Slashdot Mirror
Photoshop CS5's Showpiece — Content-Aware Fill
Barence writes "If you're looking for reasons to upgrade to Photoshop CS5 when it arrives, a new demo video might just persuade you. Narrated by Bryan O'Neil-Hughes, a product manager on the Photoshop team, the video shows the new content-aware fill tool, which has the potential to revolutionise the way you clean up photos. If you're not happy with an item in your picture, select it, delete it, and Photoshop will analyse the surrounding area and plug the gap as if it never existed."
Stalin would have just loved that content-aware fill tool.....
I wouldn't call this an ad. This is legitimately really fucking cool.
"16MB (fuck off, MiB fascists)" - The Mighty Buzzard
Fox News and the Texas board of Education.
No, no. Goodbye to all my exes in my vacation photos. Stupid real dolls blocking the scenery.
Photoshop currently sells at a "lightweight" $700. How many photos would I have to edit to make that cost effective? It entered the land of exclusive pro tool years ago.
Photoshop has had that capability natively (ie, not requiring a plugin) since CS4, this is the ability to select an object in an image - litter on a lawn was the example given in the article - and replace its former location in the image with content derived from the surrounding areas. Basically it's like an intelligent, automated version of the Clone Brush tool on steroids.
UNIX? They're not even circumcised! Savages!
Sounds like something Google Street View could use to remove people from their views and make them more acceptable.
I wish there was a paper on the core algorithms behind it (cmon Adobe, SIGGRAPH), but I could see why Adobe may be sitting on every aspect of this tool; because it sure seems to bring some real photoshop wizardry to the common user. It was really an example of "delete this thing" and it just works. Takes a common complex task and massively simplifies it. One of the most impressive marketing demos I've seen in a while.
Sure, there are some cases in which I doubt it works, but from what I could see, it seemed to have some vision and perceptual rules built in to guide how to fill in the deleted area. And frankly, it's a feature that for professionals, makes the price tag for the upgrade worth it. For some tasks, it'd pay for itself in labor alone. What would take a expert hours to do, this could do in minutes. If I was Adobe, I'd seriously consider taking this and make a Photoshop Elements Extended Edition (or whatever) and add about 79-99 bucks to this price for this feature alone. Arguably, it'd be worth it for many.
Next up is the "CSI Enhance" tool. Take a photo of 10x10 pixels, and make it a perfect 2MP image.
I wonder what sort of a fig leaf it will use to plug the gap in the goatse photo...
I want to play Free Market with a drowning Libertarian.
Color, I mean content-aware fill me impressed.
Wasn't this core technology discussed on Slashdot a number of years back? If you google "Seam Carving" you'll find some nice wikipedia articles that discuss content-aware image resizing. This may be a variant on the same technology, and i actually doubt that this is an early release of an April Fool's Day joke (no matter how Star Trek this technology seems).
Liquid rescale is an implementation of the Seam Carving technology which was incorporated into Photoshop CS4 as a feature titled Content Aware Scale.
This new feature comes from an algorithm titled PatchMatch which was presented at SIGGRAPH 2009:
http://www.cs.princeton.edu/gfx/pubs/Barnes_2009_PAR/index.php
Welcome my "getting modded Insightful for not RTFA and spreading bullshit" overlords at Slashdot.
Content-aware scaling has been included in Photoshop since CS4.
But this is no scaling, it's filling. The first 3 minutes of the video are not so interesting (nothing one could not do with a clone stamp in 2 seconds), but the last 2 minutes are breath-taking.
GREYCstoration or liquidrescale don't even come close.
I wouldn't call this an ad. This is legitimately really fucking cool.
So is the Mr. Clean Magic Eraser...
Using Photoshop to remove lens flares? Oh! Brave new world!
-Peter
I see a 12 megapixel image in hand of a before and after and not a tiny less than 400 pixel overcompressed youtube video.
I have seen this automatic stuff before and when you look carefully at it it's not very clean unless you re-sample down to 1/4 the resolution or go small for web use.. it's never clean enough to print out at 11X17 or larger.
Do not look at laser with remaining good eye.
I once did a writing contract at Adobe. You know how when you pose for an ID photo, they put you in front of a curtain or something to hide the background? When I got my Adobe badge the security guy just posed me against a regular wall, then Photoshopped the wall out of the picture!
It's not the same. Liquid Rescale moves the pieces by rescaling around them. This actually replaces just the exact area. I think it's quite a bit more useful, but in different ways.
"If you make people think they're thinking, they'll love you; But if you really make them think, they'll hate you." - DM
noone wants to shell out $60 for 200 functions 20 of which they will use from time to time.
Personally, I'll shell out. I make a living using photoshop and I support the idea that a bunch of extremely talented software engineers ought to be able to make a living developing it.
Nope not put you out of a job, just lower your wages to $8.95 an hour because monkeys can now do your job.
Welcome to what us in Photography have had to deal with. I just saw an AD on the TV for a mall photo studio that will give you 20 shots in their studio for $9.95. and 8X10 prints start at $4.95 each. Yes I know it's done by no talent kids or minimum wage people, but the average consumer does not know that. They still think that it's all in the cost of the equipment and has nothing to do with skill and experience.
Do not look at laser with remaining good eye.
I can't wait until a crime show gets ahold of this.
"Delete that wall and see what is behind it. Enhance. Enhance. Enhance."
or else!
Yeah, and cars have already had the native ability to drive, turn and stop for a century. The DARPA Grand Challenge isn't really adding anything new.
Those robotic cars are basically just intelligent automated versions of cars, on steroids.
Just because it happens in software does not make it trivial.
Porquoi?
Includes more detail about the algo
- Developed with researches at Princeton
- Demo'd at SIGGRAPH in Aug. 2009
- Old spot-healing tool tried to find one match for the hole; new tool copies multiple patches from the surrounding BG to fit into the hole, as well as finding & copying surrounding patterns
http://www.youtube.com/watch?v=S9vbHRcrbdQ&feature=related
The English word fart is one of the oldest words in the English vocabulary.
So, what about Resynthesizer?
Ezekiel 23:20
That capability has been available for a while in the Gimp as part of the Resynth plugin:
http://www.logarithmic.net/pfh/resynthesizer
It lets you resynthesize a texture, fill in a selection with surrounding content, and synthesize images "in the theme" of another image.
A few days ago I was reading about some of the algorithms for doing this, shown at Siggraph in recent years. I think it's real.
GCHQ Quantum Insert installed. If only our tongues were made of glass, how much more careful we would be when we speak
Yes, clearly the images where Photoshopped. I could tell from some of the pixels and having seen quite a few Photoshops in my time... er, wait a minute...
I support the Slashcott and will not be reading or commenting from 2/10/14 to 2/17/14. Beta is steaming pile of dog shit
I'm usually among the first in line to call Slashdot on its thinly-disguised slashvertisements, but this goes beyond product upgrade and into the realm of William Gibson novel.
I'm kind of staggered just trying to wrap my head around the uses and implications...
The key thing is getting it not to chatter or flicker, which it probably will- as I doubt it will generate the exact same results frame to frame. Nevertheless- expect it to make matte painting, wire removal, etc a lot easier. I expect they'll use it to generate a quick starting point for clean plates, which will then be given further refinements and then composited in normally.
After watching the video and seeing obvious problems even at 360p, it seems unlikely it'd hold up at 2k without some love at least.
www.GrenadeHop.com
Just "wow". Everyone who has spent tedious hours "fixing" some piece of "almost" perfect photography just fell off of their chairs.
I haven't bothered upgrading anything but InDesign in recent years - the old Photoshop (or even GIMP) was good enough. This is a reason to upgrade!
Enjoy life! This is not a dress rehearsal.
http://www.youtube.com/watch?v=dCr96ldClz4&feature=related
Same stuff except with real time video. Amazing.
It's a very cool GIMP plugin for some things, but...
This is my source image:
http://s3.images.com/huge.28.142421.JPG
I want to remove the lady on the right, so I select her:
http://img714.imageshack.us/img714/1346/resynthesizerselection.jpg
And then, per the Resynthesizer page's recommendations, I use "Script-Fu/Enhance/Smart remove selection..."
http://img121.imageshack.us/img121/228/resynthesizerresultradi.jpg
Oh dear.
Anybody with access to the Photoshop beta feature want to give that image a stab? For all I know it fails just as spectacularly - but from the research it's based on, I highly suspect it'll fare better.
Considering that the regular image clone tool did that in one click (well, not counting tracing a selection and setting the clone source), I don’t think it’s a very good test image.
(I used GIMP, not PS, if it makes a difference.)
Alexander Peter Kristopeit bought his basement from his mommy for one dollar.
darn blockquote fail :)
I know I picked it out - what about it?
I didn't pick it out to specifically make Resynthesizer fail - it's image #2 on images.google.com for 'person in field' (sans quotes).
For an example that does work with Resynthesizer, try:
http://media-cdn.tripadvisor.com/media/photo-s/01/0f/33/e2/so-cool.jpg
Select the top-left dark thing, run the Resynthesizer script-fu - voila... dark thing removed, and sky filled in pretty well.
The problem is that this is entirely hit-or-miss.. and it's far more often miss than hit.. and then -when- it is a miss, it's a spectacular miss (as in that person-in-field image).
This is what I got very quickly with Alien Skin's Image Doctor: http://img530.imageshack.us/img530/1785/edited28142421.jpg
My binary is an ELF! It was different before, but that's a.out.
Switch back to Slashdot's D1 system.
This is what I get using the plugin on its own: http://shishnet.org/ufufuf/resynth2.jpg
Do note that the script-fu wrapper works better for larger images, which this isn't
Also, the example from the video, done with gimp instead, the results are pretty similar (IMO, better, but I'm pretty sure that the "improvements" are just luck): http://shishnet.org/ufufuf/panorama-synth.png
Having been using the resynthesizer for years, I've developed a knack for which source images will work well and which won't, and the thing that struck me about that video was that the source images are pretty much ideal conditions -- I'll be impressed when they can get good results on the images that aren't so clean :-)
I mod down anyone who says "I will be modded down for this", regardless of the rest of their comment
Just tried expanding a panorama as in the demo too, and resynthesizer does that pretty well too (please ignore the fact that the source material is REALLY shittily exposed, it was shot on a mobile phone with no manual exposure mode :( ):
original
expanded
I mod down anyone who says "I will be modded down for this", regardless of the rest of their comment
obvious question: workflow / parameters?
Generally I use the foreground select tool* to select the smallest area to cover the object, then grow selection by a few pixels so that none of the object's edges are poking out and confusing it, then filters - map - resynthesize (ie, use the plugin rather than the script), and have the "tilable" options disabled since they tend to grab samples from the opposite edge of the image (if I want a tilable image, I'll use the tiling filter separately...)
Probably the biggest factor for simple success is to have the object you want to remove be on its own (surrounded on all sides by similar textures) -- if it isn't, then you need to do things the long way -- eg, if you want to remove the leftmost wheelchair from this image, and you want it to be replaced by grass when three of its borders are touching non-grass, then you'll find that it ends up somewhat messy since it attempts to merge four different edge textures. In this case you'll need to copy a section of your desired fill texture (ie, a rectangle of pure grass) into a separate image (specifically, a single layer image with no transparency); then on the original image select the object to remove, open resynthesizer, and select the "fill texture" image as the texture source; this way the generated texture will both match the surroundings of the original as much as possible, while being filled with the "surroundings" that you've specifically chosen. Having taken a sample of "pure grass" and a sample of "pure stone", then removing the top and bottom halves of the wheelchair with each respectively, the results are nicer. (with the exception that the first two images were produced with a mouse and twenty minutes of careful selecting, and the final one was 5 minutes work with a laptop nipple, so there are still some bits of wheelchair poking out of the sides...)
Incidentally, does photoshop have SIOX yet? Having the features "vaguely scribble in the general area of an object to have the object selected precisely" and "automatically and realistically remove a selection" could potentially combine to form "one-click realistic object removal" \o/
I mod down anyone who says "I will be modded down for this", regardless of the rest of their comment