 |
||
|
23. A Human Skull - Image Preparation 1 |
||
 MODELING A HUMAN SKULL Download Unfinished Front/Bottom/Right Image files This tutorial will allow you to create the front back and right views of a human skull which will be used to perform the next tutorial. Before we start, though, I must remind you that you must ALWAYS plan an object before you create it. Just as few paintings are arrived at without pre-planning, sometimes it will mean doing sketches of the proposed object, it might mean using graph paper to scale it correctly, it may mean using Photoshop to do Front/Back/Top/Side views which will match up, but it ALWAYS mean at least a little bit of planning before you actually start creating your model. When you look at some 3D object which impressed you that someone did, you can guarantee it took lots of careful planning before they arrived at the finished product that you see. It may also take research on textures, possibly purchasing books to better understand muscle groups, leaf surfaces or hair patterns. I have given you partially touched-up images with the text and lines removed and a white background. You will need to follow this tutorial to create the skull for yourself. PHOTOSHOP, CREATING BOTTOM/RIGHT/FRONT IMAGES OF SKULL Thus, the first step to creating a human skull is find photos of one, or draw one from all three angles. Since you probably don't have a human skull in your back pocket, you will have to resort to images from books. I research and find excellent quality front/side/bottom photos in an anatomy book called "Eyewitness Books - Skeleton, Written by Stephen ParkerCopyright ©1988 pg 26, 27". I scan the images, and this is what one of the scans looks like before any manipulation.  Next, I will be working in Photoshop. I cut the outlines, adjust the levels, adjust the color balance, give it some Gausian Blur with a radius of 1.2 pixels to get rid of the halftone dots. The front and side image are about twice the size of the bottom image so I digitize the first two at 150 dpi and the bottom image at 300 dpi. PHOTOSHOP, MAKING THE FRONT OF THE SKULL SYMETRICAL I use the lasso tool to make the background white and the clone tool to remove the black lines. As usual with all 3D Sculpted objects, it will be easier if I do only 1/2 of the skull, mirroring the other half. So, I do that in Photoshop. Since I have a photo of the the right side of the head I will keep that side of the head (I can always flip the images if I need them to be the left side of the head). When I use Photoshop's 'Rectangle Marquee Tool' (The 'm' key) to divide the head in two, I find that the left side of the image is wider than the right half.  I delete the right half by pressing the 'Delete' key.  If I cloned the left side of the image now, and flipped it, the face would be too wide. I would like the center line to be in the center of the image. Thus, I look at the width of the entire image and find that it is 662 pixels wide. I highlight the left half of the skull (from my view) and copy it to the clipboard with CTRL-C.  The width of the left half of the image is 362 pixels wide. So, if I reduce its size to half of 662 pixels (331 pixels) pixels, it should mirror perfectly. I go to the 'Image->Image Size' panel and checkmark 'Resample Image'. I turn off the checkmark for 'Constrain Proportions' so it will only stretch horizontally.  Before I double the size, I do a CTRL-A to highlight the entire half of the image and copy it to the clipboard with CTRL-C. It will be there when I need it when I want to flip it. I ensure that the background color is white, I enter 'Image->Canvas Size'. I choose the left side of the image to be anchored by clicking the left-hand square in the 'Anchor' controller. I choose 'Percent' for Width and put 200 as the percent value to double just the width.  Now that it has doubled in width, press CTRL-V to paste the half into the image. It will create a new layer.  I do a CTRL-A again, dragging the pasted layer image while holding down the CTRL key to make it a floating selection. I do an 'Edit->Transform->Flip Horizontal'. I drag the floating selection to the right until the dotted line on the right exactly is on the last row on the right of the image, and the top and bottom dotted lines are still visible.  I use the 'Rectangular Marquee Tool' ('m' key) to select the exact pixels right up to the head pixels on all four sides, I do an 'Image->Crop' to remove all white background that I can. I ensure that the skull is an even number of pixels high and wide. It's 636 pixels wide by 1010 tall.  Do 'File-Save As...', choose 'BMP' as the format. Save the file as front.bmp PHOTOSHOP, MAKING THE BOTTOM OF THE SKULL SYMETRICAL I now do the same thing to the 'Bottom' Image. I find with the bottom view, however, that the left side of the image ISN'T wider than half the image. That means that the front of the skull was probably rotated a little from perfectly perpendicular angle. So, I don't shrink the width at all this time before I double its width. In the end it must also be 636 pixels wide. This time I decide to crop it to the exact pixel width before I double it so it will make it easier to exactly equal 636 pixels when I'm done.  I do an 'Image->Image Size...' on half the image, with both 'Contstrain Proportions' and 'Resample Image' checkmarked so it grows in both the width and height dimensions. If I increase the width of half the image to 318 pixels (half of the total 636 pixels I'll need it to be in the end), I will have the proper width once I double it.  I select exactly the left half right up to where it turns white and hold the CTRL and the ALT keys down, clicking with the left mouse button and dragging it to the right to both make a copy of it and make it into a floating selection. I do an 'Edit->Transform->Flip Horizontal' to flip it horizontally, moving the selection so the crawling ants are even with the top, bottom and right edges. I do a 'Layer->Flatten Image' and it is done.  Do 'File-Save As...', choose 'BMP' as the format. Save the file as bottom.bmp I check the image size to see the height of this image because that is the number of pixels I'll have to use for the depth of the skull when I do the right side of the skull.  PHOTOSHOP, ENSURING SIDE VIEW IS CORRECT SIZE Therefore, when I do the side view, it must end up being 1047 pixels wide. In this case, because we won't be making the skull symetrical in the depth dimension, it doesn't matter that the size is an odd number of pixels. But, the side view will also have to be 1010 pixels tall because of the Front view's image. This will, at least in theory, be an easy one to do. I just should be able to crop the image of the right side of the skull so it is exactly touching the edges of the white pixels all around.  I 'Image->Crop' it. I resize the image to be 1010 pixels high by 1047 pixels wide. I make sure that 'Contrained Proportions' is not checked as I want both the height and width to grow in strange proportions. I make sure 'Resample Image' is checked.  Do 'File-Save As...', choose 'BMP' as the format. Save the file as right.bmp PHOTOSHOP, MAKING SURE THINGS MATCH UP So far I have just been going by the theory that if I make the skull the same number of pixels wide and tall for front, side and bottom images that the top of the nose in the side view will match up with the top of the nose on the front view, and so on... But, photos have warped perspective relative to the straight-on world with no perspective that 3D Sculpture needs and no matter how good the three views are, there will be discrepancies. It is best to get rid of the discrepancies BEFORE you start working on the model rather than after. The more time you spend creating good separated views, the better your object will look when you're done. If the side image were tilted backward or forwards relative to the front image, pertinent points might not match. I'll use Photoshop to check correlations for key points such as the eyebrow ridge, the top and bottom of the nose, the center of the teeth, etc. and possibly move things if they don't. Since the side view is the view which has the largest dimensions, I will use it as the starting photo. I load it in, doing a 'File->Save As...' to save it as 'ThreeViews.psd'. I do an 'Image->Canvas Size...' and select 'Percent' to 200% for both Width and Height. I choose to anchor the skull in the upper left corner.  This will give me room on the right to place the front view, and room below the front view to place the bottom view.  I load the front view into Photoshop, select it all, copy it to the clipboard, switch to 'ThreeViews.psd', paste with CTRL-V. I position the front view to the right of the side view so the top pixel of the front view matches the top of the image. I find that more canvas space is needed to place the bottom view, so I guess, giving it more than I need, pasting the bottom view and position it so exact pixels match up between the side of the front view and the side of the bottom view. Once they match up, I draw a selection rectangle around all three skulls, exactly picking right to the edge of all active pixels.  Looking at it visually, it looks like all points match up very well except the projection at the back of the jaw. I do a 'Layer->New->Layer...' and call it 'Red Lines'.  I right-click on the 'Rectangle Marquee Tool' in Photoshop's Toolbar and select the 'Single Row Marquee Tool'.  I place a single row, lining it up with the top of the forehead brow, select Red as the foreground color, doing an 'Edit Stroke', choosing '4 pt' as the size and 'Center'.  I continue doing this for each key point. I note where discrepancies are. There's a very good correlation between each of the points. I have placed a red arrow to show corresponding points along each key line. The only two points which start to converge are the bottom of the front teeth and the back of the jaw. Since the back of the jaw makes complete sense, as this is perspective that is causing this discrepancy, I will call it good enough for the correlations between front and side. Since the bottom probably won't be seen as much I will just say they match in that case. I might not even use the bottom view that much anyway. Generally the top and bottom of an object are far less important than the front and the side view. Next, I will show how to create a skull from these images. Click for a Rotating 3D Version of the skull done using the Blaxxun3D Applet |
||
