It’s quite common for a client to request the delivery of all project files for a job upon completion of a contract. It seems like a pretty simple request on the surface, but the legalities of it get complicated quite rapidly, especially when the goal of a project is to quickly create high quality content at an economic price.
Project costs can get expensive. Consider for a moment that recent Pixar films have cost almost $2,000,000 for each minute of finished content, and recent Dreamworks films have cost $1,350,000 for each minute of completed product.
Needless to say, most private companies looking for 3D visualization cannot provide such budgets, but still need high quality results. In order to accommodate the 99% reduction in cost, shortcuts must be taken in order to save production time.
One of the best methods of saving time and cutting costs is by purchasing pre-made assets. Instead of creating custom 3D models and textures for each client, a contractor might invest in a pack of pre-made 3D models that they can then use for multiple projects later on.
For example – an animation project that requires an outdoor setting will probably require the use of 3D trees. A contractor might dedicate quite a bit of time and money to learn and master a specialty program such as SpeedTree ($495 – $4,995) or The Plant Factory ($495 – $1,495) that allows artists to make custom, high quality 3D tree models.
Or, the contractor could go onto a website such as Evermotion and purchase a package of 60 pre-made, high quality tree models for $165. This makes a lot of sense, as it saves the contractor (and in turn their clients) a good amount of money.
While the cost savings of using services such as Evermotion are quite apparent, they do present some legal issues if the client wishes to own the completed project files. Every purchase agreement for a 3D model pack states that while the 3D models may be used to create images and animations that can then be commercially sold to clients, the contractor who bought the 3D models (the license holder) cannot legally distribute the 3D models themselves to anyone else.
This means that if the contractor were to deliver the finished project files to their client at the end of production (project files that included portions of the 3D model pack), both the contractor and the client would be placed in legal trouble, as the contractor would have violated their agreement with the 3D model distributor, and the client would now knowingly be in the possession of illegal property without a license.
One solution to the dilemma might be for the client to spend the money needed to purchase their own license to the 3D models. However, rarely do complex animation projects require the use of only a single model pack. An outdoor scene will require trees, mountains, rocks, bushes, saplings, grass, and underbrush; indoor scenes could require hundreds of different objects depending on which room in the house we need to see.
Each of these models also requires accompanying materials and textures which are applied to the objects so that wood looks like wood and stone looks like stone. Many contractors have extensive libraries of calibrated materials and texture files which enable them to create realistic metal, wood, plastic, fabric, and just about everything else you can think of. Many of these textures have some form of rights protection (as they were pre-purchased), and while the final images that they help produce can be sold and given to clients, the raw files themselves cannot.
In addition, contractors frequently make use of plug-ins – or special, custom toolsets developed to dramatically speed up production time (i,e – reduce the price of the project) or to accomplish specific needs – such as simulating the movement of liquids and gasses, or even allowing for incredibly realistic rendered images. While these specialized toolsets can be used to generate images and video files that can then be sold to a client, the source code of the toolset cannot be distributed. Quite often, very important aspects of the project files are tied to the operation of these plug-ins, to the degree that the project file is rendered useless if the plug-in is not available.
Furthermore – most contractors consider the contents of their scene files to be intellectual property, as they have spent years of time developing very technical skills and setups, which means that the scene files are essentially filled with trade secrets. One could compare it to the manufacturers of Pepsi or Coke – who are more than willing to generate and sell you their bottled beverages at a low price; but they never would include their secret recipe on the back of every label.
Lastly – the file sizes can be massive, totalling in the hundreds of gigabytes in some cases, which would require the purchase of a special hard drive so that they could be sent to the client.
To summarize, if a client wants to stay legal, it isn’t always cost effective to require the delivery of the completed project files. Either the contractor will need to spend an excess of billable hours creating each and every project asset from scratch without the aid of any timesaving toolsets, or the client will need to purchase a license for every program, plug-in, texture file, and 3D model that the contractor decides to use if they’d like to make use of the project file in the future.
If you want to stay legal, keep the quality high, and keep costs within your budget range, your first step in doing so it to just leave project files out of the equation.
Using proxies is one of the best ways to speed up your workflow in Cinema 4D. Using simple geometry in the viewport allows you to navigate around your scene and animate your models very quickly. When it’s time to render, all one needs to do is quickly switch out your simple proxy geometry for the detailed, high resolution mesh and you’re off to the races.
Many users have developed fantastic workflows for performing the geometry swap. Over at helloluxx.com, Tim Clapham details how the instance object can be used to easily substitute proxy geo for render geo; and just recently Matt Frodsham wrote a great article on how one can use layers in conjunction with Niklas Rosenstein’s Layer Access Xpresso node to switch between the two.
An average proxy switch.
Both of these techniques make use of a boolean switch to turn the proxies on and off – hit the button to turn the proxies off and the high resolution geometry on, hit the button again to turn the proxies back on and the high res geo off. Very simple.
So simple in fact, that I often forget to hit the button before rendering. I become so used to working with the proxies in the viewport that the button fades to the back of my memory, only to come flying back when I try a test render.
Wouldn’t it be nice if that was done automatically at render time? You work with the proxies in the viewport, hit render, and voila: your high resolution geometry automatically appears in the picture viewer!
Well it turns out our solution is rather simple. The workflow is based around Florian Sepp’s FS Preview Accelerator plugin, which can automatically turn cloners on & off at render time. With a little bit of Xpresso, we can repurpose this to switch out our proxies at render time.
The render time proxy workflow.
The above screenshot demonstrates the workflow; our FS Preview Accelerator tag sits on the cloner object, automatically disabling the cloner while we’re in the viewport and re-enabling it when we render to the picture viewer. Once brought into our Xpresso window, the cloner’s “Enabled” port under Basic Properties becomes our switch: outputting a value of 0 when we’re in the editor, and outputting a value of 1 once we hit render to picture viewer. We wire this up to our “Visible in Editor” & “Visible in Render” ports of our proxy & render geometry, and we’re good to go.
Download the Render Proxy Demo Scene – Requires the FS_Preview_Accelerator Plugin.
To get FS_Preview_Accelerator, go to Florian Sepp’s website, click on “Downloads” on the top bar, and scroll down until you see the “FS_Preview_Accelerator” plug-in.
You could combine this workflow with the Layer Access Xpresso node to have a “Proxy” layer and a “Render” Layer – then all you’d have to do is add your proxy geo to the proxy layer and the render geo to the render layer, and it would automatically turn them on & off at render time.
This basic technique has quite a bit of potential in other areas of C4D as well – I’d love to hear what you come up with!
Thanks for reading,
The Global and Local Matrixes within Cinema 4D can be both lifesavers and death eaters at the same time. A recent project forced me to learn more about them, and I thought I’d pass along some of that knowledge to you.
The first thing I discovered was that you can can convert a Normal Vector into a matrix by use of the Vector2Matrix xpresso node found within the xpresso nodelist under New Node –> Xpresso –> Calculate. Robert Leger has a fantastic guide on the node, which you can find here.
But a problem quickly arose: when I set this new matrix as the Global Matrix of an object, the object locked itself to the origin of my scene (position 0,0,0).
After spending many hours consulting the C4D Python help guide and experimenting on my own, I figured out both the cause of the problem and a method of fixing it.
In Cinema 4D, a matrix is composed of three vectors named V1, V2, and V3. When used as a Global or Local Matrix, these three vectors contain the rotation and scale values associated with the object. In addition, every Global and Local Matrix contains a fourth Vector called ‘Offset’ (abbreviated as ‘off’); this Vector contains the position values of the object.
Therefore, if one converts a normal vector to a matrix, and wishes to use said matrix as the global matrix, the Offset Vector needs to be changed.
To accomplish this task, I created the Matrix Splice xpresso node.
The node is very easy to use: first, feed your current matrix to the ‘Matrix’ input of the node, then feed the Global Position value you’d like to splice into the matrix into the ‘Position’ input. Your new and improved Matrix can be accessed from the “New Matrix” output tab.
Matrix Splice Python Node
To install on OS X, drop the XMA file into the following location: User –> Library –> Preferences –> MAXON –> Cinema 4D [Your Version] –> Library –> xgroup.
Once it’s placed in that folder, you can access the node within your Xpresso Editor window by clicking on File (the ‘File’ button in your Xpresso window, not the ‘File” button located at the top left of your monitor) –> Load X-group.
If you have any trouble with it, let me know.
It’s a hot topic on film blogs everywhere: 48 frames per second. Hollywood’s current standard of 24 frames per second dates back to the 1920s and 30s, when engineers were trying to figure out how to encode sound within celluloid film.
In recent months, filmmakers such as Sir Peter Jackson, Mr. James Cameron, and Mr. Douglas Trumbull have been proponents for an increased frame rate of 48fps, which would eliminate strobing, flickering, and other artifacts present in most modern day films.
Many argue that these artifacts are fundamental to the experience of cinema as we know it, and that their elimination would result in the disappearance of the undeniable magic found in film.
However, an argument in either direction cannot be made without experiencing several films in both mediums. And as it stands right now, our sample sizes are a bit lopsided.
Sir Jackson aim to remedy that situation, and has captured his latest film – The Hobbit – at 48fps.
The response to that knowledge has been mixed, to say the least. Some like it, many hate it, half don’t even know what it means, and the only people to have actually seen any Hobbit footage at 48fps are those lucky enough to have attended a special screening at CinemaCon.
I thought it time to assist in the matter, and give the average person a chance to see how the final product might look. So as a personal project, I converted The Hobbit’s teaser trailer into 48fps.
While it doesn’t truly reproduce the effect you’ll see on the big screen in December, it helps to demonstrate how The Hobbit will be different from a normal 24fps film.
“Now we see but a poor reflection as in a mirror; then we shall see face to face.”
Since there aren’t any online services currently available for streaming high frame rate footage, you will need to download the video clip in order to watch it.
Due to the sheer quantity of data stored in each second of 48fps footage, many computers will struggle to play it smoothly. For this reason, I have provided three different download links to choose from:
High Quality – for those with fast computers*.
Medium Quality – for those with average speed computers.
Low Quality – for those with slower computers.
DISCLAIMER: This video is an UNOFFICIAL presentation of the trailer in the 48fps format, and contains occasional visual distortions that are a result of the conversion process. This footage does NOT truly demonstrate how the final version of the film will appear; it merely helps to demonstrate how it will be different from a normal 24fps film.
*The High Quality video is presented as an FLV file, so you’ll need to get the popular VLC Player (free for both Mac & Windows) in order to watch it.
My Nuke script used for the conversion.
Reducing the Strain
What did you do upon hearing that The Hobbit would be presented in 48fps? I jumped for joy, as it would relieve a problem I’d dealt with for the last several years: eye strain. Anything with a flicker caused me to go cross-eyed: CRT monitors, fluorescent lights, even the strobing seen in a film. The thought of watching a movie without hurting my eyes was exciting to say the least.
But what would it look like? I had experienced a higher frame rate on the small LCD screen of my Canon 7D, but I’d never shot anything serious in that format, nor had I seen any kind of professionally crafted footage at that speed.
This curiosity intensified in December of 2011 when the first teaser trailer for the Hobbit appeared. I kept wondering, “How was this footage going to look at 48? Better? Worse?”
The screenings in late April polarized the media, and left the rest of us to speculate on things not yet seen.
My interest turned to impatience, and I began to explore the possibility of post-converting the trailer into a higher frame rate. Several methods of accomplishing this task presented themselves, each with with their own distinct strenths and weaknesses.
Just as a stereoscopic post-conversion will never truly match up to a film shot natively in the format with two cameras, the techniques that I’ll outline in the following paragraphs will never recreate the effect gained by shooting natively at that frame rate.
As far I know, the following article is the first public repository of information on this topic. If you know of any others, or have any further knowledge related to this concept, please drop me a line – what I have written here just scratches the surface of what it could become.
Without any further ado, let’s get going.
Those familiar with post-production are no doubt aware of plug-ins such as Twixtor and Kronos that allow users to turn their regular speed video footage into super slow-motion extravaganzas. Essentially, these programs examine each frame of footage you provide and figure out what happens between each pair of adjoining frames. It can then use this data to create entirely new frames that are placed between the originals.
Kronos examines the original frames (black) before creating new frames (grey). The source frames are then discarded so that flickering doesn’t occur in the new footage.
When used at its default settings, Kronos will take a piece of source footage and reduce it’s speed by 50% – a process that doubles the amount of frames present, meaning each second of original footage would have it’s 24 original frames converted to 48 new frames.
Now it’s just a simple matter of repurposing the frames’ metadata so that they increase the frame rate of the footage instead of slowing it down.
To best accomplish conversion via Kronos or Twixtor, the user must first separate their original footage into individual shots, then apply this effect onto each of them one at a time, taking time to tweak the plug-in’s settings for optimal quality.
After the shots have been rendered, they can be recombined into a new sequence, re-synched to the original soundtrack and exported into a rudimentary form of 48 fps.
While the footage may play back at the correct speed, it may not look anywhere near correct. Software packages such as Kronos usually create some form of distortion on clips that contain significant motion.
This can be seen throughout my post-conversion of the trailer: watch closely as the dwarves toss plates over Bilbo’s table, and also as Bilbo hops the Hobbiton fence in the following shot.
When Kronos attempts to process clips with a significant amount of motion, distortions are usually created.
Kronos provides several tools within the plug-in for dealing with many kinds of distortion. Most of these techniques require some time to execute, and may or may not work within the paramaters of the particular shot you’re working on. Like a difficult color key, trial and error may serve to be the only plan of attack.
Some shots have so much motion between frames that nothing can be done within the plug-in itself to achieve a clean result. It’s at this point that a brute force technique must be implemented to generate new frames. Working by hand is incredibly slow, but it produces high quality results on tough shots.
My technique of choice was the use of SplineWarp nodes in Nuke which allow the user to “warp” between two master frames, thus allowing the creation of additional frames in between.
A SplineWarp node at work. Pink dots represent a point’s position on the first master frame, blue dots represent where those points moved to on the second master frame.
Two trees of SplineWarp nodes; each grey rectangle represents a single new frame.
To convert one second of regular footage, 24 new frames need to be generated. On average, it took me 15 minutes to do one frame. That’s 12 hours to convert just two seconds of footage. Needless to say, this technique is best saved as a last resort.
In situations where you don’t have much time (i.e. always), it may be necessary to cut corners by eliminating time consuming steps. If one was attempting to quickly convert a sizable sequence that didn’t contain notable quantities of fast movement or require a significant amount of brute-force conversion, it might be worth while to run the entire sequence through Kronos or Twixtor in one go instead of separating it into individual shots before conversion.
This can be a bit of a gamble, but if your settings are good you can convert a significant amount of footage right off the bat. You would then have to sort through the resulting render and hand pick sections that need more attention.
Minor distortions could be fixed by painting them out frame by frame. This technique is quite forgiving as each frame is seen for only a fraction of a second, and the eye doesn’t have much time to discern minor changes in sharpness or tonality.
Distortions on the title shot were perfect candidates for cleaning via RotoPaint nodes in Nuke as there was plenty of material in the surrounding frames to clone from.
Converting a multi-shot sequence all at once creates a unique form of distortion whenever the footage cuts from one shot to the next. Essentially, the plug-in will attempt to create two new frames in the space between the two shots, resulting in a highly deformed transition.
When sending a multi-shot sequence through Kronos, “transitional distortions” are automatically created by the program in an attempt to create a smooth transition between each shot.
Considering the quantity of individual cuts present in an average three minute sequence (the trailer contained over 75) a method must be created by the VFX house to automate the process of fixing the transitional distortion; otherwise the time saved by converting the entire sequence at once will be greatly reduced.
I eventually settled on a series of Nuke nodes that replaced the distorted frames with a duplicate of the nearest clean frame. This results in the last clean frame of the previous shot and the first clean frame of the following shot appearing on screen for 1/24th of a second instead of 1/48th. A cheat of sorts, but it does work: our eyes are too distracted by the change in image to notice the longer frames.
My grouping of Nuke nodes used to remove the transitional distortion.
Several Python expressions were written to automate the process, driven by a user-submitted frame value that highlighted the last clean frame before a transition. This value is checked against the current frame of the composition to determine the values for the fades & mixes.
Placing the nodes into a group and saving them as a Nuke ToolSet significantly sped up the process, allowing me to clean a single transition in only 10-15 seconds.
When the desired result has been attained and all frames have been rendered, it’s time to re-synch the audio and attain a quality export. Nuke doesn’t have a proper timeline to do the sound synching, and Final Cut 7 doesn’t work well with image sequences (I haven’t used FCX, so I’m not sure of that). Adobe Premiere & Media Encoder CS5.5 don’t work natively with 48fps material (though they do work with 60fps).
After exhausting those options, I settled on performing these tasks in After Effects, which handled the job decently. As there aren’t a whole lot of compression options directly in AE, I opted to export a lossless Quicktime file that I could bring into MPEG Streamclip for the heavy lifting.
I discovered that (as always) it’s very important to test your 48fps exports on different computer systems so that you know whether or not they’ll play smoothly. Highly compressed footage, though smaller in size, can choke up an average computer system. This is why my “Medium Quality” clip is 100 mb larger than the “High Quality” clip.
If the 48fps format is as successful as it’s proponents believe it will be, then it won’t take long for Hollywood to realize that converting older films into 48fps will generate more income for the studios and provide more jobs for the VFX industry.
Now is the time for companies such as The Foundry and The Pixel Farm to develop new technologies that assist in the process, making it a smoother experience for everyone else down the road.
Sir Jackson: You’ve got my full support on this endeavor, and I can’t wait to see The Hobbit in it’s full glory on the big screen. What a glorious day that shall be.