Is there a reason many of these crazy image processing technologies never seem to have actual demos or releases? The only exception I can think of it the "smart erase" idea, which has been implemented in Photoshop as well as Gimp.
3-Sweep: Extracting Editable Objects from a Single Photo [video]
21–30 of 75 posts
Re: 3-Sweep: Extracting Editable Objects from a Single Photo [video]
#22This is the single most impressive example of image processing I've seen to date.
I think they patchmatch algorithm they use to fill the background is cooler, to be honest. Check out their video: https://vimeo.com/5024379#at=0
Re: 3-Sweep: Extracting Editable Objects from a Single Photo [video]
#23The key here is really complementary use of ‘what humans are good at’ and ‘what machines are good at’. 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 her…
The perspective analysis is done pretty darn well by the machine in these examples.
Re: 3-Sweep: Extracting Editable Objects from a Single Photo [video]
#24I want this + i❤sketch now, but unfortunately I suspect that jumping up and down and shouting isn't likely to help. http://www.dgp.toronto.edu/~shbae/ilovesketch.htm
Thanks for the reference. Are you aware of any (commercially available) handdrawing to CAD software? I need to do a bunch of relatively simple figures for my thesis, and it would be much easier if I could just use handdrawings and than transform them to professionaly looking figures.
Re: 3-Sweep: Extracting Editable Objects from a Single Photo [video]
#25This is amazing. My first thought is this could allow F1 teams to get a much better idea of what new packages their competitors are bringing to races early on just by looking at photos and video footage and modelling the new parts.
This doesn't tell you any more about the contents of the photo than what is already visible. It can't actually see what's behind an object, it just synthesizes a plausible fictional fill.
Re: 3-Sweep: Extracting Editable Objects from a Single Photo [video]
#26Also awesome is that it handles the background replacement so well. This could also be used to just remove an ugly lamp post, telephone pole, etc from an otherwise good photo. (assuming you can remove objects and resave the image) 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.
If you're just removing part of the image after cutting around it with a tool like this, having the object interpreted as 3D isn't really going to be of any benefit. 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.
Yes... I know the 3D part is the more impressive part. But I was also impressed with its ability to back fill the background.
Re: 3-Sweep: Extracting Editable Objects from a Single Photo [video]
#27This indeed is very impressing and I see the how much work passion is into this project.But I still have to say it almost only about round or cylindrical objects, there is still a long way to go
Re: 3-Sweep: Extracting Editable Objects from a Single Photo [video]
#28This is awesome - but how do they reconstruct the backgrounds that the objects previously obscured? There must be more photos?
Re: 3-Sweep: Extracting Editable Objects from a Single Photo [video]
#29Also awesome is that it handles the background replacement so well. This could also be used to just remove an ugly lamp post, telephone pole, etc from an otherwise good photo. (assuming you can remove objects and resave the image) 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.
AFAIK, Photoshop uses essentially the same method -- half the authors on the original Patchmatch paper were from Adobe.
Re: 3-Sweep: Extracting Editable Objects from a Single Photo [video]
#30This is awesome - but how do they reconstruct the backgrounds that the objects previously obscured? There must be more photos?
i thought about that too - i think the background is simply a mirror image of the foreground, and that the object 3d shape is symmetrical