What is OctaneRender®?
OctaneRender is a physical-based render engine developed by OTOY. It’s available as a standalone product and as a plug-in for DCC (Digital Content Creation) host apps like Maya, 3D Studio MAX,, and Cinema 4D.
In this guide, we’re going to focus on Octane as a plugin for Cinema 4D. This guide was written using Octane 2020.2.3, but it's basic enough that it should work fine with pretty much any subscription version to come out in recent years.
Within C4D, Octane provides a real-time render viewport window (called the Live Viewer), a full material system that can be edited in the Material Editor (like you’re used to) or in the Node Editor (which you should definitely learn how to do, but isn’t covered in this guide), tags that augment C4D’s native camera and lights, and then a few other objects that offload processor-intensive tasks directly to the GPU.
How do I download and install OctaneRender for Cinema 4D?
Go to otoy.com and click the downloads link in the red navbar at the top. From there it will have you sign in to your account if you aren’t already, and then take you to the downloads page.
Depending on which version of Octane you have (Enterprise or Studio), the Select Software list will populate with a number of options. Choose either OctaneRender Enterprise for Cinema 4D or OctaneStudio for Cinema 4D.
You don’t need to download Octane Standalone unless you want to play with that separately.
You can use either the Operating System dropdown to filter between the Mac and Windows versions or you can just look at the list and click the Download button next to the latest version for your operating system. macOS (Metal) is for newer Macs (Catalina or Big Sur, with supported AMD cards, and soon (as of this writing) for the M1 chips. macOS (CUDA) is for older Macs that still had NVIDIA cards and older versions of the OS.
Once you check Accept and hit Download, you’ll get a ZIP file (Mac) or a RAR file (Windows). You’ll need some sort of unarchiving utility like 7zip to decompress the RAR file.
On the Mac, you’ll have a folder that says the current version. On Windows, you’ll have a folder that says c4doctane. In both of those, there are several files.
This is important
You need to delete the .xlib files (Mac) or .xdl64 files (Windows) for the versions of the software you are not using. Say you have C4D R23. You will need to delete the R15, R16, R17...cdl and xdl (or xlib on Mac) files for everything except c4dOctane-R23.xld64 or c4doctane-R23.xlib prior to installing.
If you have several versions of C4D, you’ll need to make several copies of the folder and only leave in the corresponding version for each plugin folder. Probably best to do this process one version at a time to avoid confusion.
Next up, open C4D, go to Edit>Preferences, and in the bottom left, click the Open Preferences Folder... button to reveal the preferences location in Windows Explorer or Mac’s Finder.
find the plugins folder, and move the c4doctane (Windows) or C4D-OctaneX-(blah blah version number for Mac) into the plugins folder, then restart Cinema 4D.
If all goes well, you should be prompted to log in when you launch C4D again, and you’ll see an Octane menu at the top of the interface.
Part I: The Absolute Basics
If you installed Octane correctly, you should now see an Octane menu at the top of C4D toward the right hand side. Drop that down and choose “Live Viewer” - this is where pretty much everything is in Octane.
Go ahead and dock that to the right of the viewport.
The Live Viewer works similarly to the other viewports in C4D. You can navigate around in it using alt-left click/middle click/right-click, drop materials on items directly in it, and do some more advanced things like pick focus.
There’s a menu bar at the top which is where all of Octane’s objects and materials and stuff live, a button bar below that which you use to start/stop the render and help with visualization, the main window where your scene shows up and starts rendering, and some information overlays.
Enough chatter, let’s jump in. Get open the starter file, or just drop in a standard C4D Plane, Cube, Sphere, Tube, and Platonic (set to Bucky) and then arrange them in a pleasing way. If you’re doing this from scratch, make sure the primitives are all sitting on the plane (make the P.Y value of the sphere the same as the radius, or half the height of the cube and tube in the Coord tab of the Attributes manager). The starter file also uses a resolution of 1000x700 - you set that in C4D’s render settings the same way you do with any engine.
As with most engines striving for realism, Octane works best when objects are at real-world scale. An average person should be somewhere in the 5-6’ height range (150-180cm or so). A billiard ball should be about 2” in diameter (~6cm). The C4D default objects are relatively large - the cube is 200cm to a side, or about 6’5, which is fine for the start of a piece of furniture, but definitely shrink it down if you want to make a sugar cube out of it. The light bounces, camera interaction and texture sizes will just look more realistic this way. It’s also a no brainer when you’re trying to scale objects to one another - you can just Google “billiard ball diameter” and then “billiard table size” and enter those values rather than wing it and hope you’re right.
Our scene is abstract, so the C4D default sized primitives are fine.
Ready for your first Octane render? Hit the Send Scene button (looks like an Octane logo) toward the top of the Live Viewer. This packages your scene up and sends it to the GPU and starts rendering. The green progress bar at the bottom tells you how long it will be until the render is complete.
Immediately this looks a lot better than a standard render, and on most GPUs, it’s really, really fast. If you start moving around in the scene, you’ll see that the viewport constantly updates to match. The button two over from Send Scene is Pause. It won’t update the scene again until you unpause it.
Now that your render is done, you can go to the File menu in the Live Viewer and choose “Save image as...” and save it as a PNG or something.
Congratulations - you’ve now rendered something in Octane. Ready to make it look better?
Part II: Making it Look Better
Easy Lighting and Environment
Octane comes with a default white void environment. To make the scene more interesting, we can put in some image-based lighting (HDRI) which also doubles back as an environment to give our reflective surfaces something to reflect. Get the Adams Place Bridge HDRI here (download both the 4k and 8k versions). Just like in C4D, these go in a /tex folder in the same place your scene file is (you did save your scene, right?).
Let’s drop in an Octane environment. You get this by going into the Objects Menu in the Live Viewer and choosing Hdri Environment. This creates a standard C4D Sky object with an Octane Environment tag. You can ignore all the settings in the Sky object itself - the tag takes over the whole thing. Go ahead and click the tag (the little blue half-circle thing) and let’s see what we’ve got.
First thing we need to do is link to the image we downloaded. This is important - the HDRI goes inside of the Image Texture node that comes preloaded in the environment. It doesn’t replace it. So let’s click the ImageTexture field to get inside of that and then replace that with the HDRI. We’re going to use the 4k version just while we work to keep things fast, and then if we want to keep this as our final rendered background we can switch to the 8k.
Now that we have an environment in, we can go back one level to the environment tag, and use the RotX slider to spin it around. A value of -0.31 seems about right so we can put our objects on the floor. Since this is an HDRI, the lighting of the scene and the shadows will change.
Time to add some materials. All of the various materials are located in the materials menu under the Create submenu. This can be a little daunting at first, but it can be broken down into four easy categories.
The Diffuse, Glossy, Specular and Metallic materials are all old, legacy material types. Many of the materials created over the years available online use these, so they need to stay in, but you can pretty much ignore them if you’re just starting out.
The Universal Material is the best material type ever, and you can create everything in it that you could in all of the previous four put together. There’s a whole guide just on this one material, and it’s mainly what we’re going to focus on in this guide.
The Mix, Composite and Layered materials are for creating complex materials that are difficult or impossible to make in any other material type (yes, even the Universal, though it does try). There’s a guide and a stepthrough about these too.
Hair, Toon and Portal materials are for special use cases. The names should give you some clues, but these are all way out of the scope of this guide.
The first thing we’re going to do is make our objects look like they’re part of the scene. Right now we’ve got this environment with a plane on the ground. If we hide the plane for a moment, we’ll notice that all our shadows are gone, and our objects look like they’re floating in space and aren’t really integrated into the scene very well. There’s a special property in any Octane Material called Shadow Catcher. This hides the object from view, but still shows transparent shadows from the other objects - it’s great for something like this.
Go ahead and make a new Universal Material. (Materials menu in the Live Viewer > Create > Universal Material). By default, the Universal material looks like chrome. If you apply that to the plane, it looks like a mirror.
The Shadow Catcher property overrides everything in the material, so we don’t need to worry about altering it. Let’s double-click the material in the Material Manager. This should look somewhat familiar if you’re used C4D before. It’s the same Material Editor window you’d use to alter a standard C4D material, but the channel list on the left is quite a bit different.
What we’re interested in is the Shadow Catcher checkbox which is in the Common channel, so let’s click that, and then hit the checkbox for Shadow Catcher. We can immediately see in the Live Viewer what happens - The chrome goes away and the shadows are left behind, and they fall on the boundaries of the plane (which is at world zero, so it lines up with where the “floor” of the HDRI is). One thing to be aware of here - the plane needs to be large enough to accommodate the shadows, or they cut off. If we make the plane 200x200cm, we can see this clearly. Let’s make it 800x800cm just to be safe.
Let’s name this material Shadow Catcher - the same way you’d name a normal material.
Before we continue, now that the plane is transparent, the scene looks a little dim. We can change the power of the HDRI either in the Environment tag itself, or if we drill down, we can change the power of the image texture node. It’s usually a better idea to change it in the environment itself since it’s easier to remember where that is if you need to adjust it again, so let’s click our environment tag and change the Power to 2.
Let’s get another Universal Material in the scene - same process as before - Live Viewer Menu>Materials>Create>Universal Material and apply it to the Platonic in the Perspective Viewport.
Channel Stacking and Blending
Now let’s double-click this new universal material and get it open. You should recognize some of the channel names on the left from regular C4D materials, and some of them are going to be new. The first thing to know about the Universal material is that there’s an order to how the channels are applied to the object, and some override others. The Metallic channel is something that throws a lot of first-time users. As long as this is set to 1 (default), the material will look metallic. If you want your material to be plastic or wood or glass or anything aside from a metal, you’ll want to reduce this value (usually you’ll either use 1 for a metal or 0 for a non-metal). Let’s reduce this to 0. Now the material is a very light gray and shiny, but not metallic looking.
Changing the Color Picker
The Albedo channel in Octane is similar to the Color channel in C4D. If we click this, we’ll see the Octane Color Picker on the right, which has 0-1 values for R, G, and B. This can be changed to the default C4D color picker by clicking the Settings (gear) icon in the Live Viewer, then going to the Settings tab, then the Other sub-tab, and choosing Cinema4D native from the Color gui type dropdown and hitting Apply. Once you close and re-open the Material Manager, the Albedo channel should now look just like the Color channel in a regular C4D material.
Anyway, let’s change the color to orange like H:30, S:100, V:100. Now it looks like shiny orange plastic. If we bring the Metallic value back up to 1, we’ll see the difference - it looks like a Christmas ornament. Let’s bring that back down to zero again.
What the Float Does
In a lot of these channels, there’s a color, a float value, a texture field, and a mix field. The color does what you’d expect. The texture works the same as in C4D - you load in a texture or shader of some sort and it takes over the color. The mix slider does kind of an opacity mix between the texture and the color. There are much better ways to do this type of mixing, but this is a quick and dirty type.
There’s also usually a float field. Float is a 0-1 slider that’s basically a grayscale with 0 being black (the channel has no contribution, or is “off”) and 1 being white (the channel is completely “on”). Anything in between is a percentage (so if your roughness is at float=0.75, the roughness effect will only be at 75%). Float is always overridden by color, and this is where some of the confusion comes in. If we have our orange color applied, moving the float slider around does nothing. If we set the color to pure black (0,0,0) - it’s the same as turning it off, and then the float slider makes it lighter or darker gray. If you want a channel to have no contribution at all to the material, make sure both the color is black, and the float is at zero.
When we set our color back to H:30, S:100, V:100, it overrides the float value again, and we’re back to our orange material.
Roughness does what you’d expect - it roughens up the surface so it spreads the light across it and the highlights aren’t as sharp. Let’s set the float value to 0.5 to make it pretty rough. A little goes a long way toward realism - almost nothing in the real world is perfectly smooth, so often you’ll want to use values as low as 0.01 just to break the surface up a touch.
Sheen is an interesting one - that puts a reflective coating on the object that depends on the angle that light is hitting it (think satin fabric). This has a similar effect to a reflectance layer with a Fresnel shader in C4D, but this is much easier to set up. If we crank that up to 1, we’ll see that some of the hexes on the edges of the platonic get really shiny, but the ones you’re viewing straight on don’t.
That gets us a nice basic plastic material - let’s move on.
Let’s make another Universal material and apply this one to the tube by grabbing it from the material manager and dropping it on the tube in the Live Viewer Window. Yep, you can do that.
Again, there’s a whole guide that talks about every channel and what it does, so we’re just going to touch on a few interesting things here.
Thin Film layer
As we go through the other channels, we’ll see some familiar options from C4D and some new ones. One new one that’s kind of a draw to Octane users is the Thin film layer. This creates really pretty iridescence which is a pain to get in other engines. To see how easy this is, let’s set the float to 0.27. This isn’t exact science (well, okay, it actually is, but who is going to read all those material science papers?), so you just want to grab the float slider and the film IOR slider and move them around until you see a color combo you like.
Another cool thing Octane can do on a material level is create a round edges effect that doesn’t add geometry to the object. By default, this channel is deactivated, so let’s activate it by clicking the checkbox next to it and then click the Round edges channel and click the Create Round Edges button. The last step is to click the new RoundEdges field to drill down, set the radius to 2 and the mode to Accurate. Voila, our tube has a small bevel on the edges without having to add a lot of extra processing to the scene.
Using an Octane Shader
The last thing we’ll do is show how to drive a channel with an Octane shader. You can actually drive most channels with most C4D shaders as well, but it takes a little doing - the native Octane ones are typically faster and more stable. Octane’s shaders are located in the c4doctane menu in the dropdown you’d normally choose to get a shader.
Let’s go back to the Roughness channel and under the Texture dropdown, choose c4doctane>Checks. This overrides the 0.5 value we gave it earlier. This is a good place to see what the Mix slider does. As we drag that to the left, we’ll see the rough tiles get smoother and the smooth tiles get rougher until the Mix gets all the way down to 0 where the checks shader has no effect. Let’s bring that back up to 0.9. When you get into more complex materials, you’ll want to do the mixing in the node editor using more advanced methods, but this works for now.
There are a few ways to scale textures in Octane, but the easiest way to scale the entire material is to click the texture tag on the object (like you would with the normal C4D material) and then change the number of tiles. Let’s do 8 TilesU and 4 TilesV. Let’s also do 10% on Offset V. You can also change the projection here and then use the Texture mode to position/scale/rotate the texture. That’s probably good for now - let’s move on to another way to add materials.
Wood Material - Using the LiveDB for Pre-Built Materials
Next thing we’re going to do is have a look at the LiveDB. This is the first entry under the Materials menu in the Live Viewer. Let’s get this open. The LiveDB has a collection of materials available for you to use, as well as an area called LocalDB that lets you store your own.
First off, SAVE YOUR SCENE. The LiveDB window recommends it, and it’s always good practice.
Now, Expand the tree on the left until you see LiveDB>OTOY>Absolute Textures>Wood. Depending on your internet connection, this might take a little bit, but it’ll eventually populate with a number of wood textures. Select the Bamboo Detail material, and then hit the Download button. A new material should appear in the Material Manager. Close the LiveDB window and apply that material to the cube.
This downloads the textures associated with the material and puts them in your /tex folder, so if you’re trying a bunch out, be sure to delete the ones you don’t need after you settle on one.
The absolute fastest and easiest way to make glass in Octane is to create a new Specular material and drop it on an object. But really, what’s the fun in that?
Setting Up a Transparent Universal Material
The best way to do it is to make a new Universal Material and adjust the properties. Let’s make one more Universal Material and apply it to the sphere in whichever way you like best, then get open the Material Editor window for it.
The channel we’re after here is Transmission, so let’s go there and bring the float to 1. We’ll see.. well... nothing changes. The issue here is that Transmission is pretty far down the stack order, so we need to get rid of a few things above it before we can see this effect.
The first thing that needs to go is Metallic. This is toward the top of the channel stack and it overrides Transmission. If we drag the metallic float down to 0, we’ll see that the material turns to a plasticky white. Still not what we’re after.
The next thing that has to go is Albedo. Now by default, the Albedo’s float is already at zero, but as we learned earlier, the Color value in the channel will override the Float. The solution here is to make the Color value pure black (0,0,0). You might think that this would give us a black glass, but the Albedo is actually overriding the transmission, not affecting the color of it. What we’re really doing by setting this to black is removing all of the contribution that the Albedo channel has to the material. Once Albedo is removed from the equation, we now have what looks like glass. If you want a black glass, change the color or float of the Transmission channel. If you want a cloudy glass (or milky liquid), that’s also done in the Transmission channel. Albedo really should only be on for matte or plastic type materials.
Octane is a physical-based render engine, meaning it simulates real physics of materials, so if we want to make a realistic glass, we’ll need to do a little research. Through the magic of The Internet, we can find out that the real IOR (Index of Refraction) for Borosilicate Glass is 1.5046. Let’s head over to the IOR Channel and set it to that. We’re after the Dielectric (non-metallic) IOR here. Metallic IOR works with the Metallic channel. The higher the IOR, the more light bends as it passes through the material. Most real-world materials are in the 1 (air) to 2.something (gems and stuff) range. You can crank it as high as 8 for unrealistic but interesting looks.
One more thing to know about glass is that the wall thickness affects the look a lot. Right now our glass ball is a solid hunk of glass. This may or may not give you the look you’re after. To quickly see the difference wall thickness makes, let’s head up to C4D’s Create menu, go to Generators and choose Cloth Surface. Let’s nest the sphere under that, and then click it to see the attributes. We don’t need it to add any more subdivisions, so let’s set that to zero. What we do want is thickness, so let’s make that 2cm. Immediately we can see a big difference in how the light travels through our sphere. This is very useful to know if you’re going to be making bottles or wine glasses. Always try to find the real-world wall thickness of the object you’re trying to make.For now, let’s un-nest the sphere and delete the Cloth Surface object so the sphere is back to being solid.
Finally, all real-world refractive materials have some sort of dispersion to them. Dispersion splits light up into different wavelengths as it passes through a transmissive material, causing different colors to appear inside. This is a processor-intensive effect, and as a result you don’t usually see it in an engine like Physical or Standard. The GPU is good at this type of calculation, so we can usually get away with adding it without too much impact on the render times (unless you have a ton of glass in complex geometry). Let’s head to the Dispersion Channel and change the value to 0.00420. If you really want to amplify this effect, you can set it to 0.1, but just know that most real materials rarely get anywhere near that high, and it can cause weird artifacts.
Adding a Camera
The Octane Camera (located in the Live Viewer’s Objects menu) is a C4D camera with an Octane Camera tag. We have a C4D camera in our scene already, so let’s add the tag ourselves by selecting the Camera object we have, then going to the Tags menu in the Object Manager, going to the c4doctane menu and choosing OctaneCameraTag. Voila, our camera is now and Octane Camera.
Next up, you’ll want to hit the little white reticle next to the camera so you’re looking through it. Just as a first step in troubleshooting things like depth of field, always check that you’re looking through the right camera. Even those of us who have been doing this for a while forget that sometimes and it’s frustrating to waste 15 minutes just to realize you didn’t have the right camera active.
Unlike the Octane Environment, there are still relevant settings in the C4D camera object itself that need to be adjusted. The focal length is still changed here, so let’s click the Camera object (not the red camera tag) and change the focal length to 50mm to frame it up nicer, and then let’s just reposition the camera a bit so that the objects look like they’re on the floor.
Depth of field, motion blur, post effects, and a few other settings are located in the OctaneCamera tag. For now, let’s get some Depth of Field in there so we can test this out. Click the red Octane Camera tag and under the Thinlens tab, let’s make sure that both the Physical camera parameters and the Depth of field sections are rolled down so we can see what’s happening.
Depth of Field
So this part trips up new users... By default, the camera starts with the F-stop set to 2.8. In a real camera, we’d see some of our objects out of focus, but everything here is in focus. This is because the Aperture defaults to 0 (which is infinite focus). In order to see the depth of field effect, All you have to do is start moving the F-stop slider or the Aperture slider, and depth of field will kick in. Let’s set the F-stop to 1 - now we can see that the sphere went a bit out of focus and the Aperture moved to 2.5. These two sliders are linked.
The Aperture is what really controls the depth of field effect in Octane. The F-stop slider is like a built-in calculator that helps people coming from the photography world who are used to thinking in those terms. The F-stop slider only covers real-world camera values - f/0.5 is the low end of the slider and is about as low as you can expect to find in a real lens, and only a few lenses go as high as f/80, which is Octane’s top-end of the slider. Most of the time real lenses sit in the f/1.2-f/32 range.
If you want to go beyond this and make lenses that really amplify the depth of field effect in a way that would be impossible or impractical to make in the real world, you can adjust the Aperture slider directly. Let’s try making it 35cm, and now it looks like we’re looking through some crazy tilt-shift lens or something. This overrides the f-stop to a setting you can’t set manually (0.071429).
One final note here, even if the f-stop is brought to 80, Octane is still doing depth of field calculations. In order to turn this off completely, make the Aperture 0.
Ok, back to our scene - let’s set the f-stop to 1.8.
There are a number of ways to set the focus in Octane. By default, the Auto focus checkbox is on in the Depth of Field settings. This works like autofocus in a real camera, and it will pick a subject based on an algorithm and focus on that. If you want more control, there are several ways to manually focus.
We can use the F icon (focus picker) in the icon bar at the top of the viewport to choose what we want to focus on. This will disable Auto focus. Click the sphere, and then the tube again to see this in action. The Focus Picker is an on/off thing, so make sure you turn it back off (click it again) when you’re done selecting focus or you might accidentally change that next time you click in the Live Viewer window.
Whenever you click around in the scene like that, the Focal depth value (which appears after Auto focus is disabled) in the Octane Camera Tag will change as well. You can manually set a value in this field as well, but usually that’s a hassle unless you’re working with exact values.
Octane’s Focal depth field completely overrides the Focal depth value in the C4D camera object itself, so make sure you’re on the Octane tag when you’re looking at this value.
The Focus Object field in the C4D camera (not in the Octane tag) will actually override all of Octane’s focus picking, so if you head over to the C4D Camera object in the Object manager (not the Octane tag), and then drop the sphere into the Focus Object field, that will always set the focus point to the sphere, even if you try to use the Focus picker or set Octane’s Focal depth field.
If you want a fast, easy way to set a focus point, you can create a null, name it “Focus Target” (just so you know what it is and don’t delete it later), and then drop it in the Focus Object field. Then, wherever you move the null, the camera will focus. This makes it easy to pick different focus points during an animation.
There’s a lot more to the camera, and you can see a more in-depth guide here as you get more comfortable.
For now, let’s make sure the Focus Object area is cleared out (little down-arrow button to the right of the field in the C4D camera object), and check the Auto focus button to activate it again in the Octane camera tag.
Adjusting the Settings
By now, our scene is still fairly fast, but is starting to look a bit crunchy, especially in the areas where the dispersion effect is pretty apparent (the rainbows in the glass). Even if you have a very powerful computer, there’s always something you can do to bog it down, so it’s worth learning how to optimize render settings (especially if you’re going to get into animation where every second of render time for every frame counts).
Let’s hit the gear icon in the Live Viewer to bring open Octane’s settings. If we expand this window down, we’ll see there are a quite a few, and that’s just in the Kernels tab where we start. There’s also the Camera Imager, Post, and Settings tabs, and then subtabs in some of those.
Octane has a few different methods of calculating a render called kernels. As of this writing, there are three main ones (Direct Lighting, Pathtracing and PMC), and then a few others that are used for troubleshooting or compositing.
The gray bar at the top of the settings window where it says “Directlighting” is actually a dropdown and you can pick from the other kernels there. You can also pick from the Live Viewer window in the dropdown that says DL.
Octane defaults to the Direct Lighting (DL) kernel, which is what we’ve been using, and will continue to use in this guide. Direct Lighting is the fastest of the three kernels, but is missing a few (processor-intensive) things that makes a scene look more realistic. Pathtracing is probably the best all-arounder, but it’s more complicated than DL. There’s an entire guide on Pathtracing and how to optimize that. PMC is the slowest of the lot, but can look the best in certain situations.
What is a Render?
Whenever you start a render, Octane first loads all your geometry and textures on to your GPU (in the VRAM).
Then it calculates how your objects interact with each other based on things like the materials applied to them (can you see through this object? does it need to distort?). It uses some of the brute force settings like the Specular, Glossy and Diffuse depths to determine how much effort it needs to put into getting an accurate result for each type of interaction.
It also uses tricks, hacks, and capabilities of your particular GPU to speed this process up in settings like Parallel Samples, Adaptive Sampling, Path term power, and the other ones in the lower section of the Kernels window.
Once that’s figured out, it starts to show you visual information by sampling. The first sample is a super rough interpretation of all the calculations, and most of the time it’s unusable. Then it continues to go over the image in passes (creating more samples), and refining the image until it’s told to stop (via the Max Samples setting.)
After that, it applies Post processing to your render. This is as simple as color correction adding bloom and glare, or more complicated like denoising and up-sampling. Once this is complete, you have your finished render.
What we’re after here is an acceptable result. This means wildly different things depending on the use case for the render, how critical you are, and how much time and available resources you have.
Right now, our result is... okay. It renders pretty fast on a GTX 1660ti in a laptop (OctaneBench score of ~120). The image quality is pretty good overall, but it starts to get really grainy in the glass, especially around the rainbow areas where the dispersion effect is the strongest.
Setup for setting the settings up
The first thing we need to know here is how long the render is really going to take. Up until now, we’ve been in sort of a look development mode where Octane just renders whatever pixels the Live Viewer is taking up in your interface. If you undock the window and make it really small, it starts rendering a small cropped area of your scene, and it only takes a couple of seconds. If you make the Live Viewer full screen, it renders only the pixels of another crop, and it might take several minutes depending on your screen’s resolution. How do we see what the final render will look like?
This is where the Lock Resolution button in the Live Viewer comes in. Let’s turn that on now. When this is active, Octane renders all 1080x1080 pixels every time, regardless of how much space the window takes up. This gives a pretty accurate estimate of how long the frame will take to render. Now if we make the Live Viewer window tiny, Octane will still render 1080x1080, and you can grab the image and move it around to see the whole thing. If you make it full screen, you should see the whole image unless you’re on a 1080p display or smaller, in which case it crops the top and bottom a little, but you can still pan around in the image to see the rest. Let’s dock the Live Viewer window again.
Depending on your layout, you may or may not see the whole thing. When we activated the lock icon, two new fields appeared to the right of the HDR/sRGB window (both should say 1).
The field on the left is a resolution multiplier. If you make this 0.5, it will render again at 512x512px. This is faster and easier to see if you don’t have a lot of screen real estate. If you have a very large scene, this helps you work faster and you can see your camera framing easily, but it doesn’t give you a sense of how long the final frame will take because it’s actually rendering half of the number of pixels. If you set the field to 2, it renders at 2048x2048. This is useful if you want to really look at an artifact or problem area, or if you’re wondering what kind of hit your system would take per frame if you wanted to render larger than your current resolution. Let’s set this back to 1.
The field on the right is zoom. If you make this 0.5 (while the other field is 1), Octane will still render all 1080x1080 pixels so you can get a sense of how long the final will take, but will zoom out so you can see the whole thing (similar to how you’d use a magnifying glass in a 2D editing app). If you set the zoom field to 2, Octane will blow the image up to 2048x2048, but now what you’re seeing is similar to when you make an image 200% in a 2D image editing app. This is good for inspecting parts of the final rendered image (rather than increasing resolution). You’ll notice Octane doesn’t re-render when you adjust this setting because it doesn’t have to - it’s not generating new data, it’s just zooming in and out of the image. Let’s set this to 1, or less if your layout is tight and you still want to see the whole thing. It’s usually not a good idea to leave it at more than 1 because you’re not really seeing the final pixels that way.
So now we’ve locked the resolution and have our resolution multiplier at 1, we have a good idea of how long the final render will take. It’s about 14 seconds on the 1660ti. If we were going to do a 2000 frame animation meant for mobile and we weren’t going to really focus on the glass much, this might be acceptable. If it’s a still image, probably not so much.
The Denoiser: Cheating is good...
...if you can get away with it. Brute forcing (throwing time and resources at) a render so it looks good might be the noble approach, but in the end, it’s really the end result that counts. Again, we’re after an acceptable result, and we don’t have all day, so let’s concentrate on getting this out the door.
Octane has an excellent denoiser. Prior to denoisers, you’d have to keep upping the number of max samples (sometimes into the tens of thousands) to produce a relatively noise-free render. This issue was especially bad in difficult calculations like dispersion, scattering, and caustics (we don’t have caustics in Direct Lighting, but very much do in the Pathtracing and PMC kernels).
Let’s turn on the denoiser and see what happens. This is in the Octane Settings (gear icon in the Live Viewer), In the Camera Imager tab, in the Denoiser sub-tab. Go ahead and click “Enable denoising” and close the settings window. Octane will re-render the scene.... and.... nothing’s really different except the render time went up a bit (a lot if you’re on a 1660ti, a tiny bit if you’re using a 3080).
If we look closely at the Live Viewer on the bottom, we’ll see that two new little boxes appear - Main and Denoise. These represent the passes that Octane did on this render. First it did the normal render (Main), then it did a Denoise pass (DeMain). We can actually save these passes as separate files which is great in compositing, which we’ll see the benefits of quickly here. Let’s hit the DeMain square.
We’ll notice a few things happen. First off, the dispersion in the glass looks SO much better. This would be totally acceptable.... if not for the fact that most of our shadows disappeared. This is one of the limitations of the denoiser (as of this writing). It doesn’t play nicely with the Shadow Catcher material that we set up on the ground plane. There are a few other quirks to the denoiser (like denoisers, in general, don’t work well on very highly detailed tight patterns), but for many scenes, it’s great.
So we have a few options:
- We could just fix it in post: Octane has a ton of compositing options that we don’t have enough time to get into in this guide, but in this case, it would involve rendering a couple of passes and masking out the areas we like from the passes we like.
- We could ditch the shadow catcher and try to recreate the floor as a material and somehow blend it into the background, or just use the HDRI for reflections only and set up real objects in the scene to act as a background.
- We could see if we can do without the denoiser. Depending on your system and scene, this might actually be faster than messing around with passes and compositing. Let’s see what this entails.
Brute forcing with Max Samples
Sometimes if you encounter a limitation with cheating and optimizing, it actually takes less time to just throw computing power at it. The brutest (?) force way you can do this is just up the max samples value. Just for reference, on the 1660ti, this scene took 12 seconds. With the denoiser added in, it took 27 seconds.
Octane can do an A/B comparison at this point by going to the Compare menu and choosing Store Render Buffer. Then you get a slider that you can move to compare different settings. There’s also a “render region” tool that lets you render just a small part of the image (like the crunchy rainbow area of ours) to speed this process up a bit. We’re not going to worry about these for now since our scene is simple, but just know those can speed up your workflow.
Let’s open the Octane settings, disable the denoiser (in the Camera Imager tab under Denoiser), and then go back to the Kernels section and set the max samples to 1024. Now the main rainbow dispersion area looks pretty good, but the area in the upper right of the glass ball is still a little crunchy. This also upped the render time to 1 minute, 40 seconds on the 1660ti. If we double the samples (protip, just add a *2 to the end of your number and C4D will do the math for you) to 2048, the main rainbow looks great and the crunchy area looks a bit better, and it brings the time to 3:14. If we double it again to 4096, it looks even better, but now the time is 6:40. If we double it again to 8192, it’s probably fine for pretty much any application, but it took about 13 and a half minutes to get there. Again, we’re looking for an acceptable result, and that’s completely subjective, so it’s up to you where to stop.
Triage and Analysis
Denoising is the most extreme post-production/cheating way to make the image better. Upping Max samples is the most time-consuming, but “best” way to add more usable data to the image. Between those, there are a host of other settings you can change to make your render more efficient based on the content of your scene. So at this point, you have to play the triage game and figure out what will give you the best result in the time you’re willing to spend waiting for it.
First off, is this a still or an animation? If it’s a still, 13 and a half minutes isn’t that bad really - it may have even taken you longer than that to do the compositing work in post to fix the denoised version. Also, did it actually take 13 minutes on your computer, or are you rocking a 4xRTX3090 rig where it only took a few seconds? If that’s the case, why bother optimizing, right? Or is an animation, where the difference between 11 seconds and 14 seconds could mean an extra hour of overall render time, and cranking the samples to 8192 is just way out of the realm of reality? If that’s the case, is there anything else you can do to optimize to get it to a place where you’re happy enough with it? Ultimately, what is considered acceptable is entirely your call. These are all things to consider on every render, and why there isn’t just one big red “figure it out for me” button.
Bloom and Glare
You can’t really talk about Octane without talking about Bloom and Glare. These settings make reflections sparkle and shine. These are post effects and are found in the Post tab of the Octane settings. They can be overridden on the Camera level as well, in case you have multiple cameras in the scene and want to control the amount of bloom/glare for each one. These overrides are in the Post Processing tab of the Octane Camera tab. Let’s set them in the Camera tag for now so you can see how it’s done.
If you crank the Bloom up to 300, the hottest light sources (the white reflection on the sphere in this case) will start to get some glow around it. You can adjust the cutoff so the whole piece doesn’t bloom out and look like you have vaseline on the lens.
If you up the glare, you start seeing streaks come off the hot spots. Let’s try 67 for this piece. This is all a matter of personal preference and playing with the settings until you’re happy.
Saving the image
So now we have an image we’re happy with. We can directly save this from the Live Viewer to our hard drive by going to the Live Viewer window, to the File menu, and choosing Save image as.
Or... we can also send the image directly to C4D’s Picture Viewer by right-clicking in the image area somewhere in the Live Viewer and choosing “Show in Picture Viewer”.
Or... we can hit C4D’s Render Settings button (clapboard with a gear in it), select the Renderer dropdown, choose Octane Renderer, and then set up all the normal C4D output settings from there like save location, image size, etc. Then we can close that and hit C4D’s Render to Picture Viewer button and it will render the way you expect.
In fact, once Octane is set as your renderer in the settings, you can also use Takes or the Render Queue or any other normal C4D way of starting a render.
This is really just the tip of the iceberg - there’s so much more to learn. There is a growing number guides here (HTML) and also here (Behance), with more being added all the time. There’s also a manual if you want to learn more about each feature, and a forum to ask specific questions. Happy rendering!