3-Sweep: Extracting Editable Objects from a Single Photo [video]
youtube.com
youtube.com
In this case, it’s fair to say the machine, by analyzing pixels, can’t figure out perspective very well. The human can do that just fine, given an interface mechanism.
The machine is good at detecting edges and seeing similarity between pixels. Given hints from the human that ‘this point is within an object’ and here is the perspective, the machine can infer the limits of the object based on edges/colors and project it into 3 dimensions. Amazing.
This is just in case you want to throw a few upvotes their way for being first. This also illustrates that late night (PDT/UTC -8) posts don't get a whole lot of votes and proper timing is crucial to getting lots of votes.
Personally, I'm just glad to see this video finally getting traction. It really is such a cool demo. It even stands out in the field of consistently high-quality SIGGRAPH demos. Can't wait to read the paper!
It's weird that it has received quite a few votes each time and never made it to the front page. Was it a timing issue (late night, early morning, non-American hours) or is YouTube "weighted down" somehow?
This is indeed magic. I'm so happy to live in this age, and be part of the "Sorcerers' Guild".
Also, with the shiny objects, could you specify the material properties and have it "back out" the reflection such that the reflection was recomputed as you moved the shape around?
Forget the Photoshop stuff, this needs to be integrated with 3D printing immediately.
Spit out a design file into Tinkercad[1] for some minor adjustments and BAM, you've made a printable 3D model.
No, it's not.
How many people connect through social networks
That's roughly quantifiable. FB has roughly 1.15 billion users, not sure of its daily use stats. Some numbers: http://expandedramblings.com/index.php/resource-how-many-peo...
which is an obvious benefit
Now _there_ is a questionable assumption. Given that increasing numbers of people are _leaving_ FB in saturated markets, and peak membership seems to top below 50% of the population, there seems to be a limit. And I could turn up studies showing negative effects of social networking / media saturation ranging from social isolation and depression to broken marriages and lost jobs to health and life-expectancy loss due to inactivity.
How many people benefit from research papers about 3d model generation from photographs?
First: a false equivalence and shifting goalposts. Your initial claim was "most of the academic research".
Secondly: academic research covers a huge range of areas, from improved health and diet to better machines and alternative energy sources to faster and more accurate computer algorithms.
Third: what you see as a useless toy has some pretty evident applications that I can consider. Attach this method to a 3d CAD/CAM or printing system and you have manufacturing or parts replacement from a 2D photograph (AutoDesk has demonstrated similar modeling/capture systems but based on multiple images, but these can come from any camera). Art interpretation, archaeology, X-Ray modeling, geological imaging, and astronomical applications come to mind. There might be applications in protein folding or other microscopic imaging applications.
And the beneficiaries of such technolgies could extend far beyond just those who are currently plugged in.
Blindly claiming social media vastly exceeds the value of such research fails to pass the most casual of sniff tests.
Your analysis focuses only on Facebook. Of course people are leaving Facebook. But is the total user population of all social networking apps decreasing? I doubt it.
> First: a false equivalence and shifting goalposts. Your initial claim was "most of the academic research".
Poor phrasing on my part. My original goalpost was "the academic research like this," which is admittedly vague. What I meant was research projects focused on image processing and interpretation.
> Third: what you see as a useless toy has some pretty evident applications that I can consider.
I don't see it as a useless toy. I just think it's far less useful than social networking services, which have a very practical obvious benefit.
> Blindly claiming social media vastly exceeds the value of such research fails to pass the most casual of sniff tests.
It's not a blind claim, it's what I feel is an extremely obvious claim.
It's reasonable to question ALL assumptions.
Your analysis focuses only on Facebook.
No it doesn't. I pointed at FB as the largest of the present SNs, but referenced other SNs as well. FB is a leading exemplar of the field. My use of it isn't intended as exlusionary of other SNs.
My original goalpost was "the academic research like this,"
Which largely moots the rest of the argument. Though as I pointed out, "research such as this" actually does pose some reasonably interesting and useful applications. We can argue over those magnitudes, but I'll stick with my initial assessment that the net benefits of such research are likely to be high.
Also, but narrowly identifying what you feel is and isn't valuable research, you're sharply skewing the results to your favor. It's as if I said "but I meant by 'social media' 4Chan and HotOrNot".
it's what I feel is an extremely obvious claim.
And it's what I feel requires citation.
Which you've failed to provide, being rather more inclined to engage in rhetoric.
HAND.
This technology is awesome. If it's as user friendly as they make it looks, I could see a lot of application for that!
For example, I have only tried my hand at 3d modelling once or twice (and sucked at it enough to give up), but just watching this I feel like I could model vases and lamp posts with a bit of practice.
These guys/girls know what they're doing.
Indeed, and it's very impressive work.
It makes sense that this is the case, because this system is doing edge detection with fairly strict constraints: the edges must match the outline of a fairly simple shape which you roughly know the size and orientation of. That seems like it's inherently going to yield better results than completely-unconstrained edge-detection as in photoshop....
I wonder if it's just a coincidence, or whether the mega-bucketloads of money the film industry throws at CGI are a major factor in funding related research even in academia?
Sure, understood.
The thing is, I imagine film VFX guys are already doing this kind of task—making 3D versions of real objects from the movie and doing CGI additions from them—and tools like this (with, as you say, refinements) could be a great help in speeding up that process...
Edit: I am aware that Photoshop has some of this available. I've not played with it so I don't know how they compare.
The impressive thing here, imho, is the seemingly effortless and seamless transition and replacement. The background is fixed and the surface texture is stretched in what seems like real time.
I vote for this to be used with 3D printer
however, it seems strange in the first example how mountain ranges appear where none were before... how did the algos know to put it there?