Monte Cristo wrote:
On Sat, 5 Jan 2008 15:27:55 -0000, "\(not quite so\) Fat Sam" wrote:
James wrote:
Greetings all,
For quite some time I’ve been working in Photoshop to perfect the art of "swapping heads" and other forms of "Photoshop surgery" (some may call it "Photoshop trickery"). After much time, trial and error, I find that I can do it if a person has the patience. But still more times than not, there seems to be a lingering "fake" aspect to the touched up images. I do my best to match lighting, color, grain, texture and resolution, but yet they never seem to be quite perfect. I know it can be done because I’ve seen other examples floating around here and there, usually on the web, where you know a Photoshop master got in there and did a completely believable composite. Even sometimes of something that can’t possibly exist but yet it looks totally believable, and you findyorself asking, "is that real?"
(perhaps some of you have seen the picture of the "jet airliner," where what is normally the main long passenger section is one big engine, and under the wings where the engines would normally be, are small pods or "passenger sections.")
Seems like I’ve been trying forever (a very long time) to find good information, books or tutorials (preferably video tutorials) on this somewhat obscure subject regarding Photoshop. Some books like "How To Cheat In Photoshop" by Steve Caplin, touch on the subject briefly but are otherwise very lacking and usually devote only 1 or 2 pages to the matter. Who are the masters? Is there anyone that puts out training or tutorials on this specific subject? I would be very grateful if anyone could let me know. Thank you.
James
The best advice I can give you is practice practice practice. And also remember that you’re always going to be your own worst critic.
You know what you did to create the fake, so your eye is naturally drawn to the flaws. Other observers won’t necessarilly spot them. You can bet that the guy who did the example you’re quoting with the airliner probably wasn’t entirely happy with the job, but realised that to his own eyes, it was as good as it’s going to get. He also probably realised that to other folks, those imperfections weren’t as obvious, and might not get noticed at all.
It’s interesting you say that as I’ve thought about this quite a few times. I’m sure every artist goes through it. What is the litmus test as to whether something looks "real enough" or not? It’s interesting… I snap a picture with my Nikon D2X, print it out, show it to people and it looks "real," because it is real. But in all my years of working with Lightwave (and more recently, 3DS) I have seen only ONE rendering of a human face that looked absolutely real. And I guess it took many hours of rendering time for the creator to render that one frame, as I understand it. I’m sure you guys have had the experience where you had to save the project, shut off the computer overnight, and wake up the next morning so you could take a "fresh look" at your image. We sometimes seem to get too involved in our projects, thereby temporarily diminishing our capacity to see it objectively. So we have to step back for a while and then look at it again.
Yeah, I’ve been there too.
In a previous episode of my life, I worked as an illustrator, doing traditional artwork with paints, pencils and inks. I’ve encountered this so many times. Yuo go to bed of an evening thinking "I’ll finish that tomorrow" only to wake up the next day and realise that it doesn’t need any more work. But the opposite can be equally true. I’ve found myself going to bed thinking a job was finished, but when I viewed it again in the morning with fresh eyes, I could see tons of errors and realised I didn’t like it much.
I don’t make any claim to be an expert on photoshop, but I regard myself as a competent user.
These are a couple of examples of my work. Because I did them, my eye automatically gets drawn to the flaws which other people probably wouldn’t notice. I know they’re not perfect, but I also know that I’m much more critical of them than other people.
http://flickr.com/photos/swampy_bogtrotter/302675533
http://flickr.com/photos/swampy_bogtrotter/318974850
http://flickr.com/photos/swampy_bogtrotter/303645420
http://flickr.com/photos/swampy_bogtrotter/306190732
Hey Sam, first I want to thank all of you for your advice and input as it is much appreciated. In my estimation your projects are quite superb. I can’t look a them and really see any "seems" where the alterations took place. And with all respect to your work I want to point out that your images are clearly meant to be perceived as a "photo trick," if you take my meaning.
Hi, yeah, I regard these as photo trickery. Trying to create a surreal scene which looks like it could have been real.
This is sort of what I do for fun in photoshop.
When I’m working seriously in Photoshop, I do restorations and repairs of old or damaged photos.
You can see some of my restoration work here…
http://flickr.com/photos/swampy_bogtrotter/sets/721576000458 24330/ I think I skewed off a bit on a tangent in my initial post talking about the airliner. The main focus of what I want to do is composite heads and/or whole bodies if possible, so my initial post may have been a bit misleading (my own fault).
I know what you mean, but the principles are the same regardless. There are some things that you can do to give yourself a head start and make the job a bit easier.
Before you start to mask, cut out and superimpose the head or body, work with the hue and saturation, levels, curves etc. Whatever it takes to match the colours before you even start merging them. This really does make the whole job easier.
Also, carefull selection of your source images is a must. Attention to details like the direction of shadows, quality of light, tanned skin tones, noise in the images etc can make or break your project.
A lot of folks make the mistake when superimposing heads of not aligning the new head correctly, or having it either too big or too small. This can be easily sorted by reducing teh opacity of the superimposed head, and then playing with the alignment, position, angle and scale until the facial features line up with the image underneath. In particular, try to make the eyes line up, and the top lip or top row of teeth if you can see them. This should get the head positioned in a believable and realistic looking position.
Yuo can also correct any problems with shadows and highlights by using the burn and dodge tools. With carefull work, you can really make the face more or less 3-dimensional feeling, and even alter teh direction of light to match the underlying scene.
Don’t be afraid to experiment with different mask limits on the face. Sometimes you’ll have a defined chin which allows you to mask to the edge of the face.
Sometimes the chin won’t be so defined, so you’ll have to extend the overlay down onto the subjects neck or even chest.
And sometimes you can get away with feather masking the facial features only, allowing the original face shape to form the outline, aith only the features contained within being replaced.
It also helps if all your source images are as high-res as possible.
It’s hard to give an exact blueprint because every job is different, but the three key elements really are colour, lighting and alignment. Get those right and you can find work-arounds for the rest.
I’m sure I will find what I’m looking for at worth1000.com. There’s some pretty awesome stuff there.
There’s some pretty good tutorials on there.
Sadly however, in recent years, their contest entries have become more concerned with the idea or interpretation of the theme rather than the realism of the final image. As a result, some stunningly realistic, and technically perfect images have been getting beaten by images that look like they were done using MS Paint.
At this point I think the problem with head swapping isn’t so much lighting, color and grain (which is definitely an art unto itself) as much as photo perspective. When dealing with Photoshop "cosmetic surgery" and head swapping, a person knows that each picture was shot with a different camera, using different lenses of different focal lengths. All lens systems distort a captured image regardless of equipment quality. Not so much because of inferior equipment (as in the blatant pin cushioning of a $40 lens) but because you’re taking a 3D world and eventually portraying it on a 2D medium of some sort. There’s also the angle and azimuth at which the photos were taken.
I think this is what mostly causes the image to look "fake" when swapping/compositing heads. I think it’s mostly a camera perspective challenge. Would this seem to be the case to you?
I agree about the lens distortion to an extent, but to be honest it’s not as much of an issue as you would imagine.
Photoshop has a whole raft of tools to overcome lens distortion, and when you’re only working with small elements from images, the effects of lens distortion become less relevant and much less obvious.
Good luck with your projects. I hope you’ll share links to your results when you’re ready to share them with us.
