Modeling muscles of (biological) mice with Maya

Useful Links:

Adobe Photoshop

Quick Selection Tool: http://help.adobe.com/en_US/photoshop/cs/using/WSFD9BA8C5-31BA-4fec-81F3-CF04EE5295FCa.html

Layers: http://help.adobe.com/en_US/photoshop/cs/using/WSfd1234e1c4b69f30ea53e41001031ab64-78e3a.html

Painting: http://help.adobe.com/en_US/photoshop/cs/using/WSfd1234e1c4b69f30ea53e41001031ab64-780aa.html

Autodesk Maya

Polygon Modeling: http://download.autodesk.com/us/maya/Maya2012_Getting_Started/index.html?url=files/Modeling_a_polygonal_mesh_Introduction.htm,topicNumber=d28e6265

Subdivision Surfaces: http://download.autodesk.com/us/maya/Maya2012_Getting_Started/index.html?url=files/Modeling_a_subdivision_surface_Introduction.htm,topicNumber=d28e11833

UV Texturing: http://download.autodesk.com/us/maya/Maya2012_Getting_Started/index.html?url=files/Polygon_Texturing_Introduction.htm,topicNumber=d28e23782

Autodesk Mudbox

Painting: http://download.autodesk.com/global/docs/mudbox2012/en_us/index.html

NOTE: Different Maya versions may look different. However, you can use the Help system to find all the functions. Specifically, for menu items, select Help->Find Menu… and you can search for a menu command (shown below).

1. Project Setup

A project manages all the resources. Since we are going to have many texture files as reference images, project saves a lot of trouble of searching directories for texture files.

Select File->Set Project, and choose the project folder.

NOTE: Though not necessary, it is good to set a new “project” every time you start a new project.

2. Set coordinate system.

Maya’s default coordinate system is Y-up, while a lot of other modeling tools use Z-up. Confocal data can also be seen as Z-up. So we also set Maya’s coordinate system to Z-up for better compatibility.

Select Window->Settings/Preferences->Preferences. In the Preferences window, choose “Settings” from the left column of categories, and select “Z” as the up axis under “World Coordinate System” on the right.

NOTE: Do not forget to set this setting back to “Y” when you work with your normal Maya projects. When you see your human model in a prostrate position, probably it’s because of this setting.

3. Visualizing volumetric confocal data in Maya

***This step can be done via script, see end of step 3.5***

Different from 2D reference images, which are used for modeling human characters, confocal data are already 3D! We usually volume render 3D images, which Maya does not have a good support. However, we can implement a basic volume renderer of our own in Maya.

3.1 Create a polygon plane

NOTE: If you still have not turned off interactive creation, maybe it is the time to consider turning it off. Select Create->Polygon Primitives->and uncheck “Interactive Creation”.

Select Create->Polygon Primitives->Plane. By default, a plane is created at world’s center, with 10x10 subdivisions and 1x1 of size. You can set the settings before creation by clicking the square to the right of the menu command. Setting properties after creation is preferred.

Show the channel box if it is not shown. With the plane still selected, its transformation properties are shown in the channel box. Expand the creation properties of the plane by clicking on the name under “INPUTS”. We need to change several settings. Basically we want the plane the same size as the confocal image slice, and one of the corners aligned with the origin (0, 0, 0) with the plane itself in the positive XY quadrant. Now check the perspective view to see if it is transformed as described above.

If you cannot see the plane (you probably don’t, since it is too big now), in the menu of perspective view, select View->Camera Attribute Editor…, and then in the Attribute Editor, change camera’s “Far Clip Plane” to a larger number.

Press “A” key when the mouse cursor is within the perspective view, and it should look like the picture above.

3.2 Apply texture to the plane

Select Window->Rendering Editors->Hypershade. In Hypershade, create a “Surface Shader” as shown below.

Double-click the icon for the newly created surface shader, to show its attribute editor.

Click the checkerboard icon to the right of property “Out Color”, and in the “Create Render Node” window, select “File”.

The file node’s attribute editor will show automatically. Choose the confocal image of the first Z slice. Notice the image file’s path is relative. That’s because we set the project earlier. If the path is not relative, Maya may not find the proper file when the project is copied from one computer to another.

Now click the Hypershader toolbar icon “Input and Output Connections”, and the icons in the Work Area are clearly laid out.

Select the surface shader’s icon, and its attributes are loaded. Middle-drag the file node’s icon to “Out Transparency” of the surface shader.

Select the plane in the perspective view. In Hypershade, right-click-and-hold on the surface shader icon until the marking menu shows up. Without releasing the right mouse button, move mouse cursor up and choose “Assign Material To Selection”. Hit “6” key when the mouse cursor is in the perspective view. The perspective view should look like below.

Select the surface shader again. Under “Hardware Texturing”, set “Texture channel” to “Combined Textures”, and “Texture Resolution” to “Highest (256x256)”.

3.3 Duplicate the surface shader

In Hypershade, select the surface shader node if it is not selected. Then select Hypershade’s menu Edit->Duplicate->Shading Network. An identical surface shader (except for name) will be created. Repeat the duplication N-2 times with N = number of confocal Z slices, so that we end up having N surface shaders in total, including the two already created.

NOTE: Menus can be torn off by clicking the dotted line at its top. It’s good to tear off the menu and do repeating commands.

Click “Texture” tab in Hypershade, and we can see all the file nodes. It is better to sort the icons by time, otherwise “file1” is always followed by “file10”. Double-click each icon and bring up its attribute editor. Change “Image Name” to the corresponding confocal slice image. For example “file10” is going to be the tenth slice.

3.4 Duplicate the plane

Select the plane in perspective view. Select Modify->Freeze Transformation.

Select Edit->Duplicate Special and click the square instead of the text.

In the option window, set “Z Translate” to the distance between two confocal slices.

NOTE: MAKE SURE THIS NUMBER IS CORRECT BEFORE MODELING. THE MODEL WON’T ALIGN IF THIS NUMBER IS NOT CORRECT.

Set “Number of copies” to N-1 with N = the number of total Z slices.

Now the perspective view should look like below.

3.5 Assign textures to planes

Now we should have the same number of surface shader nodes and planes. Select each plane from the perspective view and assign the corresponding surface shader in Hypershade. Then you have implemented a volume renderer for confocal data in Maya!

USING A SCRIPT: Most of the repetitive work above can be done with Maya’s script language, MEL. If you happen to know Python, you can also use it to automate the job. For your convenience, a MEL is provided and you can download it here. To run the script, copy its code to Maya’s “Script Editor”, read the code and make sure you understand, modify the variables in the code, and click “Execute All” button in the toolbar. You still need to set far clip plane of the perspective camera and the texture resolutions. Refer 3.6 for setting texture resolutions.

Example of settings:

int $isizex=2765;//width of one z slice(unit=pixels)

int $isizey=4506;//height of one z slice(unit=pixels)

float $fspcz=16;//space between two neighboring z slices(unit=pixels)

int $iznum=29;//total number of z slices

3.6 Manually increase texture resolution

Open Hypershade by selecting Windowà Rendering Editorsà Hypershade

Select first surface shader node. Double click the node to open Attribute Editor. Go to Hardware Texturing, and then select Texture Resolution. For muscles, we use the highest available setting (256x256). Select surface shader nodes one at a time and change the Texture Resolution. For tendons and nerves, we make the resolution even higher (Windowà General Editorsà Attribute Spread Sheet; The last column is Resolution. We set this to 1024 for tendons and nerves).

NOTE: increasing the texture resolution consumes more graphics memory. Keep the settings low if you are using low-level graphics cards.

3.7 Group the stacked confocal images

Open Outliner by selecting Windowà Outliner. Then select all planes in the Outliner. To group the selected planes, select Edità Group.

Open Channel Box/Layer Editor, under the Display tab of the Layer Editor, create a new layer and rename the layer “Confocal” (or whatever you choose).

4. Create a prototype muscle

In this example, we will create a spindle-shaped muscle, which is used as a prototype for further modeling of different muscles.

4.1 Creating a rough model

Create a polygon cube. Set its XYZ subdivisions to 4. Scale properly and make it look like above.

Right-click-and-hold on the cube until the marking menu shows up. Select “Vertex” to go to the component mode. Move the vertices of its two ends and make a shape as above.

4.2 Smooth the model

Select the object from the marking menu. Make sure the menu set is set to “Polygons”. Then select Mesh->Sculpt Geometry Tool, and choose “Relax” as the Operation under “Sculpt Parameters” of the Tool Settings. Adjust the brush radius to a proper value and paint on the cube mode. Smooth angular edges and make them smooth as above.

4.3 Create UVs

Make sure the menu set is set to “Polygons”. Select Create UVs->Cylindrical Mapping. Modify the cylindrical mapping gizmo and make it look like above.

Select Window->UV Texture Editor and open the texture editor.

Select “UV” from the marking menu and fix the UV as above.

4.4 Convert the model to Subdiv

Go to object mode again and select Modify->Convert->Polygons to Subdiv. Hit “3” on keyboard. It is now smoothed as above.

Use the marking menu can choose “Polygon”. Then use the marking menu again and choose “Vertex”. It should look like above. The prototype muscle has been created.

NOTE: A prototype model can be downloaded here (if you have a Maya of version earlier than 2012, try this one).

Use the prototype model for spindle-shaped muscles. Import the prototype muscle into the volume renderer generated in Section 3. It may be too small compared to the planes. Scale, rotate and translate it to fit the reference image. Tweak the vertices and model its shape.

5. Modeling structures of general shapes

For structures that can’t be easily modeled from the prototype muscle, we use more general modeling approaches. Here is an example of modeling a muscle from a box.

5.1 Create a box

Create a polygon cube. Keep its XYZ subdivisions to 1. Translate, rotate, and scale the box to cover just part of the muscle.

Go to vertex component mode. Move the vertices of the box so that matches the part of the muscle it covers.

5.2 Extrude the faces of the box

Right-click-and-hold on the object. Select Faces.

To extrude the selected face, select Edit Meshà Extrude.

Select the Move tool from the toolbar on the left side of the screen.

Repeat extruding faces until your polygon model approximates the confocal rendering.

Making the model more transparent can help with modeling. You can do so by first go to Windowà General Editorsà Hypershade, then select the shader of the model. Set the transparency of the shader in its attribute editor.

Delete the histories on finishing the model. To delete history, select Edità Delete by all typeà History. The history will no longer appear in the Channel Box/Layer Editor.

6. UV unwrapping and finalize the model

6.1 Unwrap UVs

To apply a texture onto a polygon model, you must first unwrap the existing UVs. UVs, or texture coordinates are generated by projecting the vertices of the polygon model to a 2D plane. Here we use Maya’s Automatic Projection first and stitch the UVs together.

For an Automatic Projection: Go into object mode and select the object. Then select Create UVsàAutomatic Projection, select 6 planes, which will create 6 planar projections, and select “Optimize for fewer pieces”.

To see the UVs, open the UV Texture Editor by selecting Windowà UV Texture Editor. The model will look like a puzzle, and it will need to be pieced back together.

In the UV Texture Editor window, you can select an edge of one of the “puzzle pieces,” and it will be highlighted, as will its corresponding edge. Then use the “Move and Sew Selected Edges” button to stitch together the pieces.

When all the UV pieces are stitched together, move the stitched together flat model back into the unit square in the window. Make sure all the UVs are within the range of the unit square.

To check the quality of the UV unwrapping; Windowà General Editorsà Hypershade. Select Lambert, and set its surface to the checkered pattern and the color to red. Look at the effect of checkering on the model.

6.2 Convert the polygon model to a subdivision surface

Select the polygon model and make sure it’s in object mode.

Select Modifyà Convertà Polygon to SubDiv.

Hit “3” on keyboard. It is now smoothed as above.

6.3 Fine tuning of model shape

Use the marking menu can choose “Polygon”. Then use the marking menu again and choose “Vertex”. It should look like above.

Use the vertices to further adjust the model.

7. Texture

7.1 Create texture in Photoshop.

Create a new Photoshop file. Set the resolution to 2048x2048.

Draw fibers based on volume rendering of the muscle.

7.2 Export the model from Maya as OBJ

In Maya, select the subdivision surface that we want to export, and then select Modifyà Convertà SubDiv to Polygons (Select square for creation options). Set the options to: Uniform tessellation, Level 3, and then click Convert.

Fileà Export Selectionà Export the OBJ file.

7.3 Paint texture in Mudbox

Open Mudbox.

Open OBJ file of model.

In object list, select the model and right click on the selection. From the popup menu, select Assign New MaterialàMudbox Material. Now we created a new material for the model.

In the material property box, select “All channels” for “Opacity Affects”.

In the Layers panel, select Paint Layers. Click the “Create new paint layer” button to create a new paint layer. Set the paint layer type as “Diffuse”.

In the case of limb muscles and tendons, it is important to paint transparency on tendons (while in Mudbox) so that the muscle and tendon intersect properly. To create an opacity layer, do the same as creating the diffuse layer, but choose the paint layer type as “Opacity”.

Now we need to use the projection brush to paint the texture. Select Paint Toolsà Projection.

Select the image browser tab and then open the folder with the texture file in it. Select texture file within Mudbox and click “Set Stencil”

Orient the texture on the model.

Paint the texture.

Save the Mudbox project after painting.

8. Ready the models for FluoRender

Find the diffuse and opacity (if present) files within Mudbox’s project folder, and open them in Photoshop.

Use the opacity files as the alpha channel of the diffuse file.

Add a new layer to the diffuse file. Use the bucket tool to fill a solid color to the new layer.

Set the blend mode of the new layer to “Color”.

Flatten the layers and save it as a TGA file.

Put OBJ, MTL, and TGA files in the same folder. Use a text editor to open the OBJ file and the MTL file.

Edit the MTL file as follows:

newmtl lambert

illum 4

Kd 0.00 0.00 0.00

Ka 0.00 0.00 0.00

Tf 1.00 1.00 1.00

map_Kd [the TGA file name]

Ni 1.00

Find the line starting with “usemtl” in the OBJ file, and replace the material name after to “lambert”.

Now the OBJ file can be opened in FluoRender, with its material and texture properly linked.