Monday, October 5, 2009

Ikena review - part 2

In the second part of the Ikena review, we'll look at the recovery of fine details. In my first test, the camera focusses on some text in low light conditions. As Ikena likes 30fps, I've set my camera to record at 30fps.

I started by importing the footage into Ikena and then scrubbed the timeline to find the beginning of the test. I defined my region of interest in the scene using the Define Region of Interest tool found below the assets.

From that point, my text is spotlighted within the preview window. I clicked on the Live Preview and got to work.

Working from top to bottom in the Enhancements tab, the first stop is Resolution. The Resolution section has three adjustments, Matching Frames, Super Resolution, and Fusion Level.

According to MotionDSP, "The Resolution Enhancement filters work well on video sequences that have high correlation between adjacent frames (i.e., when a number of frames capture the objects or area that you want to enhance). The more frames you have, the better. In our software products, we call this function "matching frames"."

With that in mind, I selected a high number of Matching Frames as I had more than a few seconds of footage to use for the test. My text was relatively close to the lens, so I didn't need to worry about Super Resolution or Fusion Level.

There was a slight improvement, so I moved on. I didn't need to stabilise the video in this test, so the next stop was Deblurring. "Setting a higher deblurring level (above 50) will result in a sharper image as more iterations of the deblurring algorithm will be used. Higher deblurring levels require higher processing times. The Focus slider controls the radius of a point spread function (psf) used for deblurring," says MotionDSP.

With the Live Preview enabled, I used the sliders to find the best results possible.

Good results thus far. I wanted to see how the Contrast controls would work as well as the Super Resolution. I wasn't too excited about the results. So, I un-checked the boxes and reverted to the previous result.

The time to capture and load the footage was just seconds longer than with the competition. In all fairness, the math heavy functions may have been too much for my older processor. In the end, I was able to clear up the compression noise and brighten the image such that the date was clearly visible. The program's report recorded the settings (more on that later) and produced a rich text file to be included with the final package.

Verdict for this round: A-/C-

The A- is for ease of use. It's easy to get footage in and the Live Preview helps dial in the enhancements. The average user will have no trouble getting up to speed fast and producing decent results with this program.

The C- is given for two reasons. Power users will find that they can produce comparable results as fast or faster in Photoshop with their favourite plug-ins. Power users will also want a little more meat on the documentation bone when preparing for trial. Folks in Daubert states might not feel comfortable with lack of users or published studies (though in fairness, that argument can be made about any new product - regardless of its effectiveness). The second reason is price. For the price of one license of Ikena, departments can get two licenses of Adobe's CS4 Master Collection and two copies of ClearID from Ocean Systems. In today's tough economy, price has to be a factor.

Stay tuned for the next round ...

1 comment:

Mike said...

Hi, Jim,

Thanks for taking the time to review Ikena.

Just a couple of notes about your test footage... the super-resolution factor and fusion controls aren't really related to how close your lens was to the subject. You could probably get even better results by manipulating those controls.

"Fusion Level" tells the algorithms how to compensate for camera movement or subject movement. (This is something we'll do a better job of clarifying in our documentation.) In this case where you had a moving camera (thanks to small hand movements) but a stationary subject, you could absolutely benefit from turning up the fusion (all the way to 10). That's probably where you'll see most of the forensic benefit; you could even turn down the Matching Frames to ~15 to reduce the time required for processing.

Then, by adjusting the Super Resolution factor, you could absolutely see some improvement in details. This is one of the special abilities of the software to literally reconstruct images with more detail based on the analysis of multiple frames.

As you've noted, Ikena is quite easy to use "out of the box," but still requires a bit of introductory training to really tap into its incredible forensic potential.

Looking forward to the next installments!


Mike Sonders
Director of Product