What is Digg?120 Comments
- hamstu, on 10/10/2007, -0/+81Good for him. It's nice to see people rewarded for their hard work.
- shadowspawn, on 10/10/2007, -1/+63Photoshop get what photoshop want!
- reckless8594, on 10/10/2007, -3/+38I can't wait for this, it's pretty crazy.
- bjs3171, on 10/10/2007, -1/+35holy ***** crappers. and i can come up with tons of times i've wished i could do this, as a graphic designer.
- inactive, on 10/10/2007, -1/+28It's almost amazing that nobody thought of this before. And props to Adobe for hiring the guy rather than taking the cheaper route and stealing the algorithm he explains in the video.
- Tenetri, on 10/10/2007, -0/+24If I do it quickly, I can remove someone from an image like a beach in a minute or two. If its from a street with complex backgrounds, can take 5-10 minutes depending... this program did it in half a second... I'm certainly anticipating the next release of Photoshop
- evilregis, on 10/10/2007, -2/+17It would be nice if browsers themselves could do this on-the-fly. Making fluid designs with masthead graphics that adapt to the users' resolution would be incredible.
- akaii, on 10/10/2007, -1/+13Perhaps Adobe tried and failed? So they took the easiest route..
- glasnostic, on 10/10/2007, -0/+11This will greatly reduce my disdain for my idiot clients.
- SparQy, on 10/10/2007, -0/+11Funny how the brain works. Or doesn't.
I spent a good 20 seconds reading the title as "Adobe Hi-Res ..." trying to figure out what the hell the author was trying to say. - shaunmadams, on 10/10/2007, -1/+11I'd buy it just to never have to search for a panoramic photo for a web layout ever again.
- philz, on 10/10/2007, -1/+10Now I can tell everyone that first there was my girlfriend on the picture, but I removed her.
- lucasgarsha, on 10/10/2007, -1/+9This would be a great addition to widescreen TVs... SDTV stretched as it is now is awful, so are black bars on the side.
- akumal999, on 10/10/2007, -2/+9I could resize all my old wallpapers 1024x768 to 1280x720 in a snap, glad i didn't deleted'em.
- RecoDesign, on 10/10/2007, -0/+6newbies dont buy photoshop...they steal it
- Fordi, on 10/10/2007, -0/+6Seriously? That's your question?
Ask any graphic designer; just that 'stuff removal' procedure is pure gold. Especially if they can figure a way to localize it (create selection, demote object, remove item). - NAvAP, on 10/10/2007, -0/+6an offer he couldn't refuse
- ahawks, on 10/10/2007, -0/+6Don't you guys realize how business works? You think they'd just release this as a free plugin, when this could easily be marketed as a key feature of CS4?
- burymobile, on 10/10/2007, -2/+7wow, who says magic isnt real?
- Sandurz, on 10/10/2007, -0/+5Are you suggesting that TV's should do this to each individual frame of video on the fly to fill up some vertical bars on the side?
- brundlefly76, on 10/10/2007, -0/+5I dont understand why he went to work for Adobe - why didn't he just patent and license this to them??!!
I hope he didn't shortchange himself - its an outstanding invention. - Razster, on 10/10/2007, -0/+4You and me both, this will help out not only my Design aspect of my job but the photographer part of my personal business.
When I first took a look at this video I nearly wet myself and asked how come no one in all these years has thought or done this sooner... Glad it is now and not 10years from now.
I'm very excited! - Error601, on 10/10/2007, -0/+3Interesting algorithm. I'd like to see it applied to expanding 4:3 images to 16:9. I think it would look much better than the stretch algorithm of changing the aspect ratio as you move away from the center of the picture.
- narutometal, on 10/10/2007, -0/+3Good luck with that.
- kevisazombie, on 10/10/2007, -0/+3maybe they made him a realy nice offer
- jcims, on 10/10/2007, -0/+3It looks like it would be useful for fixing/tweaking composition after the shot is taken. Also, if you consider how it works, it should be possible to select a specific area within the image and move it around a bit, using this technique to compress/expand the void left.
- simpleid, on 10/10/2007, -0/+3I'm sure people have 'thought' of it, I've had and do have a lot of good ideas, as most programmers do, but the math... the math is the hard part! (Not really the math though, more like the system...)
Sometimes the best ideas no matter how simple, have some obscure complex algorithm behind it which is so difficult to narrow down to. A person has to understand how all the variables change in relationship to one another.
The only practical way of doing anything is via an algorithm, and to do that you need to understand the underlying concepts fully, enough to know what variables map to what aspect of your problem. Then how to write formulas that describe the way those variables 'behave,' sometimes in multiple contexts.
Besides all that you need the code to runs efficiently enough to pull it off quickly...
If this guy really did it alone that's very impressive since he would know how to write the algorithms to be as optimized as humanely/machinely possible. :-)
Thinking of it... that's the easy part of inventing. It's not coming up with an idea that's hard. That's easy.... It's the above which goes in to designing the framework for a functioning solution that becomes the challenge. - Shadow6363, on 10/10/2007, -0/+3Hrmm, I only now realized it says hires after reading your comment, I too read it as hi-res...
- McBradd, on 10/10/2007, -0/+3I'm in the Games industry and I use photoshop constantly to generate textures from photo reference. This tech would make cleaning up ref photo's a breeze. For example, you have a nice wall texture but you need to take out a door, or you have a nice sand texture but need to take out a rock...
- bouche, on 10/10/2007, -0/+3incredible. i hope he retains the license. this is going to make him RICH!
- mrgulabull, on 10/10/2007, -1/+4I find it amazing someone HAS thought of this. That algorithm sounds and looks incredibly complex. There's very little 'common sense' to the way in which it evaluates the images.
- lazyfisherman, on 10/10/2007, -5/+8so true
wish photoshop want run native on linux - iOsiris, on 10/10/2007, -0/+2although seriously awesome tech, you guys should check out the potential problems outlined on the slashdot comments
- slamtv7, on 10/10/2007, -3/+5Wow that was amazing!! I can't wait for this!
- glasnostic, on 10/10/2007, -0/+2Personally, I am constantly asked to put some image that the client has selected into a spot that is the complete opposite format. This application would save me many many hours.
- simpleid, on 10/10/2007, -0/+2Oh snap... I'm very freaking impressed, this kicks some serious ass. Great work!
- regeya, on 10/10/2007, -1/+3Considering how often I'm given a customer-provided vertical print (like, 3" wide by 5" tall) and am told to not crop anything out but make it fit in 2 col. x 2 in., this would be not only a Godsend but an argument-ender.
I'm just amazed that it took this long for someone to invent it... - daphreak, on 10/10/2007, -0/+2go GIMP coders go! get this under a GNU license before adobe can put out a product of their own.
- geminitojanus, on 10/10/2007, -1/+3"Steal it"? The guy invented it. He has the option of patenting it. Adobe hired him so that if he /does/ patent it, Adobe will control the patent.
That's how these companies think today, and it's a royal PITA. - pleiadianagenda, on 10/10/2007, -0/+2Visual media will be forever changed. Wait till these algorithms are applied to digital video and video games. Depth perception will become so much more realistic.
- inactive, on 10/10/2007, -0/+2I viewed this in person at Siggraph, and it's better than this video shows. the level of control you have over the image resizing is incredible. the quality of the resized images is nothing short of amazing. plus, it's lossless, in that the information is still part of the image, not discarded, and can be restored easily. It provides a level of intelligence to the image that is built around design and presentation rules. Can't wait to see production, plug-ins etc. developed around the technology.
- pcsperson, on 10/10/2007, -0/+1that was impressive
- Fordi, on 10/10/2007, -0/+1Actually, it's not that hard to do this; the algorithm to calculate seams looks pretty simple to me; and would only process a picture in linear time oN-nN times in each direction (where oN is the original dimension, and nN is the target dimension). Video compression and decompression codecs do a LOT more work than this, and can be implemented in hardware to handle *astounding* resolutions in real time.
- Fordi, on 10/10/2007, -0/+1You can derive the algorithm from the description in the video. Difference of Gaussians is reasonably simple, as is lowest-point pathfinding. The innovation here is in the assembly of parts, not in the parts themselves.
- Guspaz, on 10/10/2007, -0/+1Why not? It's probably not practical due to distortion (the algorithm isn't perfect), but it's probably technologically feasible. I don't way it can't be implemented in hardware real-time, or even in software.
In fact, this may validate Sony's desire to put the Cell processor into televisions. With that kind of processing power, it could calculate seams for 8 or 9 frames at a time. Delaying video isn't usually a problem in non-interactive uses such as TV or movies (as long as audio is delayed too). This would give each cell SPE/PPE about 300ms per frame to do the seam calculations. That seems mroe than sufficient.
However, a more likely proof-of-concept would be implementation as a PC video filter. This should be relatively easy to implement as a plugin. For example, implemented as a dscalar filter, you could load it with ffdshow and use it with any media player. Besides, the performance of PCs is raising at an exponential rate. Even if today's dual-core 2.6GHz processors can't do it in real-time, perhaps tomorrow's 8-core 3.2GHz processors will be able to. - bieber, on 10/10/2007, -0/+1You poor, naive soul. Welcome to the world of software patents, and the exploitation thereof.
- Zero82z, on 10/10/2007, -0/+1I would think they hired him more for his creativity and intelligence than this specific method. If he's capable of coming up with something as ingenious as this, they're going to want whatever else he can think up too.
- Fordi, on 10/10/2007, -0/+1Fortunately, sufficient prior art exists ('rivering') that any patent can be reasonably challenged given a similar but differing implementation.
- Fordi, on 10/10/2007, -0/+1Ideally, such an algorithm applied to motion would include the prior and next frame in its difference of gaussians check.
- crimson117, on 10/10/2007, -0/+1I hope he licenses it free to mozilla or for general open source use.
-
Show 51 - 100 of 117 discussions
Check out the new & improved 
© Digg Inc. 2009
— Content created and posted by Digg users is