A couple of weeks ago I was asked to do more write ups. Usually I am still at my wall with minutes to go so by the time I post an entry there is precious little time to be writing about it, so I saved regularly to at least have a progression (attached).
The first thing I did when I saw the source was to turn it upside down. That is when I saw the head of my little buddy. A nose, a chin and two big cheeks. The rest of the source was a manipulator's dream, so many bits and pieces :)
The next thing I did was cut out the parts I was going to use. Usually, I will enlage the image by 200%, zoom in quite close and use a point-to-point tool to select the parts. Yes. it is a long way around, but I can be more precise. Once the part is cut out and reduced back
(resize to 50%) to normal size, the edges mostly smooth them selves out on their own (bonus) and the quality of the image relatively unchanged.
Finding a face right away is a big advantage to building some kind of figurine. It creates a focus.
The head is composed of about 25 layers. I keep the layers separate for as long as possible because it makes the later on shading a lot easier.
The shading is done with three layers. The bottom layer (1) being the original, the next layer (2) a darker but slightly see through copy (multiply blend mode) and layer (3), copy of (2) with brightness turned to -255 essentially making it black. Firstly I turn off layer 3 and erase parts of layer 2 where the light would hit adjusting the opacity as required. Next I turn on layer 3 (the black one), reduce the opacity (very low) and also erase the parts where the light would hit and anywhere else the image needs to be lighter. Giving the image a slightly matty and a more 3D look. At this point I merge the three layers but still keep the various components of the face separate just in case.
"Some things once merged can not ever be unmerged :)"
Around here is when I finally found out why I keep crashing PSP8. Even though I have 32GB RAM sitting in my machine, PSP8 seems to be limited to 2. Time to upgrade me thinks.
The next step was to cut out the rest of the components from the source. Take two half oranges and make them a whole. A round ball can be essentially manipulated into any shape, ears, fingers, hands, arms, boots.
To best retain the quality of the image I first use the system tools like deform, straighten and change perspective. These tools seem to retain the pixel relationship like a vector. Essentially if you resize re shape and then bring back to the original shape there is no punishment :) the image retains the original composition. For the finer detail I use the warp brush, but a lot depends on the various settings as to how satisfying its function is. Either way there is a lot of distortion and often I end up having to come back to re-select and cut the unhappy edges back. Softening may work, but that depends on how well the image blends into the background. The starker the contrast the more the edges need to be re-trimmed rather than blended.
Sometimes it is much easier to use multiple, overlaying images and erasing where they join rather than trying to force all the manipulation on the single image.
The steel bowl was the easiest to manipulate and had I started with it, the little dude may have turned out to be man of steel of sorts.
Originally I thought to have 3 to 5 of these little dudes, but building one took over 4 hours so unless I was going to make just copies, which wouldn't look very interesting, this dude took up the center stage.
I've hidden another head in the jungle and added some of the flowers onto the background to give the two images some relationship. Could have done more but once again my expectations of my self and time available did not match very well :)