Compiling and Importing a Shader:

Once you have a coded *.sl file, you must compile it using the “shader” command. This is as simple as calling the program with the name of the file after it like “shader file.sl”. The file it outputs will be a .slo file with the name of the shader defined within the .sl file (not necessarily the name of the .sl file itself) which you can then pull into a RfM node. In my case, I will be using a combination of a displacement and a surface shader to create a look like a shiny coat of paint that is chipped away.

To pull in a Renderman shader, you use a Renderman shader node which RfM creates an entry for in the Hypershade window with the rest of your materials. Here is an example of the attribute windows of the two shaders I used, followed by another earlier one that has a varied opacity:

The attributes are empty when the nodes are first created and the “Shader” text box right under the Material Sample must have a valid .slo file loaded before it reads and presents the proper attributes. My surface shader has several fields to control color of the paint and the underlying surface while my displacement shader controls how much to displace and how much of the surface should be chipped.

Linking Surface, Volume, and Displacement Shaders:

Using a Renderman displacement shader at the same time as a surface shader is the same as it is in Maya normally–one must link the displacement shader into the “displacement shader” attribute of the Shading Engine. To make this link, from the Hypershade window, hold right-click over the surface shader and select “Graph Network”. From here, drag the displacement shader into the work area and use the connection editor to connect the “displace” attribute to the “displacementShader” attribute:

The same process can be used to link a Volume Shader to the Shading Engine as well. With all that complete with some tweaks of the attributes of the shaders, applying it to the object, and adding some lights and shadows we can see a sample of the shaders:

The Scene:

The scene used is #18 from the same site:

http://www.3drender.com/challenges/

I applied the shader from before to the robot’s main body. For the rest of the scene we should keep in mind that we can’t use the mia_material_x we’ve relied on for the earlier projects as they were MentalRay nodes. However, we can still get results by tweaking the render settings of RenderMan and setting up some environment lights.

First thing to do in a scene like this which will have plenty of reflections (for me, at least) is turn on Ray Tracing in the render options:

Also, I wanted global illumination effects because RenderMan’s color bleeding looks very good so at the bottom if you hold right-click in the text box next to “Environment Light”, you can create a environment light directly. Here’s an example of my environment light attributes I set up for the scene, of which most important is the shadow type change to “color bleed” and the sampling mode to “sampled”:

Another important thing to do is set up shadows in the scene. RenderMan will work with standard Maya depth map shadows (though I am not sure about Ray Traced shadows) but I prefer the RenderMan custom shadow attributes. To access these attributes, with a light selected under the lightshape tab, click the “Attributes” dropdown menu right above the tab listing and you will see “RenderMan” at the bottom of the menu. Hover over this and you can see “Add Shadow Attrs” as well as “Create Custom Shadow Map”:

The new attributes will be appended to the very bottom of your attribute listing, under Extra RenderMan attributes. The DeepShadow shadow type is generally better as it allows for a few extra features such as transparency. NOTE: Right under where the “Create Custom Shadow Map” option was is “Create Custom Deep Shadow Map”. From what I was able to tell, custom deep shadow maps are extremely expensive and not recommended until final render if you must use them.

Finally, when you are mostly done with your scene and are ready to do a full render, you should turn down the Shading Rate in the render options menu for a higher quality render:

This option usually defaults to 5.0, but 1.0 to 0.5 generally works from my experience. However, the extra time taken for a lower shading rate seems relatively small so turning it all the way to 0.1 or just rendering all the time at 0.5 might not be too much of an issue depending on your scene and setup.

Here is the end result of my scene:

http://www.3drender.com/challenges/ Challenge #21: The Bedroom

Lighting:

Once again, I went for a more natural look in the same fashion as I did with the King’s Treasure scene using physical sky and portal lights. This time, however I had to use Final Gathering to get enough light into the room.

One note is that if you decide to use an ambient light to brighten an overall scene like I did, turn off its Ambient Shade attribute so objects won’t be affected by shading coming from it:

Texturing and Materials:

Again, the general techniques are the same as for the King’s Treasure scene. One issue one may arrive upon when texturing the posters is that due to the posters’ deformations, the UVs don’t line up with a rectangular image. While doing the export and texture in Photoshop would work for these posters I thought they were too simple and instead directly textured in Maya. To solve the UV problem, select the object and go to Window->UV Texture Editor:

Here when you seelct and stretch your UVs to size, you can see the corner doesn’t form a proper rectangle. You can tweak the UV vertices directly into place in order to rectify this problem. However, have your actual mesh visible somewhere because you want to keep the UVs as close as possible to the relative shape of your mesh to avoid distortion.

Another issue you may run into is that some of the objects come as a single mesh (such as the Laptop). While you could create a single texture for the whole object it is more work and the whole object would still have to share the same material properties. An easier way is to select the object, enter Face mode and select the faces in question. Then, while in Polygons mode, go to Edit Mesh->Extract:

This will split the faces off the old mesh, allowing you to texture these as seperate objects. I usually delete the History on both the resultant meshes, but that is optional for this scene.

Render Layers and Ambient Occlusion:

One feature of Maya is that through the use of render layers, you can render your objects in multiple different ways (or different objects altogether) in one pass and have Maya do a simple composite of them all in one pass. For this project I used a render layer for Ambient Occlusion by selecting all non-light objects and adding them to a new render layer.

After doing so, right-clicking on the render layer and select Attributes. From here go to the Presets menu and select Occlusion. Also, in the bar directly at the top of the render layer section, go to Options and then the settings box for “Render All Layers”. Here you can choose whether you want Maya to do compositing for you or just render the layers for you to take care of yourself. Finally in the dropdown box above your render layers, you can choose the method by which they are composited by Maya. For now, use Multiply for the Occlusion layer.

General Tips and Issues:

My Ambient Occlusion composite makes the scene way too dark!

This happened to my original scene. For some reason, the AO pass had only greys and when multiplied, had the effect of darkening everything. If you encounter this, the easiest workaround is simply to change the compositing mode on the render layer to “Overlay” rather than “Multiply”.

There are way too many objects!

Easiest way to work around this is simply to first determine your camera angle and then texture only the objects visible. However, I failed to follow my own advice and ended up assigning materials to everything.

My textures are way way too bright!

This also can happen. This can be fixed by going to the File node of the material you assigned the texture to. Then you can see a Color Balance attribute category. In here, if you turn the Color Gain down, you can darken the texture. Also, changing the Default Color can make it easier to keep track of which objects in your scene are textured without having to turn on Textures for the viewport.

Sample Final Product:

Object Close Ups:

http://www.3drender.com/challenges/ Challenge #19: King’s Treasure

Lighting:

For this file I decided to use a Physical Sun and Sky, which one can turn on through the general render options in MentalRay:

However, one can see that just this lighting is extremely dark so we will take advantage of the mia_portal_light light shader, which essentially re-emits the environment in a direction you specify. First, you create area lights (I made one for each window, totaling 4) and size them so that they cover the entire window. Next, in their mental ray attributes, you must turn on the light shape (be sure to check “visible” as well):

After that you can go into the Hypershade window and switch to “Create mental ray node” and go down to the light shaders where you can click to create a mia_portal_light shader. Once you have one in the Work Area, select an area light and find its light shader attribute and middle mouse button drag your portal light shader into the “light shader” label:

Now you do this for each area land and your portal lights are set up. To increase the brightness of the scene, go into the utilities tab of Hypershade and from there you can select your Physical Sky and increase the multiplier value. You can also change the multiplier on your Portal Light shaders (I used 2 different ones for the windows in front of and behind the camera to independently adjust light intensity from each side of the room.

Also you can select the new sun light automatically created and have it emit photons for caustics, etc.

Texturing and Materials:

Since most of the things in this room are metal/glass, there is not much texturing to do. Most of the shading in this one I handled with the mia_material_x mental ray shader, which has many presets available to give you a start. I used this for everything in the scene except for the gemstones, chair wood, and bricks in the background. The best way to learn how to use them is really to look through the presets oneself and play with the settings (which each attribute if you hover over their label gives a surprisingly detailed description in the Maya status bar at the bottom). For a few examples I started with chrome for the sword metal and lamp bronze and copper for gold and worked from there.

For the gemstones I simply refined the Phong_E shader that was assigned to them from the start. From there the main things to change were color and transparency tweaks but most importantly, to turn on refraction in the materials:

Another, more physically accurate approach is to use dielectric_material, found in the mental ray nodes to model the gems. I found it unnecessary for this scene but a good tutorial to look at is:

http://www.3dtotal.com/team/Tutorials/diamond_caustics/diamond_01.php

For the bricks, bag, and tablecloth I used a procedural bump map to create a texture for them. Sample brick network:

Very basic network which works. You might notice no matter how fine you tell your noise/fractal/etc shader to be, it is still too big. That is when you select that place2dTexture node and in its attributes adjust its RepeatUVs (near bottom of the picture):

Finally for the coin, I created a custom bump map. The inital UV layout that comes with the file is fine for this purpose so you can simply take a UV snapshot from the UV Texture Editor window and import it into photoshop and draw or paste a picture with the lines as a reference. My coin:

General Tips and Issues:

I want depth of field!

Look in your camera’s attribute editor and find these attributes:

Now do not turn it on until you are done with everything else. It massively inflates render times.

Everything looks grainy!

Turn up your sampling in your general render options:

This is another thing that will majorly affect render times however. Max samples 2-3 is generally plenty.

Render times are still too long!

You can turn off final gathering and try decreasing reflection refraction trace depth. You’ll generally want higher refractions than reflection depth unless your image has very obvious places where people can see multiple reflections very clearly.

Sample final product:

http://www.3drender.com/challenges/ Challenge #1: Fruit Bowl

We will use Maya’s obj export plugin and Photoshop’s 3D abilities to texture and render the fruit bowl.  The first thing to do is decide which of the fruits we wish to do procedurally (create wholly with Maya’s hypershade nodes) and which we want to use custom textures. When we do, we convert the ones we wish to use custom textures on to polygons (recommend quads).

Then, set up the UV map for the objects (easiest, but imperfect way is use Automatic Mapping) and export a selected object as an obj. You can then read in this file into Photoshop and get something like this:

The 3d tools are near the bottom of the toolbox. In this mode you can freely paint onto the objects or project a texture from another image onto the 3d model by merging a 2d layer on top of the 3d one (most easily done with “Merge Down” which is Ctrl/Cmd+E) Continue to texture the model and when finished, right click on the 3d layer in the Layers box and select “Export 3d Layer”, select “obj” file type again and save the file. You will note in the same folder a psd file with the texture map has now been created in that folder as well. I found it easiest to refine the texture by working directly in this 2d represenation directly since Photoshop’s 3d representation can leave holes in the model. In this view, patch up any holes with a brush/clone stamp/healing brush and you will get something like this (texture for banana):

You can continue to do the rest of your custom fruits in this manner. For an example of a procedural texture, we can look at the orange. Since its geometry is highly detailed as it is, we can use procedural textures to easily add an extra bit of detail. Using a combination noise nodes in Hypershade, we can pipe them into the LayeredTexture node and arrange their opacities and colors relative to each other before sticking it in to the shader. You can also use the alpha channel of the node to create a bump map. For the orange, here is the sample graph:

From here, the rest is really just constantly toying with the attributes within each shader and constantly test rendering until you get the results you want.

Sample Issues:

Among some of the problems you may run into:

MentalRay won’t render at all:

This problem most likely occurs when you use a regular File node to apply a custom texture rather than the PSD node, which is specifically required for PSD file types. Be sure you are using the right type of node.

It is so boring constantly having to paste and project small slices of texture into Photoshop and heal their edges:

Easiest way I did it was defining a block of relatively consistent texture as a pattern and using the Pattern stamp tool as a replacement for my paintbrush. You may notice edges appearing still (though a lot less often than if you were to use pure projection), but those are easily fixed when editing the texture in 2d, where whenever you see an edge appearing you can toggle the “keep alignment” box at the top and painting over the seams. (However, this simply changes where the seams appear, so be sure to turn it back on in places you don’t have seams)

I am seeing seams when rendering with MentalRay.

This happens sometimes because Photoshop isn’t perfect with how it handles the model and UV’s so the most robust way is to simply extend the textured regions in the 2d view. However, doing this manually is still relatively time consuming so the easiest way to do it is to create a new layer UNDER the texture layer and using the pattern with the texture mentioned in the previous faq, simply paintbucket it onto this layer, covering all remaining holes in the textures. Thus, our banana from before now looks like this:

While this isn’t easy to read, the essential parts are still there and there is no chance of transparent holes or seams. If you need to go back and further edit something, simply toggle the layer invisibility.

Final Image:

This is a simple exercise where we simply stabilize a shaky camera. This is a minimal node tree for just stabilization:

The stabilize node only moves the frames so that they will remain as close to the tracked points original locations as possible so we need the Window (Crop works just as well) node to cut out the excess to give the illusion of a stable camera. Unfortunately we have no choice but to lose some of the original image as a trade off.

Here is an overview of the tracked nodes:

I chose these points so i would have two in the foreground and one in the background, creating a triangle in perspective. Once you have tracked your points, you need to activate “applyTransformation” in the properties of the Stabilization node in order for it to move the camera.

The result:

02 – Skilift Tracking

Now we try to matchmove a composited element with a moving live action element:

This is our node tree:

The tracking in this shot is much more difficult since we’re tracking a foreground skilift that crosses the path of other elements twice, muddling the tracker. To amend this, we can allow it to reach the end of its camera presence and delete the errant frames and allow shake to interpolate the motion in the problem areas. However, we can also see that as our skilift leaves the camera, the tracker refuses to move and follow it off the camera.

This can be addressed in two ways: we can manually keyframe the tracking at a later frame and use the curve editor to try to match the interpolation line to the video. Or, we can use the offset feature of tracking and tracking something else on the same object. I went with the former for this since at the time I was not aware of the alternative.

I tracked both sides of the lift as well as used an average of the two for this result:

Notice in this one the tracking area is very large. At this time I also had not realized the area is per frame, not across the whole duration of the video. Keeping this area as small as possible makes your tracks even easier to be accurate.

Once you use a move node to position yourself properly and try to correct color, you can tell the MatchMove node to applyTransformation and see that we still need to mask the hands so that they don’t show through the ski lift:

Here I simply make a rotoshape and animate it along with the ski lift to mask my hands:

Finally, our original color key was bad because of our source video so there’s a lot of flickering in the arms and legs so I over’ed several more rotoshapes of appropriate color to hide them.

The result:

03 – Camera Matchmove

Here we composite a cg element on top of a moving camera.

Node Tree:

Similar process to the last one, we find several good points on the source and track:

Position and MatchMove the turret and check the perspective. Relatively simple now that we’re used to it after the last composite. However, now we need to take care of the shadow from the tree so we create a DropShadow and Over it on top of the original to create a shade. The DropShadow is a little smaller so we tweak its size with a DilateErode node so it completely covers the original. Looks plausible now, but boring. We notice several patches of light that shine through the tree’s leaves so we mimic it with a RotoShape and an Outside node to create several holes in the DropShadow:

We add some blur to the rotoshape to soften the edges but still see that its too smooth and perfect compared to the surroundings so we add a pixelize to it as well. Looking at it, the light difference is too great so we also throw in a Fade node on the RotoShape to dim the bright areas a little further.

The result:

Combining these two videos:

This is the shake tree:

First we key out the green screen with chroma key, adjusting the sliders while viewing the alpha channel so that only the figure is white. In this video, since the figure will not move, instead of creating a garbage matte, it is easier simply to crop the video.

Then we go ahead and overlay it over the background video and adjust the huecurves for green suppress, looking at the background video. Noticing the quality of the background video, also added a SMALL amount of defocus and some pixelation and FilmGrain.As an example of how much different it would look compared to the original:

to

Now we scale him, referencing the people in front of and behind him and move him into position. Notice that there’s not enough space between the man and the 2 girls so we will have to roto him. We create 2 rotoshapes for the man: 1 for the head and the other for the body, since one will not change nearly as much.

Now since the movement of the escalator is constant and perspective is of a minimal factor, you can simply keyframe the first frame and last frames for scale, and it will interpolate in between. The position, on the other hand, will need more work since you will need to nudge the figure sidways so his hand stays on the railing. Simply move him into every 20 frames or so and keyframe. Be sure to look at the actual escalator stair he is standing on and make sure he stays on that stair.

Now the result:

02 – Render on Live Action

Combining this render animation with this video:

This is a closeup of the model, an extremely simple ufo:

This is the tree:

We start by simply overlaying the ufo with the beach and aligning the animation to what we want. This will be a simple composite–the object floats and is further in the background, freeing you from the burden of detail (and lack of modeling skill).

First we colorcorrect and find that not much adjustment is needed–only a little darkening and slightly increasing the blue.

Then we create a matte for the house that the ufo flies behind:

We find that the background video’s camera is actually moving so we will need to animate the rotoshape just a little–for the 5 frames or so the ufo is in contact with the left and right sides of the building.

Now we have the ufo, but we need a shadow or else it really will be “floating” over the picture:

You can probably tell from the tree that I originally intended to use a drop shadow but since the water is a flat plane and the ufo moves towards and away from it, it doesn’t actually work out. Therefore, I roto my own shadow again with a dark green color so I can control the way the shadow moves. We animate the movement to follow a little behind that of the ufo due to the direction of the light. Remember when the ufo dips down closer to the water, the shadow will move forward and vice versa as well as grow darker and lighter.

Now the result:

03 – Live Action on Render

Combining this video with this still render:

This is the tree:

Note that there are a lot of node groups for the shadows, will cover those soon.

This time the figure is moving so we can’t simply crop so we make a garbage matte for the lights:

The bottom left one will be animated when the figure’s legs cross over.

As always, color adjust the video and add defocus/pixelize as necessary. Since our background is a render, we won’t need as much filtering on one, but I added a little to both to try to make them fit more together.

The main issue with this one will be the shadows, as they are very visible in the render. Here are is the tree’s shadow groups expanded:

And here is a sample of the result overlaid on top of the original video to show the relative placement and effects of the garbage and holdout mattes:

The wall shadows are easiest, we simply reposition a drop shadow and adjust color and opacity to match those in the render and add a holdout matte. I created different ones on the different walls as they will have different properties due to their angle. Also, note that since the render shadows are hard-edged, we reduce fuzziness on our shadows to 0.

For the ground shadow, the perspective is much more obvious. At first I had been using a Warper node, but that did not work out past a certain point, so I made another ground shadow group that “takes over” past a certain frame using the CornerPin node. Remember to keep the feet of the shadow touching the feet of the figure and the legs connecting to the shadow on the walls.

I also quickly roto’d the shadows on the streetlight using the same animation principles as in the ufo composite.

The result:

Combining these three pictures:

Everybody likes Vince.

This is the node tree:

First we key out the green in the frame and use an Over node to superimpose it over the Mona Lisa, adjusting size of either picture to suit our needs:

Somewhat messy since the frame I decided to use was rather low quality, but hopefully filter and file size can mask this error later. We roto Vince’s hand next:

This we use this on the picture of Vince in order to separate his fingers, allowing us to put it over the picture he’ll be holding. Later we will also copy this rotoshape and soften some of the edges in order to create the mask for his drop shadow as well. Next we chain Over nodes so the fingers are above the picture, which is above the rest of Vince. After seeing the initial result, insert move, scale, and rotate nodes and adjust until it is in the proper position in Vince’s hand.

Now add a dropshadow to the RotoShape of Vince’s hand. Copy the roto as previously mentioned and use it to mask the parts of his hand that should not be in the shadow. Move the shadow as desired to match the light source:

02 – Render on Live Action

Without MentalRay this would have been a lost cause for me. We will be making a shiny toaster and put it into this scene:

Since we will be making the model shiny (which is awesome), we will need to create a minimal environment with it that mimics the photo (which is not awesome). Thus here is a wireframe overview of the result:

We have created a plane for the table face, added several walls, and an apple bowl. We texture these with small clips we crop from the original picture such as the tiles and the apple skins. The result isn’t pretty in the modeling window, but when viewed through the toaster’s reflection, it should be enough. Also notice that though in the original picture only 1 light seems to be visible, we have used 4 ambient lights in Maya. This is because I have not yet figured out how to properly set up lights in MentalRay and it is the only way I could get the results to look acceptable. Here is a closeup of the toaster and bowl:

And here is the resultant render:

The apples look terrible, but all we need is the toaster. Also notice in the original picture that the table surface is somewhat reflective so we need a shot from below the toaster too since the pattern on model table is not exactly the same and would not merge too well.

Now we are ready to composite. Here is the node tree:

First we roto the two model renders:

As before scale, rotate, and move the pictures as necessary. We also want a very minor application of Defocus on the main toaster so it doesn’t appear too sharp. Now we need to adjust the reflection toaster as the reflection on the table is not perfect. We want it to fade as it goes towards the bottom of the image, so create a Ramp node and mask it over, creating your fade effect. You will also want to apply the roto for the relection you made to the ramp to prevent it from affecting the rest of the image later:

I also use a IBlur node to blur on the X coordinates with another ramp in order to disperse the reflection as it goes down. Now over the toaster over the reflection and the reflection over the table:

The shadow was created with a DropShadow node and since I couldn’t figure out the Warp node, I used the Twirl node instead and adjusted until it looked acceptable, though it is obvious when thinking about it the shadow extends too far to the immediate left.

03 – Technically Not Following Instructions: Another Render on Live Action

It wasn’t until I was far into this one before I realized that while the environment was rendered, it was still technically a foreground element. Oops. I used these images.

And here is the overview, sample, and result of the render I used:

I added extra walls that weren’t visible to block light bleeding from through the plane. At this time, I used the city picture on the plane outside the window to stage the backdrop. This is actually a bad idea since i realized in Shake that I would just have to roto it out later anyway to put the person in front of it. This is the node tree:

Did not get very far so it is a simple tree. As always, roto what you need first:

I had also used a simple 4-sided roto to remove the city from the render picture as well, but that is very simple so it probably does not need demonstration. The Overed result is as such:

Now also add some motion blur and defocus (never much needed for most photos) the foreground to match the rest of the picture and stick it back over the original image of the city skyline:

04 – Live on Render

Getting back to the assignment, I use the same environment but put a parking meter in the image instead of the skydiver. The meter:

Having learned my lesson, I replace the plane with the city texture with a magenta screen:

Node Tree:

I have also adjusted the position of the outer light to better match the lighting in the city. Now we roto the meter:

Now we need to consider the lighting on the meter due to the relatively strong lighting in the render. I decided to simply roto the shaded side of the meter manually:

I combined this with a mask over a black color layer and a blur to make my own custom shadow:

Merge the render with the city before we continue:

Flip, scale, rotate, and move the meter into position. Now notice the meter will also cast a drop shadow on the ground so I did another roto on the black layer in the same fashion to create it:

As always, color correct, defocus, etc as necessary.

In this tutorial, we will be using these 4 nodes, all on the filters tab.

Emboss

The emboss node adds a depth effect to a flat image, simulating a light source from a certain direction and manipulating the lightness and darkness of the edges to make the image “pop” in or out. It is commonly used specifically with text to make it look engraved or carved in a surface. Usually, the filter does not do well with more complex images, but can still be used to similar effect.

This is our shake tree for an example where we add the text “Here Lies Johnson” to a blank gravestone:

Johnson was a good man–just another grim example of a innocent bystander caught in the daily chaos of digital image manipulation.

1. First, we import the base picture we will use.
2. We then create a text node. An image node can be used as well, but text is easier to work with for starters, so we create a text node with our desired text and match the dimensions of the node to the dimensions of our target picture.
3. For now, ignore the Set Alpha node and create your Emboss node. Link the output of your text node into the emboss node and once you double click on that node, you should see the effects of the default emboss settings. You can return to manipulate your emboss settings to match depth, material, and lighting later on.
4. In this case, our base picture was slightly blurrier than the text, so we added a Blur node to make it fit better.
5. The darks in the emboss were too dark, so we inserted the Set Alpha node after the text in order to lighten it.
6. Finally, we create an Over node and overlay the embossed product over the image.

Finished product:

Sharpen and Edge Detect

The Sharpen and Edge Detect nodes are named for exactly what they do. The Sharpen node increases the “sharpness” of an image, while the Edge Detect node creates a black image with white edges corresponding to the original image.

However, the Sharpen function generally creates unwanted side effects in the image, usually exaggerating any texture in areas that did not need to be sharpened. Therefore, these two functions can be combined to restrict the sharpen effect to the edges.

We will be using this image:

This is our Shake tree:

1. We have our imported picture at the top and create our sharpen node.
2. Sharpen the image. You will want to use a lot of it by increasing the “pixel” and “strength” properties of the node–focus on how the edges look and try to ignore any resulting unsightliness elsewhere. It should look THIS bad:
3. Create an Edge Detect node and pipe in the results of your sharpen.
4. Create a Threshold node (or edit the “threshold” property in the Edge Detect node, however this provides slightly different results) and adjust the properties until you remove most of the non-significant edges from the edge detect result.
5. Create a SwitchMatte and pipe in your sharpened image and your thresholded edge. The result should look like a colored version of the Edge Detect node.
6. Create a Over node and overlay the results of the SwitchMatte over your original image.

Result:

Note that even though the sharpen was so strong, the end result is still a little subtle. Increase sharpen as necessary for stronger results.

ISharpen

The ISharpen node is a sharpen node with an additional input for a mask it can multiply by. This and the other “I-” nodes (such as ISharpen, IBlur, IDefocus, IPodjokesaren’tfunnyanymore, etc) are all structured in this way and provide more accurate results than manually masking the normal version of this node.

Here we simply combine it with a Ramp node to create a sharpen effect that gets stronger as it progresses towards the right of the image.

This node is still essentially a more flexible version of the normal Sharpen and can be applied for similar tasks.

Details

EdgeDetect

The EdgeDetect node is great for pulling out and massaging edges. You can control
what is detected, the strength of the edge, and the ability to expand or soften the
edge.

Parameters

strength

Applies a gain to the filter. Lower strength values eliminate detail in favor of the
strongest outlines, whereas higher strength values reveal more details. The slider lets
you choose a value from 0 to 10. Practically speaking, 10 is the highest useful value, but
you can enter higher values into the number field if you need to.

threshold

This parameter lets you further reduce the amount of detail in the resulting image.
Pixels below the threshold turn black. The range is 0 to 1.

binaryOutput

When binaryOutput is turned on, all pixels falling under the threshold parameter are
made black, and all pixels falling above the threshold parameter are made white. The
resulting image is only black and white, with no grayscale values.

directionFilter

Enables an effect similar to Emboss.

directionFilterangle

This parameter changes the lighting angle when the directionFilter parameter is
turned on.

despeckle

Similar to a median filter, this parameter removes isolated pixels up to the despeckle
radius (in pixels), and can be useful for eliminating noise from the resulting image.

xBlur, yBlur

Blurs the resulting image after the edge detection has been performed. By default, the
yBlur parameter is linked to the xBlur.

xDilateErode, yDilateErode

Applies the DilateErode effect to the first filter pass. By default, the yDilateErode
parameter is linked to the xDilateErode.

rgbEdges

Inherits color from the input image instead of just black and white, and applies it to the
final output image.

addEdgesToImage

Mixes the resulting image from the edge detection node to the original input image, to
create a blend of both.

method

A pop-up menu that lets you select which edge detection method to use. Your
choices are:
• Sobel
• Babu
• Fast
Note: Babu is an algorithm that is extremely slow and cannot be used on large (2K or
larger) plates. It is maintained for compatibility purposes.

babuSteps

The steps performed when using the Babu filter. More steps result in a higher quality
image, but are more processor-intensive to render.

bytes

Three buttons let you choose the output bit depth. You can choose among 8-bit, 16-bit,
and float.

Sharpen

Un-sharp masking is used for the sharpening filter. This process blurs the image slightly,
takes the difference between the blurred result and the input image, and adds that
back over the input image. High values of x and yPixels (for example, greater than 3
percent of the image size) return undesirable results. You can also use the sharpen filter
in the Convolve node, but ringing may result.

Parameters

This node displays the following controls in the Parameters tab:

percent

The amount of mixing between the sharpened and the original image. With a value of
100, none of the original image is present.

xPixels, yPixels

The amount of sharpening as described in pixels.

channels

The channels of the input image to sharpen.