, but just to explain what actually happens. It is not a conspiracy, but it is really unfair, and probably inevitable. Resistance is futile. SIGGRAPH is not even the worst in this respect. Ask a neural net researcher about NIPS, for example.

I'm going to jump on the bandwagon and thank Michael for expressing his views. I have to say that I agree with several of his points, and several made by others in this forum.

I have had one SIGGRAPH paper accepted, and another rejected. The rejected paper deserved to be rejected because, while the ideas were good, the paper was in poor shape.

The accepted paper was much more of an interesting study. The paper recieved mediocre scores between 2.5 and 4. The average score was 3.3. Among the reviewers was one who knew our work - his was among the lowest scores. Many of the other scores were due to the reviewers not knowing the background work, and therefore not thinking that the work was new (ie. it was so clear and simple, how could it not have been done before?).

In the end, I believe that the paper was accepted despite the mediocre scores due to the nice images in our paper.

This is an example of the review process going wrong both ways. While I think it's nice to have pretty pictures in graphics papers, they should not be the sole reason for acceptance. I believe that the paper should have been accepted based on it's technical merit. In fact, the paper generated a lot of buzz, produced a lot of interest, and there have been several follow up papers in the short time since the work was presented (which justifies acceptance).

Now for some suggestions (some of which have been mentioned before):

*There should be at least one editor/reviewer who is responsible and can be contacted by the authors.

*Jonathan's idea of a winter SIGGRAPH is a good one... I'd like to take that even further:

*Eurographics has several workshops/symposia for special topics. If SIGGRAPH had such counterparts (eg. The SIGGRAPH workshop on Rendering), this would give a home to many high quality submissions. Moreover, they would provide a high quality delivery venues. I have been very impressed with EGSR, and have learned more about new rendering work attending one EGSR than 5 SIGGRAPHs.

This doesn't mean we eliminate SIGGRAPH. It's still good to get a smattering of different topics (vision, image processing, rendering, animation, etc...) in one venue, but we should provide more venues. Since they would be attached to "the SIGGRAPH name", presumably they would inherently carry weight in the community (which is perhaps sad).


One last point I would like to address: Many people have suggested to start sending submissions elsewhere. This is fine for professors who are already established/have tenure. They want to make a statement, and that's fine. This is a problem for new professors and graduate students. We need to published our best work at the best conferences to survive in academia. It's not so easy for us to just pick an appropriate venue in liu of SIGGRAPH because it can lower our percieved worth. This is why it's very important to help solve the problems of SIGGRAPH, and possibly open up new high quality (where the quality is also percieved to be high) venues.

I personally believe that EGSR is top notch, and that some of the most useful and visible(though not most theoretically novel) work is in JGT. But I am hesitant to submit to these venues because there is more recognition for publishing in SIGGRAPH.

Same Anonymous Coward as below...

I think that a major problem is that most people outside of one's specific field judge the quality of a researcher by the weighted sum of the publication list (higher weights for top-tier conferences). However, I think that it is more important to try to figure out if the researcher has made significant contributions to the field. Once you've got a topic, it's quite easy to incrementally re-spin it a few times over a 1-2 year period, and maximize the publications; however, doing this prevents you from working on something more novel and interesting in the meantime. It stifles innovation on a personal level.

For example, suppose that I discover some new application domain where an NP-Complete problem of great importance needs to be solved. Something big, along the lines of register allocation in compilers.

Paper #1 - introduce the problem. Model it as an integer linear program or satisfiability problem and solve it using a commercial tool. Minimal implementation effort needed.

Paper #2 - Solve the problem with a backtracking search. Because you know something about your application domain, you can find a nice pruning criteria to speed things up a bit compared to #1.

Paper #3 - backtracking search is exponential in the worst case, even with your pruning criteria. Now let's solve it with an iterative improvement algorithm such as metropolis, simulated annealing, or a genetic algorithm.

Paper #4 - speed is important. Resolve the problem with a polynomial-time heuristic based on a network flow or dynamic programming.

Paper #5 - #4 is still too slow. You've found an online application of the problem (e.g. just-in-time compilation), so now you need to focus on speed rather than solution quality. Greedy heuristic.

Paper #6 - You've analyzed your application domain, and found that 50-70% of all problem instances belong to a special class that can be solved optimally. (say, graph coloring on interval graphs) New paper.

Paper #7 - It's time for an approximation algorithm.

...

This is how you maximize your publication list. Beat the same problem to death. Of course, we know that you can do #1-7 for ANY NP-Complete problem., so at some point, you're not really adding much to the community (although #6 could be valuable, especially to industry).

Then, find some variation of the original problem with a different objective function, and repeat #1-7. For your typical embedded/CAD example, maybe the first iteration was optimizing performance, given some area and energy constraints. The second iteration, you optimize for area, given performance and energy constraints. The third iteration, you optimize for energy...

THIS, my friends, is how you get tenure. Taking the John Nash approach from A Beautiful Mind--find a really really really important problem and solve it--is a real crapshoot. Either you prove or disprove P=NP and get uber-famous; or you don't get tenure. Typically, this is a matter of being in the right place at the right time, and being smart.

Edited by author 08-29-2006 05:24 AM
I believe that there is a similar problem in the architecture community with ISCA. If you don't believe me, check out the ISCA Hall of Fame:

http://www.cs.wisc.edu/~arch/www/iscabibhall.html

The same people repeatedly publish there every year, and its hard for newcomers to break through.

The difference, however, is that there are some alternatives in the architecture community, although they generally aren't considered to be as important as ISCA. For example there are MICRO, HPCA, and to some extent, ASPLOS if you have the right application.

Secondly, any architecture researcher who is truly in a bind can switch to embedded applications quite easily. That opens up the possibility of publishing at DAC or ICCAD, which are generally larger (similar rejection rates, but significantly more overall sessions--in other words, if your paper is of high quality, they are likely to find a place for you).

But, back to ISCA, one of the big problems there is that the papers are not reviewed blindly when the paper is discussed at the table.

---------------------------------------------------------- ---------------

As far as getting your ideas stolen from someone on the committee, that is a definite risk. The best advice that I can give anyone is to assume the worst. If you believe in your paper, and it gets rejected, publish it online as a technical report ASAP. You can usually republish technical reports in conferences and journals anyway. This way, if someone tries to steal your idea, you can wave your technical report in the air and cry bloody murder. It is highly unlikely that the "stealer" will be able to get a submission out and published before your post your technical report.

----------------------------------------------------------- ---------------

I've heard of a number of situations where two papers proposing similar techniques have been submitted, and the one who was authored by a program committee member gets in. I have personally been involved in situations where I have developed the same result (simultaneously and unknowingly) as someone else. In this case, I think that either both papers have to be accepted (or rejected).

---------------------------------------------------------- ----------------

The community needs to find a way to promote collaboration rather than competition. The Berkeley group and the MIT group that have similar research topics should be encouraged to work together, not against one another.

www.tipcell.com sell: kyocera 2135 housing, kyocera 2235 housing, kyocera 3225 housing, kyocera 3245 housing, kyocera 3250 housing, kyocera kx17 housing, kyocera kx9 housing, kyocera kx13 housing, kyocera kx440 housing, kyocera kx494 housing, kyocera K433L housing, kyocera K434N housing, kyocera KE433 housing, kyocera KE433C housing, kyocera KX 434 housing, kyocera K7 housing, kyocera K9 housing, kyocera K433 housing, kyocera K434 housing, kyocera kx1 housing, kyocera kx1i housing, kyocera 1155 housing, kyocera se47 housing, kyocera 1135 housing, kyocera 2119 housing, kyocera 2235 housing, kyocera 2345 housing, kyocera 3245 housing, kyocera 3250 housing, kyocera 5135 housing, kyocera 6035 housing, kyocera 7135 housing, kyocera k10 housing, kyocera k112 housing, kyocera k404 housing, kyocera k433 housing, kyocera k434 housing, kyocera k454 housing, kyocera k484 housing, kyocera k490 housing, kyocera k493 housing, kyocera k494 housing, kyocera ke413 housing, kyocera ke414 housing, kyocera ke424 housing, kyocera ke424c housing, kyocera ke433 housing, kyocera kx1 housing, kyocera kx16 housing, kyocera kx2 housing, kyocera kx424 housing, kyocera kx433 housing, kyocera kx434 housing, kyocera kx440 housing, kyocera kx444 housing, kyocera kx5 housing, kyocera QCP-2035 housing, kyocera QCP-3035 housing, kyocera QCP-6035 housing, kyocera S14 housing, kyocera SE44 housing, kyocera SE47 housing
mail: sales@tipcell.com
name: keiresing
MSN: keiresing9988@hotmail.com
http://www.tipcell.com
ICQ: 239470786

Edited by author 08-19-2006 01:44 AM
My somewhat off-topic response is to message 61 (industry person 2). Upstill's 1985 thesis was on tone mapping and was ten years ahead of its time not only in recognizing HDR as an issue, but in taking a very sophisticated approach to dealing with it. Unfortunately it is only available from University Microfilms. And I would not bet against something else I haven't heard of being first! Some might argue it was Stockham. For a more well-known work that I would argue defined the field for graphics was Tumblin and Rushmeier's work in the early 90s. Clearly you had to be doing HDR in order to do tone mapping!

Obligatory SIGGRAPH comments: there are good reasons other fields have conferences with abstracts, and journals with articles. That Genie can probably not be stuffed back in the bottle very easily...

Just an aside, the story several posts below is not the first time that the
SIGGRAPH academic side has failed to mention prior work in the industry.
Academic people talk as if they invented the HDR idea, but actually it was
introduced by Glenn Kennel in 1993, as part of the Cineon file format (and
Cineon compositing system). Both were in wide use in the mid-90s, several
years before SIGGRAPH papers had heard of the idea. The Cineon compositing
system died out several years later (though other systems have picked up the
idea), but the file format lives on as "DPX", which is a superset of the
original format, (and in practice is usually the same).
No doubt, the academic side has taken this idea much, much farther.
But it's never mentioned where it came from.

Cineon/DPX is 10 bits log per channel, with an adjustable "white point"
usually at 685 on the 0..1023 scale, so values above 685
are the HDR values. Since this is a log scale, that is an adequate
range, since for typical purposes the >white values will be clipped
after processing anyway.



From wikipedia,
   "Pixel values above 685 are "brigher than white", such as the sun, chrome highlights, etc"

Now let's count the days until someone edits this wikipedia page.
But watch out, wikipedia has an revision history.

I think the idea of having online reviews/discussion will not help much
(though it will not hurt either).

The problem is that your paper is entirely reviewed by your direct
competitors, who have a shared interest in having your work rejected,
and there is no process to hold them accountable to producing a fair review.
Simply having the discussions online do