Blackmagic Fusion: Getting Started

Getting Started in Blackmagic Fusion

These articles are mostly based on Fusion 8.2 build 2 running on Windows 7 with an nVidia GPU. There are some slight differences in the interface between Fusion 8 and previous versions. I will do my best to highlight them as they come up throughout this series. For the first couple of lessons I was using Fusion 7, though, so some of the screenshots will look a little different from your interface, regardless of which version you are using.

A note for Mac users: As is typical with most software, the Windows Control key is Mac's Command key. As I am sure you know that already, I won't continue to insult your intelligence by pointing it out every three paragraphs.

This lesson covers importing footage into Fusion, converting it to linear color, viewing and playing it, and exporting it back out in the file format of your choice. Where Fusion differs greatly from Nuke and After Effects, I will make note of it. For more thorough details on the software and its tools, please consult the Manual and the Tool Reference documents, which are probably the most valuable contributions Blackmagic has made to Fusion so far!

The Interface, Briefly

We'll go into more detail about the Interface in Chapter 3. For now, we'll just look at it in very broad strokes.

Fusion 8 Workspace
Fusion 8 Workspace

The above image shows the divisions between the major work areas. The Menus work just like most other software. The Toolbar has a handful of layout buttons and shortcuts for a variety of commonly-used tools. In Fusion 7 and earlier it also has buttons for Load, Save, Undo, Redo, Copy, Cut, Paste, and Delete.

The Viewers are where you will display your footage. They are independent and can show different images and have different zoom levels, but they are time-locked to one another, so they'll always show the same frame.

The Flow View is where the Tools used to build the composite are represented visually in a flow-chart-like representation. The Control Panel displays the settings for Tools that are selected in the Flow View.

The Time Ruler controls the in- and out-points of the composite, the current displayed frame, the working time range, and has the typical playback controls you might already be familiar with in other video software.

Navigating the 2d Interface

To pan around in any panel except the Time Ruler, hold down the middle-mouse button (MMB) and drag in any direction (though the Control Panel will only go up-and-down). If you do not have a middle-mouse button (or yours, like mine, is difficult to use), you can also hold Ctrl+Shift+Left-mouse button (LMB) and drag.

To zoom, hold both LMB+RMB and drag, or use your scroll wheel. Fusion works great with a stylus. I use a Wacom Intuos drawing tablet and have assigned MMB to the second button. Holding that button and hovering over the surface pans, while holding the button and drawing on the surface zooms.

Loading Footage

When you open Fusion, it automatically creates a new blank Composition (comp, for short). If for some reason the default comp does not appear, File > New (Ctrl+N) will create it.

The first step in a typical project is to import some footage or images to work on. In most software, you would use the Import Footage command in the File menu. If you go that route, however, Fusion will create a new composition with the chosen footage as its filename. Then, when you use File > Import to get a second element to combine with the first, Fusion will again create a new comp for it instead of importing it to the existing comp. While you could copy-and-paste the Loader nodes from these comps into your working comp, using the Import method is not the most efficient way to bring footage into your workspace.

Every video or image you work with in Fusion is controlled by a Loader node, which is equivalent to Nuke's Read node. When you use File > Import, the new composition is created with a Loader already in place. If you need to make your own, though, there are four ways to do it. First, at the top of your screen, there is a toolbar with a number of buttons with two- or three-letter codes. The leftmost says "LD." That's the Loader. Click it, and you will get a file browser you can use to find your footage.

Fusion 7 interface. In Fusion 8, everything to the left of the Layout buttons is omitted by default.

You can also drag the button to the Flow View in order to place it exactly where you want it instead of having it appear at a default location.

The second method is to right-click in the Flow View and navigate to Add Tool > I/O > Loader.


The third method is to click in the Flow View to make it the active window and type Ctrl+Space to bring up the Add Tool dialogue, just like hitting Tab in Nuke. Then start typing "loader," and the dialogue will filter the available tools to display the one you want. Highlight Loader in the window and click OK, or simply hit the Enter key. This method is the quickest way to get a tool whose name you already know. The right-click context menu is the easiest way to browse the available tools when you aren't sure what you want, or don't know if it even exists.

The fourth method is to find your footage in your operating system's browser, and simply drag-and-drop into the Flow View. If you're importing a sequence, only drag one file—Fusion will automatically detect that it is part of a sequence. If you really only need one frame, hold down Shift before you drop the file. The loader will be configured to use only that frame (although you can still access the entire sequence using the Trim controls).

Fusion does not have hotkeys for tools by default, but you can customize the hotkeys to call for nodes by going to View > Customize Hotkeys… Nuke users can thus make themselves at home by assigning hotkeys to tools themselves.

Click for a larger image.

Once your Loader is in the Flow View, the rightmost column should show the node's controls. I won't go over all of the controls right now, but here are some important ones you may want to use right away:

The Global In and Global Out controls determine when in your timeline the clip begins and ends. Be careful using this control, though, as Fusion will abort the render if the Saver at the end of the chain cannot get an image at a given frame. You can get around this by using the Hold First Frame control to force Fusion to call for that frame until such time as the clip actually starts.

The Filename field contains a path to the footage or image. Click the folder icon in Fusion 7 or the "Browse" button in Fusion 8 to open a file browser where you can select a clip. Be aware that loading a new clip (or reloading the same clip) will reset the Trim and Global range controls.

Trim In and Trim Out are used to determine the first and last frames of the imported footage. That is, if you have a 3000 frame clip but only need to use frames 88 – 146, set Trim In to 88 and Trim Out to 146. Note that the Global In still controls where in your timeline the clips shows up, so if you have Trim In at 88 and Global In at 0, then frame 88 of your clip will be displayed at frame 0 of your comp. This works like using the "start at" mode for the frame control in Nuke. For After Effects users, you can switch over to the Timeline view to see a graphical representation of the timing of your Loaders. Any changes you make in the Timeline view will automatically update the controls in the control panel. The order in which the nodes appear has nothing to do with layering, though, so don't confuse this view with AE's layer-based system.

getting-started_013The Import tab stands in for After Effects' Interpret Footage dialogue. Here you can change the bit depth, the pixel aspect ratio, and the color and gamma space, among other things. Generally speaking, it's a good idea to leave this tab alone, as it is easy to overlook settings made here, and there is nothing you can do that cannot be done more explicitly with a separate node. Still, it's a good idea to be aware of it in case someone has adjusted something and you are trying to troubleshoot it. I know typical practice in Nuke is to set the color space in the Read node, and while Fusion does permit you to do so, the fact that it's hidden in a secondary tab makes it a good idea to keep the color management outside of the Loader for clarity's sake.

Depending on what kind of footage your loader is handling, the Format tab may or may not have something in it. If you are using Cineon or DPX files, it's a good idea to make sure the Bypass Conversion box is checked. I'll cover this tab in greater detail when we talk about file formats and color management.

Displays and Color Management

Now you have some footage in the flow, and you want to be able to see it. Unlike Nuke, Fusion does not require a Viewer node. By default, there are two Viewers available: The two large, empty boxes above your Flow view. To put your footage into a Viewer, you can drag the node you want to view to the Viewer. Or, if you mouse over the node, two small black dots will appear in its lower-left corner. Clicking in the left-hand dot puts the image in Viewer 1. Clicking in the right-hand dot puts the image in Viewer 2. If you create a new Viewer with Window > New Image View, a third dot will appear. In addition, I am given to believe that if you have a Blackmagic interface card connected to a broadcast monitor, a dot will be available to send the image to that output. I've never had that luxury, unfortunately. Or, like Nuke, you can simply hit 1 or 2 to put the image in the left or right viewer, respectively. Or 3 to put it into Viewer 3, if there is one.

My footage displayed in Viewer 1.

At this point, we should talk a little bit about color management. This is likely to be foreign territory for a lot of After Effects users because AE makes color management much more difficult than it needs to be! It may also be a weak point for some Nuke users because Nuke handles color conversions more automatically than Fusion does, so artists frequently don't even realize that the Read is doing a conversion. I won't get too far into it, as it is a deep topic best saved for a later chapter, but the basics are these: Fusion (for the most part) assumes that the images you are working with have a linear color response: A value of 1 represents twice as much light as a value of 0.5. For various technical reasons, cameras and monitors are not linear devices. In order to view the image the way it was meant to be seen, you need to know two things: What gamma curve is the image using, and what gamma curve does my screen use? Once those two things are known, you can apply some tools to remove the gamma of the image while you work on it, restore it before you save the composite, and apply the right curve to view it on your screen.

Image gamma:

Jpg, tga, png, avi, mov, mp4: If you are using one of these file types, or some similar format intended for viewing on a screen, it is likely that the image is in either sRGB or rec.709 color space. If that is the case, it probably looks just like you would have expected it to right now. Fusion is just passing the pixels from the file to your screen without attempting to interpret them in any fashion, just like AE does in its default configuration. While that's great for looking at it, it's not going to be so good for compositing (we'll explore why later). To turn the sRGB image into a linear one, you need a Gamut tool. Gamut isn't available on the toolbar, so you'll have to use the right-click or ctrl+space method to access it. If you use right-click, you'll find it in the Color folder. If the tool didn't automatically connect to the Loader, just drag the red square of the loader to the gold triangle on the Gamut to connect them, then put the Gamut tool into the Viewer.

Removing the sRGB gamma curve to linearize a jpeg.

In the Gamut's controls, select sRGB in the Source Space drop-down. By default, the Remove Gamma checkbox will appear, already ticked. Leave Output Space set to "No Change." What that does is to tell Fusion that the gamma curve of the input is sRGB, and that it should remove that curve and send the image out with no further change. The image is now linear, and it likely looks terrible in your Viewer—too dark and saturated. We'll fix that in a few moments.

Cin or DPX: If you have footage that came from a scanned film negative or was preprocessed for post-production, it probably uses some kind of logarithmic gamma curve. These images usually look milky—desaturated with lifted blacks. Depending on your preferences, the Loader may have automatically given a dpx a default conversion. Check the Format tab to be sure that the Bypass Conversion box is checked. Best practice is to do the conversion in a separate node in order to make it obvious what's been done. To convert a log plate to linear, use a CineonLog tool. This tool is found in the right-click menu in the Film category. Again, if it did not automatically connect to the Loader, drag the red square of the Loader to the gold triangle of the CineonLog. The default mode converts Log to Lin with a standard Cineon curve (a specification developed by Kodak for scanning film to digital format, Cineon continues to be a useful approximation for many logarithmic capture formats). You can adjust the curve with the controls in this tool, or select from a small number of preset log types provided by standards and camera manufacturers. If you need a more specific color conversion, I will cover LUTs more thoroughly later on.

EXR: EXR is usually assumed to be linear. No further conversion is usually necessary, although it's easily possible to render to EXR in sRGB mode, in which case you'd use the Gamut tool as described above. An sRGB image that has been interpreted as a linear one usually looks too bright and noisy.

Viewer Gamma:

Now that the image has been linearized, it's time to tell Fusion how to display it on your screen. In the image below, you can see that I have converted my dpx source, shot on an ARRI Alexa, to linear using the ARRI LogC curve in a CineonLog tool:


I've circled the LUT button, which is used to tell Fusion what kind of monitor I am viewing the picture on. If you aren't familiar with this concept, LUT stands for LookUp Table, and it's simply a brightness curve that maps incoming pixels to outgoing brightness values. Once again, we'll get deeper into that idea later on. For now, just know that by activating this switch and choosing a color space in the pop-up menu (click the little up-arrow next to the LUT button), Fusion's Viewer will translate the linear pixels back into something we can easily view. In Fusion 7, you can simply choose sRGB and be good to go. Fusion 8, for some reason, removed many of the default LUTs, so you'll have to do just a little bit of configuration to get the Viewer to work correctly.

getting-started_009 From the pop-up menu, choose the Gamut View LUT, then immediately open the pop-up menu again and choose "Edit…" The dialogue that appears works just like the Gamut node—set the Source to "No Change,"  the Output to "sRGB," and ensure that "Add Gamma" is checked. This time, we're telling Fusion that the incoming image is linear, and we want to add a gamma correction to turn it into sRGB.

getting-started_014In order to make sure that these settings stick the next time you open Fusion, you need to set them as default. Right-click in the title bar of the control panel—the blue text that says "GamutViewLUT" and choose "Settings > Save Default." Now the LUT will always use these settings. To make this the default LUT, right-click in the Viewer and again choose "Settings > Save Default." Now every Viewer will use this LUT. (As of Fusion 8, build 2 there is a small bug where both the Fusion View LUT and the chosen LUT will be selected in Viewer 2. Select the Gamut View LUT again to clear this condition.)

Phew! That's a lot of (boring) work just to be sure we're looking at things correctly, but trust me: this is critical knowledge, and getting it right will only help you. Once again, we'll go deeper into color spaces and LUTs in a later chapter.

Playback and the Time Ruler

getting-started_011Below the Flow View is the Time Ruler, playback controls, and a few other buttons. The playback controls are fairly standard for software of this kind: Play forward and back, step forward and back, go to in and out points, and a loop control button. If you right-click on the play buttons, you get a context menu that lets you select the step frame value.

The hotkeys for stepping forward and back on the Time Ruler are the inconveniently-located [ and ] (Someday maybe we'll standardize the playback controls for all video and film software). Holding Alt while pressing these keys will move the playhead to the previous or next keyframe of the currently selected tool(s), if any such keyframes exist.

To set the Time Ruler ranges, the numeric fields at the left and right extremes are the easiest way. In this image, I have marked the Time Ruler Range with hollow white arrows, the Render and Playback Range with solid white arrows, and the current frame with red arrows:


As you can see, the light grey bar represents the Render Range, and the dark grey bar represents the Timeline Range. The playhead is indicated by an even lighter grey line, and any keyframes on the currently selected tool will show up as green lines.


Changing the image from linear back to sRGB.
Changing the image from linear back to sRGB.

Writing the information out once your composite is complete is much the same as loading it to begin with. First, you need to add whatever gamma adjustment your finished image needs. Usually, you want to simply reverse whatever you did to make the image linear. A CineonLog tool in Lin to Log mode for dpx, or a Gamut tool with the Source Space set to No Change and the Output Space set to sRGB or rec709.

Next, you need a tool that will write the image to disk. For this we use the Saver, which is equivalent to Nuke's Write node. On the toolbar, it's the red "SV" button.

There are some quirks in the Saver node. First, if you are rendering to discrete frames (recommended), Fusion will put a four-digit frame number right before the last period in the filename. If you render out foo.jpg, your actual filenames will be foo0001.jpg, foo0002.jpg, and so on. So a filename of foo_.jpg or foo..jpg is better. If you want to specify the frame padding, Fusion will interpret any numbers or # signs right before the last period as the numbering pattern. So foo.000000.jpg, foo.123456.jpg, and foo.######.jpg will all result in six-digit frame numbers. The starting frame number is determined by the Time Ruler's settings, not by any numbers you put in the filename field.

Second, if you change the Output Format in the drop-down box, but do not change the extension that appears in the Filename field, the rendered frames will be in the specified format but have the wrong file extension. On the other hand, if you change the extension in the Filename field, the Output Format box will update automatically.

Third, if you are running Windows and wish to output a Quicktime video file, you will need to have Quicktime Essentials installed. There was a recent security warning about a vulnerability in Quicktime, so it is currently recommended to install only the Essentials portion of the software, not the Player or web plug-ins. Again, though, Fusion doesn't handle video files very well, so an image sequence is recommended. You can convert the sequence to a video format of your choice using software dedicated to that task, such as FFMPEG.

Once your Saver is configured, you can render the comp with the green Render button by the playback controls:


Unlike Nuke, Fusion will always render every Saver that is not passed through (disabled). If you have getting-started_016multiple Savers in your comp and only want to render one, you have to disable all the others. Ctrl-P is the hotkey for disabling a selected node, or you can click this button (it has a different icon in Fusion 7 and earlier, but it's in the same place):

A few closing thoughts

Fusion gets inefficient very quickly when dealing with multiple video sources. Best practice is to convert any video to frames in order to streamline your work. It's also not the best at handling audio, but if you do want to link a separate audio file to your comp, first make certain that your playback rate is set correctly in the Preferences, then right-click the audio button to the right of the playback controls. In the Saver, you can add the audio to your rendered output, assuming you render a video format, in the Audio tab. Again, though, it's usually better to handle audio and video encoding in an editing program such as Premiere or Final Cut.

That about does it for the "ins and outs," if you'll pardon the expression, of Blackmagic's Fusion! In the next chapter we'll talk about some basic compositing theory and tools.

Update, June 9 2017: In response to Ian's question in the comments, I have added a little bit more information about video formats in the Saver.


  1. Hi Bryan!

    Just wanted you to know that I am following your tuts with great interest since I am new to fusion and compositing in all together. Keep up the good work!

    Best regards from Sweden

  2. OK – Clarification please. Im also new to Fusion and Im getting to think its just one headache after another and I should be in the Adobe camp. Ive just created a short text intro based on an online tutorial. Great everything works. Like the result and now want to save it – this is when everything goes seriously wrong.

    Ive just read your page –
    Second, if you change the Output Format in the drop-down box, but do not change the extension that appears in the Filename field, the rendered frames will be in the specified format but have the wrong file extension. On the other hand, if you change the extension in the Filename field, the Output Format box will update automatically.

    NO video file available to save on. I have to save as a series of files.

    You have to be kidding me.



    1. I assume you're using Windows and do not have Quicktime installed? Fusion doesn't have its own Quicktime encoders and relies on an existing Quicktime framework in order to encode video formats. Installing Quicktime should solve your issue. If you are concerned about the reported security vulnerability, you can install only the Quicktime Essentials with the regular QT installer. I forget exactly the steps, but there should be a custom install option where you can disable the Player and web plug-ins.

      Standard professional practice is to work only with image sequences—there are several disadvantages to trying to work with video in a compositing program (yes, even AE)—so Fusion has limited capacity to create it. We're hoping that Fusion 9 will have better video support on Windows and Linux, but so far Blackmagic hasn't shared any details about what's in store for the new version.

      Personally, I recommend using a utility specifically designed for video, such as FFMPEG or Adobe Media Encoder, to encode image sequences to video when required rather than using a compositor. You get much more control over the process, and the files are usually more efficiently compressed. AME compresses h.264 as much as 20% smaller than After Effects when using the same settings.

  3. Thanks Bryan. There seems to be so much confusion on some many areas of the internet and finding a straight forward answer from yourself was excellent.
    I don't suppose in a future post you could give a pros and cons on file formats when saving a project.
    Thanks again.

    1. I've covered a little bit of it with the Anatomy of an Image article. I have been thinking that I should get a little deeper into video formats, though. There are so many video codecs, though, that it's a little daunting, especially since I don't actually know all that much about video, so I need to do quite a bit of research.
      The relevant info is about 2/3 of the way down the page, under the heading "File Formats and Compression."

  4. Is there a way to create a single node which "represents" an entire composition? This would be essentially importing models into a composition, such as a street lamp into a city scene. I know I create a duplicate node to represent subsequent copies of a structure (such as the street lamp), but is it really necessary to copy and paste the entire node structure of the object to begin with? If so, it would clean up the node/flow space tremendously.

    1. You don't usually need to create copies in order to reuse portions of a comp. A node-based compositor uses a one-to-many paradigm—you can pull as many wires off an output as you need, allowing you to reuse a Loader, a Render3D, or an entire portion of the tree multiple times.

      For instance, in your example of copying a street lamp several times, you could make several Transform3D nodes, connect the output of your street lamp geo to each Transform3D's input, then connect all of the Transforms into a Merge3D. Even though there's only one copy of the lamp in your flow, you'll get several in your scene because it's being added to the Merge3D several times with offsets determined by the Transform3D. Or for something as regular as that's likely to be, you could just use the Duplicate3D node, which can distribute the lamps evenly in whatever direction you choose and even add a little bit of jitter to their placement so they're not perfectly uniform.

      You also have the option of prerendering an element if it's too heavy to use multiple times. Then you can replace the input(s) below that Saver with a Loader. Just remember to pass-through (disable) the Saver once the prerender is done so that you don't keep processing that part of the tree when you don't need to.

      There will be some good examples of using multiple outputs from a 3D scene in the next article. I'll also try to remember to put in some information about the processes I described above. Whenever I get around to writing it; my new Oculus Rift is super distracting!

      1. Thank you for your quick reply. One of my frequent frustrations with the Duplicate3d node is that it randomly anchors in the rendering field. What I mean by this is that the grab and drag arrows don't always appear right with the duplicate object, sometimes they're separated by a hefty distance. This becomes problematic when trying to rotate the object, because it is trying to rotate on an axis that is far removed from the center of the duplicate object.

        I'll look forward to your next article to get a clearer picture of some of the other strategies you mentioned.

        Thanks again.

Leave a Reply