Libertarian Ok… posted in random order (I flipped a coin to decide which to post first) are two picks to decide between.
Please judge them as the appear in the browser…
I did the work at the final resolution… If I really wanted to do it right, I would do the work on a much higher res original, then scale the size down for publication. That should hide any missed artifacts.
This was about 25 minutes work… just FYI.
Okay, Scott. I’ve just seen your post, and am beginning my assessment now. Please stand by.
My opinion is that the original was the yellow car.
You don’t sound very certain.
But, you are correct.
How confident where you, and what were the clues.
Also, if you saw the red image by itself somewhere… Would it occur to you that it was altered?
I don’t know how much more certain I could have sounded. I stated my opinion unequivocally. At any rate, there are a number of tip-offs. I’ll start with those that are endemic to the red car itself.
[li] Almost instantly, you can see a yellowish hue over the very top-rear of the red car. This hue extends upwards, over the rear top of the car, and quite a ways toward the front.[/li]
[li] The space between the driver and the windshield appears to be the glass on the other side of the driver. It has a yellow hue.[/li]
[li] There is a yellow halo surronding the driver-side mirror.[/li]
[li] There is yellow in the headlight highlight and midtone.[/li]
[li] There is yellow reflected in the wheel spokes.[/li]
[li] Finally, there is a yellow reflection in the shadow just in front of the car.[/li]
Having both images certainly made it easier and I’ll admit that it helped to cinch what I already strongly suspected.
[li] There is an enormous loss of highlight detail from the yellow car to the red car.[/li]
[li] The side panel reflector on the red car doesn’t look right when compared to the one on the yellow car. It’s too cyan. Even if that lens were yellow originally, it should have been less red-opposite.[/li]
[li] Finally, the yellow car seems to have a more natural gamma with respect to the rest of the image than the red car.[/li]
I suppose that’s about it. I would say that I was fairly confident within a minute, and very confident after about five minutes of reviews and re-reviews. I posted as soon as I could, but the board was extremely slow. And I always preview, no matter how slow it is.
I think you did a good job. But it’s the nature of the thing that it’s devilishly hard to pull off cleanly.
Well, it is gratifying that it took nearly 5 minutes of study to really be convinced.
Clearly, it was good enough that if you saw it singly, it probably would not stand out as altered. (Which for most purposes is good enough)
As I did this at work, I was in a bit of a hurry. I will make one more attempt this evening to do a really decent job. I still think, this one would fool alot of people.
By the way, did you magnify the image at all to examine it, or did you just look at it as it was?
I thought I made it clear (but apparently not) that the red car image by itself is sufficiently questionable to derive a fairly solid opinion. Having the yellow image merely solidifies an already nearly certain thing.
The items I listed for you are readily visible without any magnification (as you can see). I saw your request that I examine them as they appear in the browser, and that is what I did. The one thing I might not have made clear is that the loss of highlight detail resulted in the red car not “shining”. That started me off on a rather distracting logic angle.
Yes, five minutes might seem like a long time, but this was no casual observation. You had place me under considerable pressure, and I was being tested. As I said, had you required an instant answer, I would have said the same thing. There is misplaced yellow practically everywhere.
In the course of an entire colorized movie, the effects are so annoying that they make me dizzy. Maybe it’s just a Melancholy thing. When I was a kid, I remember being nonplussed when I became aware of the story of the Princess and the Pea. “What?” I thought, “What’s the big deal. Of course someone would notice a pea under a stack of mattresses.”
Oh, sorry ,it was clear… I got that, but was gratified that it took actual “study” to really decide for sure.
I was just making sure on the magnification…
On the loss of shine, that was what was the most telling thing to me… I think I can maintain with a bit more effort.
ooooh, I thought I was being tested. Tested to see if I could make a forgery that would pass a competent observer.
I wouldn’t go so far as to say there was misplaced yellow “everywhere”, though.
I will agree on the colorized movie bit… They’re pretty pitiful usually, and I won’t even touch the princess and the pea comment.
None the less, interested in examining one where I pull out all the stops?
Guys, this is great GQ stuff! Trial and error, examples, cites, the whole bit.
While it hasn’t so far, it has the potential to get personal. I have confidence that no one will succumb to that temptation and keep this thread the sterling[sup]1[/sup] example that it is.
[sup]1[/sup] Or was that golden… 
Don’t know why it would get personal… Libertarian is providing great feedback as far as I am concerned. I am surprised he spotted problems as easily as he did… but just makes me want to try harder.
Point is, he spotted the fake. period… I failed to fool him. I think I can fool him, but I may be wrong.
But, as to the OP, I think we are showing that a fairly passable color substitution is possible fairly easily with at least one product. Getting it dead nuts undetectable is more difficult. Still trying for undetectable.
I have noticed one thing… cars and other shiny items are much tougher to do than non shiny things I have typically had to deal with in the past. They pick up alot more color from the environment, and that is hard to deal with.
I have redone real estate photos taken in winter with the unattractive dead brown grass. Turned that back to lush green… piece of cake compared to the car picks.
Hmmm. Obviously, one of the factors in play here is the resolution. Did you do this for a glossy, off-set printed catalogue, or a website with a 1048XWhatever pic, or what?
Scott,
As you wish. And please, call me Lib. It is the moniker of choice among practically all my friends here.
Manny,
Thank you for your comments and your caveat.
pulykamell’s four cars look pretty good.
To see it and not the Angelfire logo, first go to http://www.angelfire.com/ab/pawinski/
Then paste this into the browser and press ENTER
http://www.angelfire.com/ab/pawinski/fourcars.jpg
Angelfire will think that it was linked from the Angelfire website.
Thank you, Holden. I got to them that way. They are not horrible. They do appear to have suffered considerable compression loss, however. And in general, they suffer from the same problems as Earthling’s efforts: highlights have disappeared, elements that should not have changed hue did so, color is bleeding into where it doesn’t belong, and so on.
This is not to disparage anyone’s work. This is not a case where something is easy to do and nobody has the skill. It is a case where something is challenging beyond reason. Computers are lousy at pattern recognition at any significant level of complexity. A person could do a fairly good job, but it would take quite a bit of time. For one thing, you need to take lots of stabs at it, with “times-out” between each stab so that you can see more and more nuances with each visit.
For me, it is easier to do this kind of work using HSL rather than RGB, but to each his own. To make the blue car yellow, and satisfy myself, I would require a couple of days.
Ok Lib, you win.
I played with some images last nite and was never completely satisfied. Most interesting images have some radiocity effect in them, and that sucks to deal with.
For those that are unfamiliar with radiocity, it is when the color of one object spills over onto another object. For example, if you have a shiny red car in the driveway, your garage door is probably gonna have some reflected red tint on it somewhere, and even worse, the red car is probably gonna pick up some green tint somewhere from close by grass, trees, bushes, etc.
I am sure I could probably fool ya if I deliberately picked an image that displayed little or no radiocity OR I only shifted the objects color very little. This wasn’t satisfactory to me (cheating).
Specifically addressing the OP, yes there are software tools that will allow a person to alter the color of elements in an image AND retain the shading and highlights of the object. Photoshop is what I was using, and will be happy to say to anyone, it is one beautiful piece of software. There are limitations though.
If you have Photoshop or get Photoshop, I will be happy show you step by step to procedures to accomplish this task.
Contrary to posts in this thread, it takes almost no artist talent to do alot of work in photoshop. I have nearly none (my greatest weekness I think). I SUCK at doing creative designs in Photoshop (or any other tool for that matter), but I am very good at image alteration, and creating/replicating specific looks and effects. For example, I can make a brightly lit room look like it is candlelit pretty convincingly… But, I probably wouldn’t come up with the idea that it would look good. Does that make sense?
Let’s say we all win. We had a good time, and we all gained new and valuable knowledge.
I’ve never heard the effect called “radiocity”, but then, as with most things, I’m self-taught. One disadvantage of informal education is that sometimes you miss out on key vocabulary of concepts that you know instinctively or have come to learn. And now I’ve learned a new term that I can use to describe just exactly why working with color is so devilish to the next person who asks. Thanks, Scott.
Aside from the OP itself, I think one thing this topic illustrates well is the incredible pattern recognition capabilities of brains. Whenever computers become capable of this, it will launch a whole new revolution in computing. Just think of the applications possible when these machines become able to select a car (and all its symbiotic compositional effects) from a photograph. When this happens, we’ll be able to say that computers can see.
I suppose this thread will sink away now, so I’d like to leave it with my favorite quote about computers by, of all people, Pablo Picasso: “Computers are useless. They give you only answers.”
>> “Computers are useless. They give you only answers.”
Well, yes, that and porn.
Damn, I misspelled radiocity every time… It should be radiosity.
The other is more romantic. 
