Introduction

λ·vue is a video magnification tool developed to reveal signals in recorded video that are normally invisible to the naked eye.

Quanta Research Institute (QRI) owns and operates the λ·vue web service.

Zero, λ·vue is a free-to-use service.

For technical help, please contact tech.support@lambda.qrilab.com. For any other questions, please contact info@lambda.qrilab.com.

Video magnification processing is computationally expensive. For videos with a small frame size (640x480) and low frame rate (15-30fps) Processing takes roughly two to four times as long as the video itself, so a 20 second video will take 40-80 seconds to process. Also, keep in mind that a large frame size video (1920x1020, also known as 1020p) is about six times the size of—six times the pixels—a small frame size video (640x480), and takes about six times as long to process.

Videos magnification processing time is closely related to the total size of the video in pixels: the video width times the video height times the number of frames. If you can compute the size of your video (in millions of pixels, mp) you can estimate video processing time (in seconds) using the following formula:

  • mp = width x height x num_frames / 1000000
  • Processing time ≅ mp x 0.23 + 13 seconds

If you choose to upload your own video, we recommend that they have the following properties:

  • Encoded using H.264 in an MP4 container. If your video is not in this format, we will attempt to transcode the video, but we cannot guarantee that we can handle all input formats.

Note that some video formats, e.g., Windows Media Video (.wmv), may not be recognized by the browser as video files and not show up in the video upload dialog.

  • Be less than 45 MB in size. Be aware that large videos can take a long time to upload and to process; assume the upload will progress at most at 1MB/s (so 20 seconds for a 20MB file).

Note that using ideal temporal filtering will use a large amount of memory. You will be presented with an error if you try to process a video that is too large.

Account

You do not need an account to view works published by others. However, to create work of your own, you will need to create your own account. To create an account, go here.

You can request to have your password reset by clicking the forget password option from the sign-in page.

Copyright

You should only upload video that you have copyright to or is in the public domain. It is your responsibility to verify the video you uploaded falls into these categorie

Derived works are processed video that λ·vue generated based on the uploaded video. You own the rights to the video processed by you if the source video used to create such video does not infringes any copyrights.

Playground

In λ•vue, frequency is defined as the number of motion or color changes per second that occurs anywhere in your video.  When you specified a frequency range (such as 1-2 Hz), you are essentially instructing our software to amplify the motion or color changes that falls within this range (1-2 times/second or 60-120 times/minute).  In other words, changes within this range will have its change effect magnified by the amount you specified while changes outside of this range will remain untouched.

The frequency range that you chose thus has a significant impact on what you will see in the resulting video.  Using the right frequency range instructs our software to focus on amplifying only the changes most interesting to you and thus let you observe the phenomenon you are interested more easily.  On the other hand, using a less than ideal frequency range often produces video with too much noise, making it nearly impossible to see what you are interested.

In λ•vue, amplification specifies the number of times you want to intensify the change effect.  Take a pixel color change that changes from white to gray by 1 color gradient for example.  When you apply an amplification of 10 to this effect, the result will be a darker grey pixel that has been shifted by 10 color gradients instead.  Essentially, the higher the amplification, the larger the movement or color changes you will see which can be good or bad depending on what you use as the frequency range.

The “right” frequency range completely depends on the phenomenon that you wish to observe within your video. For example, since the normal resting adult human heart rate ranges from 60-100 beats/min., to magnify the heart rate, you should set the frequency range to 1-1.67 Hz. If you know that the subject that you are observing has a heart rate between 80 and 90 beats/min., you can further improve the quality of the magnified video by setting the frequency range to 1.3-1.5 Hz instead as this narrower value will often produce video with less noise.

λ•vue lets you specify this range (between 0-15 Hz) in two ways. In Advanced mode, you specify an arbitrary range using the scroll bar or by directly entering the lower and upper bound. In Simple mode, you specify a predefined range by clicking on one of the 5 circles:

interface

1 = 0.1-3 Hz

5 = 3-6.5 Hz

8 = 6.5-9 Hz

10 = 9-12 Hz

14 = 12-15 Hz

To get a better idea on how fast these changes are, check out the bouncing ball visualization below depicting the movement at 1, 8, and 14 Hz respectively.

1 time/sec (1Hz) 8 times/sec (8Hz) 14 times/sec (14Hz)