May Bumps 2015 Thursday Footage

Thursday footage:

Cambridge May Bumps 2015 division M1 normal and slow motion
Cambridge May Bumps 2015 division M1 normal speed
Cambridge May Bumps 2015 division M1 slow motion
Cambridge May Bumps 2015 division M2 normal and slow motion
Cambridge May Bumps 2015 division M2 normal speed
Cambridge May Bumps 2015 division M2 slow motion
Cambridge May Bumps 2015 division W1 normal and slow motion
Cambridge May Bumps 2015 division W1 normal speed
Cambridge May Bumps 2015 division W1 slow motion

May Bumps 2015 Wednesday Footage

Wednesday footage:

Cambridge May Bumps 2015 division M3 normal and slow motion
Cambridge May Bumps 2015 division M3 normal speed
Cambridge May Bumps 2015 division M3 slow motion
Cambridge May Bumps 2015 division M4 normal and slow motion
Cambridge May Bumps 2015 division M4 normal speed
Cambridge May Bumps 2015 division M4 slow motion
Cambridge May Bumps 2015 division M5 normal and slow motion
Cambridge May Bumps 2015 division M5 normal speed
Cambridge May Bumps 2015 division M5 slow motion
Cambridge May Bumps 2015 division M6 normal and slow motion
Cambridge May Bumps 2015 division M6 normal speed
Cambridge May Bumps 2015 division M6 slow motion
Cambridge May Bumps 2015 division W2 normal and slow motion
Cambridge May Bumps 2015 division W2 normal speed
Cambridge May Bumps 2015 division W2 slow motion
Cambridge May Bumps 2015 division W3 normal and slow motion
Cambridge May Bumps 2015 division W3 normal speed
Cambridge May Bumps 2015 division W3 slow motion
Cambridge May Bumps 2015 division W4 normal and slow motion
Cambridge May Bumps 2015 division W4 normal speed
Cambridge May Bumps 2015 division W4 slow motion

Spring Head to Head 2015 Footage

Playlists for the Spring Head to Head 2015.

Oarstack Analysis

Do you ever get to the end of a race and think “I wish I knew the coefficient of variation of our catch timing during that piece,” or “What was the standard deviation of our drive/recovery ratio?” Or even something esoteric like “What was the stroke rate?” Well now you can find out with Oarstack analysis!

Bandwidth

One word of caution before we start – it’s not YouTube.  When you click the play button your browser has to download the full 1920×1080 high quality video file, regardless of the resolution you’ve chosen or the speed of your connection.  It’s not ideal if you’re paying by the megabyte.

Things to do

  • You can just use it as a player.  You’re watching the files that get uploaded to YouTube, but without the extra compression that YouTube can add.
  • Change the playback speed by clicking the ‘1x’ button on the bottom right.
  • Everything revolves around markers, which you place at the current time using MARK CATCH and MARK EXTR. buttons on the control bar.  If all you want to do is measure stroke rate then all you need to do is place the catch markers.
  • Clicking on a marker takes you to it, and the MARK button becomes and UNMARK button to remove the marker.
  • The step and frame advance buttons can are handy to step through the video.  There are time buttons, where ‘0.5s’ steps forward half a second, and frame buttons, where ‘-4f’ steps backward four frames.  These buttons can be held to step continuously, and holding ‘1f’ gives a handy slow playback.
  • Once you have a few markers, you can get at the data by either hovering the mouse over the markers themselves or reading the tables below the video.
  • The tables can be cut and pasted into spreadsheets, and then used to make graphs, etc.
  • Right now the only way to store markers is by embedding them in a link using the box below the video.  If you share those links the markers will be there when anybody clicks the link.  You can edit the numbers in the link too.
  • You don’t have to use the markers just for catches and extractions.  With of bit of invention you can measure how well the crew is in time, etc.  Or at least you could if I’d taken it a bit more easy with the zoom lens.
  • Form a crack team of specialists to find out the best time to say the catch has actually happened.  Is it when the blade first touches the water?  Or when when it’s covered?  Or when it’s half way in?  Likewise for extractions, which tend to be more ropey.  In theory there’s an ideal point that gives you the lowest standard deviations, but as far as I know it’s one of the great unknowns of modern science.  It’s right up there with Dark Matter.

On that last one, for now let’s go with the catch having ‘happened’ when the spoons are, averaged over the crew, 50% in the water.

Browser quirks

The analyser is abusing HTML5 and Javascript quite a bit, and is a bit ‘experimental’.  Here are a few of the things that don’t quite work.

  • Internet Explorer: Nothing works at all until the entire video has loaded, including (inconveniently enough) the load progress meter.
  • Opera: Clicking on the video itself only toggles pause/play whilst the button is held – it doesn’t toggle pause/play per press like other browsers.
  • Firefox: Video flickers black momentarily when you use the controls.  ‘1x’ playback speed only goes down to 0.25x.
  • Chrome: Is not bad – nothing to report so far.
  • Safari: This is the only browser that can play video backwards smoothly – keep clicking 1x and you’ll get negative numbers.  Behaves badly until the entire video is loaded.
  • Mobile devices: I haven’t found one that works at all yet, although some will at least play the footage.
  • Browsers without HTML5 support: No support for these, if you can find one.

Feel free to log issues here: https://github.com/sloe/analyseapp/issues or mail to [email protected].

Extracting a photo from Oarstack video

If you’re looking for a crew photo, it is possible to extract one from Oarstack video on YouTube. Results will vary, but they can be OK, like the frame below from this video. Click the images for larger versions.

magicactions-extracted-closeup

At the time of writing I’m using the Magic Actions for YouTube extension with the Chrome browser to extract frames. Plugins can turn into a security risk, so you’ll need to satisfy yourself that it’s still safe.

If you’re happy and have the extension installed, first adjust some settings. If the extension is working you’ll see its icons below YouTube videos. Click the settings one as shown below.

magicactions-settings-icon

Most of the settings don’t matter, but the resolution should be set to the same as the source video, which for Oarstack is usually HD1080p.

magicactions-settings-hd

Setting the player type to HTML5 will activate the slow playback features, which will be handy for choosing the right frame.

magicactions-settings-player-type

Then find your chosen video – this is our example – and set the playback speed to 0.25. This will make it easier to choose a good frame.

magicactions-settings

Drag the slider to somewhere just before the area you’re interested in.

magicactions-coarse-find

Play the video and try to pause it at the right frame. If you miss it, drag the slider back and try again.

magicactions-fine-find

When paused at the right frame, click the Capture icon.

magicactions-capture

This will take you to the capture page.

magicactions-capture-large

It’s worth experimenting with the sharpen filter if the image is soft. Click the save button when done.

magicactions-capture-save

You should now have a 1920×1080 image, a bit like the one below.

magicactions-extracted-youtube

Things to consider are:

  • Video compression will produce artefacts in the images. These will be at a minimum when there’s not much ‘going on’, or not many rapid changes to compress, at that point of the video.
  • Most Oarstack slow motion video alternates between real frames that the camera captures, and interpolated frames that the software generates. If some frames are interpolated the description will say so. Interpolated frames tend to have more artefacts than the real ones.
  • Oarstack video is taken with a fast shutter speed – usually 1/1000s or faster – to aid generation of slow motion using frame interpolation. This is unusual, and makes the video look jumpy when viewed at full speed, but is good for grabbing frames. Other YouTube videos might not give such good results.
  • Oarstack video and images can be used freely for personal use.

May Bumps 2014 Thursday Footage

May Bumps 2014 highlights (extended with Thursday footage)

Cambridge May Bumps 2014 division M1 normal and slow motion
Cambridge May Bumps 2014 division M1 normal speed
Cambridge May Bumps 2014 division M1 slow motion

Cambridge May Bumps 2014 division W1 normal and slow motion
Cambridge May Bumps 2014 division W1 normal speed
Cambridge May Bumps 2014 division W1 slow motion

Cambridge May Bumps 2014 division M2 normal and slow motion
Cambridge May Bumps 2014 division M2 normal speed
Cambridge May Bumps 2014 division M2 slow motion

May Bumps 2014 Wednesday Footage

May Bumps 2014 highlights

Cambridge May Bumps 2014 division W2 normal and slow motion
Cambridge May Bumps 2014 division W2 normal speed
Cambridge May Bumps 2014 division W2 slow motion

Cambridge May Bumps 2014 division M3 normal and slow motion
Cambridge May Bumps 2014 division M3 normal speed
Cambridge May Bumps 2014 division M3 slow motion

Cambridge May Bumps 2014 division W3 normal and slow motion
Cambridge May Bumps 2014 division W3 normal speed
Cambridge May Bumps 2014 division W3 slow motion

Cambridge May Bumps 2014 division M4 normal and slow motion
Cambridge May Bumps 2014 division M4 normal speed
Cambridge May Bumps 2014 division M4 slow motion

Cambridge May Bumps 2014 division W4 normal and slow motion
Cambridge May Bumps 2014 division W4 normal speed
Cambridge May Bumps 2014 division W4 slow motion

Cambridge May Bumps 2014 division M5 normal and slow motion
Cambridge May Bumps 2014 division M5 normal speed
Cambridge May Bumps 2014 division M5 slow motion

Cambridge May Bumps 2014 division M6 normal and slow motion
Cambridge May Bumps 2014 division M6 normal speed
Cambridge May Bumps 2014 division M6 slow motion