Close

Optical Flow TOP

Summary

Optical Flow detects patterns of motion in its input. The motion detected in the X-direction is output in the red (R) channel, while the motion in the Y-direction is output in the green (G) channel. This operator currently only works with Nvidia GPUs.

The pixel values stored in the red and green channels are 32-bit floating point numbers. A value of 1.0 in a pixel of the red channel means that the horizontal motion of what is at that part of the image is moving left-to-right at 1 screen-width per second. In a 16x9 image, a green value of 1.0 at that pixel means the motion is upward at 1 screen-width per second, or 1 * 16/9 = 1.77 times the screen height per second.

The values can be scaled with the Gain parameters. The TOP can drive forces in the particlesGpu component.

PythonIcon.pngopticalflowTOP_Class


Parameters - Optical Flow Page

Grid Size gridsize - - Determines the output resolution. A smaller grid corresponds to a larger output image. Turing GPUs only support `4x4`, whereas newer GPUs can use any option.

  • 1x1 grid1x1 - The smallest grid size, which is slowest to compute. The output resolution is the same as the input resolution.
  • 2x2 grid2x2 - The output resolution is half of the input resolution.
  • 4x4 grid4x4 - The largest grid size, which is the fastest to compute. The output resolution is one quarter of the input resolution.

Quality quality - - Specify the optical flow model. Higher quality is slower to compute, and low quality is faster to compute.

  • High high - High quality optical flow, but slow to compute.
  • Medium medium - Medium quality optical flow, with medium computational speed.
  • Low low - Low quality optical flow, but fast to compute.

Cost Output costoutput - Toggle whether the "cost" of the optical flow is output in the blue channel of the TOP. Higher cost means higher uncertainty in the optical flow estimate in the RG channels. The cost values are integers between 0 and 255.

Gain gain - - A post-multiply for the optical flow in the RG channels, allowing control of output intensity.

  • Gain gainx - This number will be multiplied by the R channel (the optical flow in the x-direction) at the very end.
  • Gain gainy - This number will be multiplied by the G channel (the optical flow in the y-direction) at the very end.

Manual Timing manualtiming - The Optical Flow TOP works in both real-time mode and non-real-time mode. However, if you have a video source whose frame rate differs from the project rate, you might be able to get better optical flow results with Manual Timing enabled. For every input video frame, you would need a timestamp in seconds to drive the `timestamp` parameter on the TOP. Some TOPs related to video streaming have this `timestamp` channel when an Info CHOP is placed on them.

Timestamp timestamp - This is a timestamp in seconds of the video frame. See the discussion of Manual Timing in the parameter above.


Parameters - Common Page

Output Resolution outputresolution - - quickly change the resolution of the TOP's data.

  • Use Input useinput - Uses the input's resolution.
  • Eighth eighth - Multiply the input's resolution by that amount.
  • Quarter quarter - Multiply the input's resolution by that amount.
  • Half half - Multiply the input's resolution by that amount.
  • 2X 2x - Multiply the input's resolution by that amount.
  • 4X 4x - Multiply the input's resolution by that amount.
  • 8X 8x - Multiply the input's resolution by that amount.
  • Fit Resolution fit - Fits the width and height to the resolution given below, while maintaining the aspect ratio.
  • Limit Resolution limit - The width and height are limited to the resolution given below. If one of the dimensions exceeds the given resolution, the width and height will be reduced to fit inside the given limits while maintaining the aspect ratio.
  • Custom Resolution custom - Enables the Resolution parameter below, giving direct control over width and height.

Resolution resolution - - Enabled only when the Resolution parameter is set to Custom Resolution. Some Generators like Constant and Ramp do not use inputs and only use this field to determine their size. The drop down menu on the right provides some commonly used resolutions.

  • W resolutionw -
  • H resolutionh -

Resolution Menu resmenu - A drop-down menu with some commonly used resolutions.

Use Global Res Multiplier resmult - Uses the Global Resolution Multiplier found in Edit>Preferences>TOPs. This multiplies all the TOPs resolutions by the set amount. This is handy when working on computers with different hardware specifications. If a project is designed on a desktop workstation with lots of graphics memory, a user on a laptop with only 64MB VRAM can set the Global Resolution Multiplier to a value of half or quarter so it runs at an acceptable speed. By checking this checkbox on, this TOP is affected by the global multiplier.

Output Aspect outputaspect - - Sets the image aspect ratio allowing any textures to be viewed in any size. Watch for unexpected results when compositing TOPs with different aspect ratios. (You can define images with non-square pixels using xres, yres, aspectx, aspecty where xres/yres != aspectx/aspecty.)

  • Use Input useinput - Uses the input's aspect ratio.
  • Resolution resolution - Uses the aspect of the image's defined resolution (ie 512x256 would be 2:1), whereby each pixel is square.
  • Custom Aspect custom - Lets you explicitly define a custom aspect ratio in the Aspect parameter below.

Aspect aspect - - Use when Output Aspect parameter is set to Custom Aspect.

  • Aspect1 aspect1 -
  • Aspect2 aspect2 -

Aspect Menu armenu - A drop-down menu with some commonly used aspect ratios.

Input Smoothness inputfiltertype - - This controls pixel filtering on the input image of the TOP.

  • Nearest Pixel nearest - Uses nearest pixel or accurate image representation. Images will look jaggy when viewing at any zoom level other than Native Resolution.
  • Interpolate Pixels linear - Uses linear filtering between pixels. This is how you get TOP images in viewers to look good at various zoom levels, especially useful when using any Fill Viewer setting other than Native Resolution.
  • Mipmap Pixels mipmap - Uses mipmap filtering when scaling images. This can be used to reduce artifacts and sparkling in moving/scaling images that have lots of detail.

Fill Viewer fillmode - - Determine how the TOP image is displayed in the viewer.

NOTE:To get an understanding of how TOPs work with images, you will want to set this to Native Resolution as you lay down TOPs when starting out. This will let you see what is actually happening without any automatic viewer resizing.

  • Use Input useinput - Uses the same Fill Viewer settings as it's input.
  • Fill fill - Stretches the image to fit the edges of the viewer.
  • Fit Horizontal width - Stretches image to fit viewer horizontally.
  • Fit Vertical height - Stretches image to fit viewer vertically.
  • Fit Best best - Stretches or squashes image so no part of image is cropped.
  • Fit Outside outside - Stretches or squashes image so image fills viewer while constraining it's proportions. This often leads to part of image getting cropped by viewer.
  • Native Resolution nativeres - Displays the native resolution of the image in the viewer.

Viewer Smoothness filtertype - - This controls pixel filtering in the viewers.

  • Nearest Pixel nearest - Uses nearest pixel or accurate image representation. Images will look jaggy when viewing at any zoom level other than Native Resolution.
  • Interpolate Pixels linear - Uses linear filtering between pixels. Use this to get TOP images in viewers to look good at various zoom levels, especially useful when using any Fill Viewer setting other than Native Resolution.
  • Mipmap Pixels mipmap - Uses mipmap filtering when scaling images. This can be used to reduce artifacts and sparkling in moving/scaling images that have lots of detail.

Passes npasses - Duplicates the operation of the TOP the specified number of times. Making this larger than 1 is essentially the same as taking the output from each pass, and passing it into the first input of the node and repeating the process. Other inputs and parameters remain the same for each pass.

Channel Mask chanmask - Allows you to choose which channels (R, G, B, or A) the TOP will operate on. All channels are selected by default.

Pixel Format format - - Format used to store data for each channel in the image (ie. R, G, B, and A). Refer to Pixel Formats for more information.

  • Use Input useinput - Uses the input's pixel format.
  • 8-bit fixed (RGBA) rgba8fixed - Uses 8-bit integer values for each channel.
  • sRGB 8-bit fixed (RGBA) srgba8fixed - Uses 8-bit integer values for each channel and stores color in sRGB colorspace.
  • 16-bit float (RGBA) rgba16float - Uses 16-bits per color channel, 64-bits per pixel.
  • 32-bit float (RGBA) rgba32float - Uses 32-bits per color channel, 128-bits per pixels.
  • 10-bit RGB, 2-bit Alpha, fixed (RGBA) rgb10a2fixed - Uses 10-bits per color channel and 2-bits for alpha, 32-bits total per pixel.
  • 16-bit fixed (RGBA) rgba16fixed - Uses 16-bits per color channel, 64-bits total per pixel.
  • 11-bit float (RGB), Positive Values Only rgba11float - A RGB floating point format that has 11 bits for the Red and Green channels, and 10-bits for the Blue Channel, 32-bits total per pixel (therefore the same memory usage as 8-bit RGBA). The Alpha channel in this format will always be 1. Values can go above one, but can't be negative. ie. the range is [0, infinite).
  • 16-bit float (RGB) rgb16float -
  • 32-bit float (RGB) rgb32float -
  • 8-bit fixed (Mono) mono8fixed - Single channel, where RGB will all have the same value, and Alpha will be 1.0. 8-bits per pixel.
  • 16-bit fixed (Mono) mono16fixed - Single channel, where RGB will all have the same value, and Alpha will be 1.0. 16-bits per pixel.
  • 16-bit float (Mono) mono16float - Single channel, where RGB will all have the same value, and Alpha will be 1.0. 16-bits per pixel.
  • 32-bit float (Mono) mono32float - Single channel, where RGB will all have the same value, and Alpha will be 1.0. 32-bits per pixel.
  • 8-bit fixed (RG) rg8fixed - A 2 channel format, R and G have values, while B is 0 always and Alpha is 1.0. 8-bits per channel, 16-bits total per pixel.
  • 16-bit fixed (RG) rg16fixed - A 2 channel format, R and G have values, while B is 0 always and Alpha is 1.0. 16-bits per channel, 32-bits total per pixel.
  • 16-bit float (RG) rg16float - A 2 channel format, R and G have values, while B is 0 always and Alpha is 1.0. 16-bits per channel, 32-bits total per pixel.
  • 32-bit float (RG) rg32float - A 2 channel format, R and G have values, while B is 0 always and Alpha is 1.0. 32-bits per channel, 64-bits total per pixel.
  • 8-bit fixed (A) a8fixed - An Alpha only format that has 8-bits per channel, 8-bits per pixel.
  • 16-bit fixed (A) a16fixed - An Alpha only format that has 16-bits per channel, 16-bits per pixel.
  • 16-bit float (A) a16float - An Alpha only format that has 16-bits per channel, 16-bits per pixel.
  • 32-bit float (A) a32float - An Alpha only format that has 32-bits per channel, 32-bits per pixel.
  • 8-bit fixed (Mono+Alpha) monoalpha8fixed - A 2 channel format, one value for RGB and one value for Alpha. 8-bits per channel, 16-bits per pixel.
  • 16-bit fixed (Mono+Alpha) monoalpha16fixed - A 2 channel format, one value for RGB and one value for Alpha. 16-bits per channel, 32-bits per pixel.
  • 16-bit float (Mono+Alpha) monoalpha16float - A 2 channel format, one value for RGB and one value for Alpha. 16-bits per channel, 32-bits per pixel.
  • 32-bit float (Mono+Alpha) monoalpha32float - A 2 channel format, one value for RGB and one value for Alpha. 32-bits per channel, 64-bits per pixel.


Operator Inputs

  • Input 0: -


Info CHOP Channels

Extra Information for the can be accessed via an Info CHOP. Info Channels Common Page

Common Operator Info Channels

  • total_cooks - Number of times the operator has cooked since the process started.
  • cook_time - Duration of the last cook in milliseconds.
  • cook_frame - Frame number when this operator was last cooked relative to the component timeline.
  • cook_abs_frame - Frame number when this operator was last cooked relative to the absolute time.
  • cook_start_time - Time in milliseconds at which the operator started cooking in the frame it was cooked.
  • cook_end_time - Time in milliseconds at which the operator finished cooking in the frame it was cooked.
  • cooked_this_frame - 1 if operator was cooked this frame.
  • warnings - Number of warnings in this operator if any.
  • errors - Number of errors in this operator if any.