Those are really cool! I'll admit I wasn't quite sure if what I was looking at were miniatures made to look real, or reality made to look miniature. Some of them blur the line between the two so much it's really difficult to tell! I love stuff like this.
@ WAFS- The purist in me agrees with you. However, the part of me that loves to screw with reality, but also to do it without spending months on a single project has a different take. Xtian and I had a conversation a year or two ago regarding the gradient mesh method of creating photo-realistic art that was similar.
That the following is an illustration created in Adobe Illustrator, is amazing:
(not my work, I only wish lol)
That it was in many ways, copied pixel for point from a photograph is not as amazing. That said, it's not something I have the talent to do. If this image would have been created without the use of any source material, it would be leaps and bounds ahead as far as the talent goes. However, doing a seamless gradient mesh of a photo does in fact require a very good eye for detail, understanding color theory, proportion, layering and a slew of other things. I've seen other work that this artist did without source photos, and he is truly amazing.
Gradient mesh realism can crudely be described as a method for mapping the details and color values of a photo, and then applying them to shapes which were traced from said photo.
For comparison, here is an illustration I did in Adobe Illustrator that was not in the gradient mesh mapping method. I used one photo for the subject, along dozens of other source photos (many of which were not of the bike that is depicted):
http://www.danleventhal.com/brough.html
The main difference is that although I used photos as the basis for the illustration, it was created with endless hours of "digital brush strokes" rather than simply mapping the photo itself. (hey, I love motorbikes, but there is just no way I could store that much information about a bike that I've seen once in my life). Under all of the layers of shapes, blurs and meshes, there isn't a photo, only blank canvas. The downside is, my end result doesn't look photo-realistic. The upside is, I was able to take liberties and change anything that I wanted without losing the the integrity of the image as a whole. (the horn in the source photos was not correct for the machine, nor did it have the proper nickel-chrome finish on the fuel tank, the small fin on the headlight ring, etc.). Thankfully the other bikes in this series no longer exist, so I didn't have to follow any real world images other than for reference (Orwell's Rudge is a wildebeest of years/makes/models).
It took me years to finish that illustration, where it would take someone who does the gradient-mesh method, a photo to work from and maybe a week or two at most. For me it comes down to if I'm getting paid to do it, or for myself. If I'm getting paid, I want to bang it out as fast and accurately as possible, so I can get back to my labors of love. Also, as pointless as it seems to make a human being with a computer a glorified photocopier, the gradient-mesh method has amazing uses for photo restoration and enhancement, as well as seamless collage work. When a client wants a "photo" of the new product they envisage, especially "in it's environment", this method is a godsend.
For me, the key to all this newfangled tech isn't to cheat to get the work done easier, but to have more time and flexibility when it comes to the process. More imagination, less grunt work. It's cheating, yes, but it also makes painting with a real paintbrush a religious experience, something sacred that I won't have to use to make advertisements for toothpaste. I think the same logic follows with photography, music, and it likely will with sculpture once we've all got access to 3-d printers.
