Blender 3D: Noob to Pro/Creating Models With Photo Assistance
From Wikibooks, the open-content textbooks collection
The first tutorial is about using guide images to place vertices in their proper places in 3D space. The second tutorial is on how to take good reference pictures. These tutorials assume that you have completed all previous tutorials.
Contents |
[edit] Making A Pyramid
Note: This section may be incredibly frustrating to new Blender users. If you are starting out it is recommended you go over the earlier modeling tutorials first before going through this section.
First we are going to create a pyramid the easy way. Then we are going to show how to use different viewpoints and images as a guide to place vertices correctly in 3D space.
You should have the default cube, if not, press CTRL+XKEY to start a new project.
- Go into Edit Mode (if in Object Mode, press TAB).
- Select side (right) view (NUM3).
- Make sure no vertices are selected by pressing AKEY.
- Ensure the "Limit selection to visible" ("Occlude background geometry" in Blender 2.46 and later) icon is disabled so we can select all four of the top vertices, even the back two which aren't visible.
- Go to Box Select mode (BKEY) draw a box around the top vertices of the box
- Merge vertices by pressing ALT+MKEY → At center and the top of the box is merged to a single point
- Unselect all vertices (AKEY).
- Box select (BKEY) the bottom 4 vertices
- Hollow out the bottom of the pyramid by deleting the bottom face XKEY → faces (remember that if you want to change to face select mode press CTRL + TAB → faces)
- note: with versions 2.46 of Blender and later, you just have to select merge at center after the merge vertices step to create a pyramid, the last 3 steps aren't needed.
- note: you can do this easier by making the top vertices closer to each other (SKEY)
and then use Remove Double (in editing buttons/Mesh tools tab)
Now that we have the pyramid the easy way, let's learn how to use photos as references to build models. First, unselect all the vertices by pressing AKEY. Next, select the bottom four vertices of the pyramid and delete them with DEL or XKEY. The only vertex left will be the vertex which makes the tip of the pyramid. This will be used later.
note: in Blender 2.48 you can go to edit mode face mode select top face and size it down.
[edit] Window Layout
To make things easier, open the user preferences window and in the "View & Controls" sub-menu, turn on the "View Name" option. The window names will be used for reference inside the rest of this tutorial.
Split the Main 3D view window in to 4 windows.
- Reminder: to split windows, move the mouse to the border of the view, when the cursor transforms into arrow, right-click and choose "Split Area". (Explained in the guide: Noob to Pro/Blender Windowing System.)
Change the point of view in each window so that they end up like this:
-
NUM1 NUM7 NUM3 NUM0
And if you click on View, you can see that these windows are respectively:
-
Front Top Side (Right) Camera
By knowing which view you are looking at you may find that you quickly get the idea of what you are doing and can proceed somewhat intuitively in this section on your own without following all of the step by step instructions.
[edit] Reference Pictures
Make a picture of a white square and of a white triangle in the GIMP, Paint :) , or some other image editor.
Read the following section carefully: Make sure that the drawing of the square is square and not just rectangular. Make the triangle the same width and height as the square. Make sure the apex of the triangle is directly above the midpoint of its baseline.
Suggested method of construction#1: Make the square. Save it, but keep it open. Delete everything except the bottom line. Make the triangle from the lower corners (ends of the baseline) and make sure the two new lines meet exactly above the midpoint of the baseline. Make sure the triangle is visible against a black background as shown in the image. Save as a new file.
Suggested method of construction#2: Download the black and white triangle image on the right of the screen and use that, make a square by filling it white and resaving it.
Suggested method of construction#3: (if you are using Photoshop): make a square selection of "n by n" size, remember the value of "n". Fill it with white color and save. To create a triangle of needed properties make a rectangular selection of same (n by n) size, on a new layer, click RMB on your document, choose "Transform selection" option in the pop-up menu. Once you are in "Transform selection" mode, right-click the blank image again. This time the pop-up menu would be different. Choose "Perspective" from it, and with LMB drag one of the two top vertices toward the other. Once the vertices meet (in the top-center of the image), exit the transformation mode, and fill the resulting triangular selection with white.
Note for Gimp users: turn on the grid (View->Show Grid, View->Snap to Grid), use the rectangle select with a fixed aspect ratio of 1:1 (in the tool options panel) to select a square that you can flood fill. For the triangle, use the node tool to draw a triangular path, convert to selection (Select->From Path) and fill it. Or you could just use Inkscape...
Save the files to a place that is easy to access. Blender only supports the TGA, PNG, and JPG image formats.
[edit] Background Images
Load the white square into the top (NUM7) window by going to the 3D view window and pressing View → Background image → (click on the icon of a file) find your file and click "select image"
Load the white triangle into the front (NUM1) and side (right or NUM3) windows.
If necessary, zoom out so that you can see the whole picture.
Now you have a guide for making a pyramid.
If you can not see the picture, switch to Ortho view by hitting NUM5
[edit] Side One
The vertex that is left will be the topmost point of the pyramid. Use the GKEY to move the vertex around. To get it in the right spot, line it up at the top most point in the front (NUM1) and side (right or NUM3) windows. If you look in the top (NUM7) window the vertex should appear to be in the center. Make sure to keep the vertex highlighted for the next step.
Note: If you delete all of your mesh objects in Object Mode (and remaining vertice), then you'll be stuck in Object Mode. You can however, add a new mesh object as a base, or an empty mesh via a python script (Add → Mesh → Empty Mesh) [Blender 2.49]. Once an empty mesh object has been created, select it (right click on it or AKEY), and then change to the Edit Mode. Now you can begin to create individual vertices with CTRL + LMB.
- (note: it might be helpful at some point to zoom in and use the X,Y and Z movement restriction)
Now, we are going to place more vertices in the scene. Since we want to have the new vertices connected to the first one, we will make sure the first vertex is selected and hold down CTRL and click LMB to create a new one. In the side (right or NUM3) window, place a vertex on the lower left edge of the triangle by holding CTRL and clicking there. This should create a line between your 2 points. If needed, use GKEY to line it up in that window and also at the lower right point in the top (NUM7) window.
Next, with only one of the vertices selected, in the front (NUM1) window place a vertex on the lower left edge of the triangle using the same method. Line it up in that window and at the lower left point in the top (NUM7) window.
Now, select your 3 vertices (use AKEY) in the front (NUM1) window and press the FKEY. You should see a triangle appear.
[edit] Side Two
Press AKEY to deselect all vertices and select the vertex at the top of the pyramid again. Repeat the process from Side One to make the next side. Place a vertex in each of the other corners of the square using CTRL and LMB. Line them up in two different views with the corners of the triangle and square. Make sure only the top vertex and the two new vertices are selected and press FKEY. This will fill in a face opposite of the first face in the pyramid.
[edit] Sides Three and Four
To fill in the other two sides, select the top vertex and the two corner vertices on a side where there is no face yet. Again, use FKEY to fill in a face. Repeat this for the last side to have all four sides created.
[edit] Bottom
In the top (NUM7) window, select all four corners and make a face. You should have a solid pyramid! Now, select all 4 faces, hit XKEY, and choose "faces".
[edit] Taking the Best Reference Photos
Remember what I said about turning a cube into a puppy? We'll do that now.
[edit] Step One: Get the pictures of the model
- Tip: The images here do not line up. Some need to be rotated and others do not match in size. They will be kept so that you can get the "real feel" for this project.
If you have a puppy and a digital camera, take three pictures of the cute little rascal and upload them. If you don't have a puppy, any object or small animal will do. Ideally, the photos will be looking straight down at the top of the puppy, a side view, and a front view. It's important that the puppy be in the same pose in all three photos! Or at least close to the same pose...we all know puppies don't stand still very long.
You could use two mirrors. One is placed next to the puppy at 45 degrees to the camera and 45 degrees to the puppy. Another is placed above the puppy, also at 45 degrees to the camera and 45 degrees to the puppy. This produces three images, one of the puppy (front \ NUM1), one of its reflection seen 90 degrees to the right (side \ right \ NUM3)), and one of its reflection seen from overhead (top \ NUM7). Take the photo from a long distance away with a zoom lens to get close to an orthographic projection.
Or how about pictures of a toy wolf taken from 6 views points?:
Using your favorite image editor, such as PhotoShop or the GIMP, down-scale the images need to a reasonable size (I made mine 512x384), and then match them to each other. To match them, draw construction lines (pulled from the rulers above and to the left) on the left view for example to pick out key features. I picked the tail, the front of the back foot, eye level, tip of the ear, and the front of the nose:
I found when I picked out these features that this first image needed to be rotated slightly. That completed, I proceeded to scale, rotate and shift the other two views (top and front) until they matched fairly well as layers on top:
Once I had the proper results I saved the resulting images, and these are the ones we will use in Blender.
The results are the files you'll need for Step Two:
Just right-click and save them some place where you can find them to load them into Blender for Step Two. You may notice the photos aren't perfect, but we'll use them just to show how you should deal with your real photos. When you are creating your own pictures to import, note parallax. In this example, parallax is present, and we'll attempt to compensate.
- Reader note: I used Paint Shop Pro 7. I used layers and played with transparencies to be able to see where the views fitted best with each others and align them better.
[edit] Step Two: Get the Picture into Blender
Getting the image into blender is the easy part. The more difficult part will be creating the mesh, but first things first. Create a new file (File → New) to see the familiar default objects. Don't bother deleting the cube, we'll end up using it in the tutorial. Just as was done in the "Making A Pyramid" section, split the 3D Viewer into four views and assign them the following view ports. Note that the arrangement is different than that used in the previous tutorial.
-
NUM7 NUM3 NUM1 NUM0
Each window will show you different XYZ coordinates with the NUM0 view being what your object will look like to the camera object.
Now that the screen is set up, let's load the images. In the top left viewport (XY), click View on the viewport's header (the menu at the bottom). From the menu provided, select Background Image .... A small window will appear.
Click the Use Background Image button and several more buttons will appear. Now click on the folder icon located to the right of Image:. A new full-viewport window will appear. Explore this window a bit and end up selecting the image file of the wolf from the top view. What you should get is the picture of the toy wolf from above with the default cube on top of it. Rotate your view of the cube. The picture disappeared! But don't worry - the picture is still there. Press NUM7 to realign the viewport and see the picture again.
- Tip: Be sure to be in orthographic view Num 5 or else you won't see the picture.
Now load the front view of the wolf into the NUM3 viewport (YZ) as you did the top view. Repeat the procedure and load the side view of the wolf into the NUM1 viewport (XZ). As a note, each picture is specific to the viewport it was loaded into. When you switch between axes views, the picture will not change. Try this out by changing the top left viewport to NUM3 and NUM1. Return the viewport to NUM7 before continuing.
The pictures are now loaded into the Blender viewports. If you look at the pictures on the grid, you may notice that the front view of the wolf isn't quite center. That is okay, Blender has a way to fix it.
Move your cursor over the NUM3 viewport and press Shift + SpaceBar, this expands the current viewport (NUM3/front view) to a fullscreen view, then open the Background Image window for the front view again. Notice that there are picture manipulation options available. One of these includes picture offsets. Click on right side of the 'X Offset: 0.00' to increase the offset to 0.20. The picture will be shifted over slightly so now the wolf is more centered. Press Shift + SpaceBar again to return to the four viewport view you created earlier.
Figure 2.2.1 Viewport picture setup
The setup work is now done! Let's start on actually making the wolf model.
Reader Note: If you, like me, weren't paying that much care to how your "wolf front view pic" was aligned from left to right earlier on during the gimp editing phases, then you may need to use a different 'X Offset: ' value than '0.20' that the author recommends. If that is the case, then just use some appropriate alignment value to center it visually as best you can. Hopefully this helps with avoiding some potential confusion for other readers.
[edit] Step Three: Rough Model Fitting
This is a brute force model creation using techniques discussed previously in this book. This section is meant to help you explore and become more comfortable with them. Do not try to follow the example to the tee. Your wolf and my wolf will probably not look the same since you may want to add more or have less detail.
[Note: If you are following this book all the way through, and are just getting started with blender, the following step (step 3 as a whole that is) may likely take several hours to complete. The best strategy is to take breaks, be patient, and with time you'll figure out the best way to go about this step.]
The rough fit stage requires either some planning or on-the-spot decisions. Think about where the wolf will have parts of its body flex or require parts jutting out.
The first step is to create a blocky wolf. Start out with a column of blocks using the extrude face command (select face, EKEY). Don't worry about snapping the vertices to the grid since we are working with an organic figure.
Figure 2.3.1 Body column formation
The next step is to split the ears and legs off of the body. Do this by subdividing the appropriate faces. Save often, and if you make a mistake, go ahead and use the undo option (CTRL ZKEY). Also, if you find yourself looking at redundant faces, combine them (FKEY).
Figure 2.3.2 Appendage formation
]
If you are having trouble with this, try mousing over the perspective window (the one you designated with NUM0) and using the MMB to rotate the view so that you are looking at the underside of the wolf. Click on the face underneath the wolf that is alongside his front legs (use the side view to check this). We are going to subdivide this face in order to grow legs off the new faces. To subdivide, press the WKEY and choose subdivide. You will see that the face has been divided into four. Take one of these faces and extrude it as many times as is necessary to make the right leg. Then do the same again for his left leg. (NOOB NOTE: Be sure to extrude each leg and ear separately exactly as it says here. I multiple-selected both faces to extrude the rear legs from simultaneously and they ended up sharing common vertices at the corners where they meet, making the legs difficult to separate. 2nd Note: Actually it's fine to extrude both, just make sure to select "Individual faces" in the extrude popup menu, rather than region) Use the pictures as a guide.
Doing the ears is similar, except instead of working underneath you will start with the face on top of the wolf which is directly over the ears. Select this face and subdivide it once. Deselect everything using AKEY, then select one of these four faces and extrude it upwards once to make an ear. Do the same for the face alongside it to make the other ear.
Finally, extrude the tail end of the wolf one more time, so that your wolf has as many divisions as the picture above.
Let's start refining the model starting with the tail. Try putting your viewports in wireframe mode by pushing Z, it may make things much easier. Line up the vertices over the wolf in each viewport by lasso selecting multiple vertices (CTRL LMB, Drag). then move to the right location with grab (GKEY).
Figure 2.3.3 Working on the tail
Continue onto the hind legs of the wolf. It is trickier to manipulate the legs so keep rotating a viewport to look at the model from multiple perspectives. Remember that we are working in three dimensions.
Figure 2.3.4 Working on the hind legs
Continue working up along the wolf fitting the blocks to the pictures. If you have problems seeing the picture because the model is in the way, let's hide the model. In Edit Mode, select the entire model by AKEY or by pressing LKEY when you have the cursor over the model. Simply pressing HKEY will hide the selected items. To unhide the view, use ALT+HKEY. By hiding and unhiding the model, or parts of the model, you should be able to keep using the picture as a guide.
Once you have the first pass done, you'll notice that the model just won't fit all three pictures correctly. This is due to parallax. The most obvious example is the side view. The four feet should be level, as they are all standing on a flat surface. Since they are not, we'll just ignore some of the aspects of each picture and continue with the model. (This is a helpful example to show what you need to consider when taking your own pictures.)
Figure 2.3.5 Completed rough fit
Attention: The top right Viewport is said to be "NUM3 - Side" but the Picture shows the wolf from the front! So care and change the picture with the left window at the bottom or change the Viewportsettings
Reader Note: this is because for the pictures to line up correctly the front has to be in the side viewport and the side has to be in the front viewport.
Reader note: The author realizes that he can't line up the images this way in Blender but forget to explain why after changing the position of the wolves. If you put the TOP wolf in a window set up for a TOP VIEW (NUM7) you'll have a VERTICAL looking down wolf. In the same way if you put a SIDE VIEW of the wolf (looking right) you'll have an error in the FRONT VIEW because SIDE VIEW (NUM3) is planed to receive a LEFT SIDE VIEW of your object, the result is that in FRONT VIEW You'll have the ass o the wolf.
Put in your SIDE VIEW (NUM3) an left looking wolf and in TOP VIEW a looking down wolf and all will be in order, or use the side view to represent front and front to show the side as he does at last, wich is much more complicated and difficult to explain.
[edit] Step Four: Refining the Wolf Model
Now that the rough fit is done, let's smooth out the wolf. Add a Subsurf modifier and set the Levels to 2. The wolf will now be smoothed, but we want to add some of the hard lines back into the model. This may be accomplished with creased edges.
First, turn on the view creased edges by toggling the Draw Creases button in the Mesh Tools 1 window. Enter Select Edges or Select Faces mode (CTRL TAB). Highlight the edge or face you want to crease and press SHIFT + EKEY. Use the mouse and pull away from the center until the Crease value is close to what you want. A value of +1.000 will give you the sharpest look and is useful for places such as the bottoms of the paws. When an edge has been creased, the edge will be highlighted in yellow (positive crease) or black (negative crease). These highlights are shown due to the 'Draw Creases' button being turned on.
In this example, I creased edges along the paws, tail, ears, and nose to give them some sharpness.
Figure 2.4.1 Creased edges
The last step is to refit the model to the pictures. You may have noticed that when the model was smoothed, the result didn't quite fit to the pictures. Now is a great time to tweak the vertices to fit to the pictures or add to/modify the model.
And here is my basic wolf based on three pictures!
Figure 2.4.2 Final toy wolf model
[edit] Alternative Wolf Modelling
- (user comment) I was doing this tutorial and though I´m still a noob at this I thought I could contribute a little to this. I decided only to do half a wolf and then mirror it to create a complete wolf. I started out with creating the silhouette of the wolf body in the side view, and then in front view I started to shape the wolf head, tail and legs...
User Note: If you are attempting this method, I read in a much later tutorial that it is best to avoid using triangles in rounded areas of a Mesh, because they tend to cause artifacts in your model (artifacts are protruding edges or other things that don't look realistic). Apparently, using quads is a much better alternative. Triangles should only be used on flat areas of your mesh if they cannot be avoided.
And at any rate this is what I ended up with. Not the best wolf ever, but as I said before I´m still pretty noobish. 
From this point forward there are multiple ways to do the exact same thing, however for simplicity's sake and so that I can be more detailed I will be using one method (the one which I use) and be using GIMP.
I found it best to size all the photos to a known width, with an easy to find center. (Mine happened to be 850x638 pixels, I don’t recommend that but you can choose any size you want really, as long as all of them are the same size). Then drag the construction lines to form a crosshair in the middle of the photo. To do this, click on the top ruler, and drag down to the middle (Exact middle) of the photo, then click on the side ruler and drag across to the middle (Again exact middle) of the photo.
If you are having troubles finding the exact middle of the photo, move the cursor to the very bottom left of your photo and the height of your photo will be listed at the bottom left of the GIMP interface. The numbers are listed in an (x,y) format so you want the first number to say 0 and the second to be the largest you can make it by dragging your cursor. The second number is the height, and half of that is the middle of your photo. You can do the same with the top ruler to find the vertical middle of your photo. Only this time the co-ordinates at the bottom left of the GIMP interface should list the second number (y) as 0, and the first number should be as large as you can make it by moving your cursor (to the upper right of the photo).Once you have your width again half of that will be the middle of your photo.
Then using construction lines put one at the top of your object, and the bottom of your object. Find the "height" of your object by the distance between them. Remove the construction lines from the top and the bottom, and place a new construction line above the horizontal center line by the half of the "height"(of your object). Now place a construction line on both sides of your object and find the “width” (distance between the new vertical lines), then remove those construction lines and place a new construction line vertically half of the "width"(of your object) to the right of the vertical center line. Now cut the object out, and drag it so that the point you used as the "top" is on the horizontal construction line that is above the middle. Then Drag the photo left or right until the right edge of the object is on the vertical construction line you put in right of the middle construction line.
Now the center of your object is at the center of your photo. This is a very important thing because when blender loads in the picture you will need this so that all of your pictures match up with each other 3d. You should repeat these steps with all 3 photos. I also dont recommend doing it in GIMP's "layered mode" as that caused more pandemonium for me. I recommend opening each photo in a new window .
Taking your pictures is the most important part, because if the pictures are not all in the same scale (object size to photo size) then your photos will not line up and you won't be able to place a dot on the same location from front view, side view, and top view.
As a recommendation I would recommend making your first model from a Lego man. That is what I did and it is very simplistic easy practice. To take my photos I took about 10 minutes to construct a photo platform for my object. It consisted of a cardboard box with two sides cut out. I covered the inside area with computer paper. I then used a 2”x4” and a ruler to make sure that the box stayed the same distance from the camera for all shots, as well as marking where the Lego man’s feet were positioned inside the box with a pencil. This will provide good pictures, providing you keep the camera at the same distance and zoom for all three photos.
