< Previous | Contents | Manuals Home | Boris FX | Next >
Delivering Distorted Imagery
In this workflow option, you create and track undistorted imagery, generate CG effects, re-distort the effects, then composite the distorted version back to the original imagery. The process for delivering distorted imagery with the two-pass workflow is described here, but if you are working only with Fusion or Nuke, you may want to consult the following section on the Zero-pass workflow as well.
Here's how to use the two-pass workflow:
- Determine lens distortion via calibration, checklines, or solving with Calculate Distortion turned on.
- Save then Save As to create a new version of the file. (recommended).
- Click “Lens Workflow” on the Summary or Solver panel.
- In most cases, ensure that the "Use input's bit depth" checkbox is turned on: if the input is 16 bit or floating point, you will likely want the same bit
depth on output. If you have specifically set the bit depths on the shot setup panel, turn this off. (If the bit depth changes, the RAM cache will flush and reload automatically.)
- Select whether you want a margin value in pixels, or percentage. A margin value in pixels will result in an image that exactly contains the undistorted image, with that number of pixels around the edge. The image resolution will vary each time you run the script, based on the computed distortion values. Alternatively, select a percentage margin, in which case the image will always be that much larger, regardless of the computed distortion. This approach lets you keep working with a fixed "round" resolution throughout your toolchain, including for different shots that may have somewhat different resolutions.
- Classic workflow : Some legacy codecs, such as Cineform, may require that the image width be a multiple of 16 pixels, otherwise they will crash SynthEyes. To accommodate them, turn on the "Width=16x" checkbox to force the width to be a multiple of 16.
- Advanced lens workflow : the advanced workflow works to maintain exactly the same aspect ratio as the original shot, to avoid introducing tiny subtle mismatches. If your imagery has an irregular aspect ratio (ie reducing the width/height fraction to a “proper” fraction still results in large numbers), unexpectedly large padded sizes may result. If so, slowly increase the Maximum rounding error from its default 0.001 value to see how little introduced error will be sufficient to produce a reasonable size.
- Select the “Final output” option “Redistorted(2)” and hit OK.
- (If needed) Click “Save Sequence” on the Summary panel or Output tab of the image preprocessor (which lets you change resolution if you need to). Write out a new version of the undistorted imagery.
- Important : Save then Save As to create a new version of the file, call it
Undistort for this discussion.
- On the edit menu, select Shot/Change Shot Images.
- Select the “Switch to saved footage” option, hit OK.
- You will now be set up to work with the undistorted (fixed) footage. If you tracked and solved initially to determine the distortion (or without realizing it was there), the trackers and solve has been updated to compensate for the modified footage.
- Track, solve, export, work in your 3-D app, etc, using the undistorted imagery.
- Render 3-D effects from your 3-D app (which match the undistorted imagery, not the original distorted images). You should render against black with an alpha channel, not against the undistorted images.
- Re-open the Undistort file you saved earlier in SynthEyes.
- Do a Shot/Change Shot Images, select “Re-distort CGI” mode on the panel that pops up; select the rendered shot as the new footage to change to. (This updates a number of resolution and aspect settings to facilitate applying rather than removing distortion.)
- Use Save Sequence on the Summary panel or Image preprocessor’s output tab to render a re-distorted version of the CG effect.
- Composite the re-distorted imagery with the original imagery.
Tip : You may be able to perform the image undistortion and redistortion in your compositing application, to avoid round-tripping the images through SynthEyes, if the exporter to your application has the necessary support, or if you use Lens Distortion maps or static or animated projection screens. Such exporters include After Effects, Fusion, and Nuke.
Tip : Exporters that build an undistortion/3D pipeline and a redistortion pipeline in a compositing app (those big 3) will generally put an undistorted version of the imagery at the back of the undistortion pipeline, as a reference when lining up the 3-D scene. Be sure to turn off that lineup backdrop before doing final composite renders from the redistortion pipeline. Otherwise, you will be compositing not over the original imagery as intended, but over an undistorted then redistorted version of the imagery, with its attendant blurring due to repeated resampling.
The two-pass workflow is more complex than delivering undistorted images, but it is a consequence of the more sophisticated end product desired, and it is usually the preference of the end customer for feature film and television projects, instead of the single pass workflow more typical for drone shots and architectural work.
©2024 Boris FX, Inc. — UNOFFICIAL — Converted from original PDF.