"Once you have the closest object in focus, tapping the screen on your Lumia will start the capture process – which only takes a couple of seconds (try to be very still). During this time, Nokia Refocus performs a ‘focus sweep’ and creates a depth map.
This new way of taking photos is a great example of computational photography. We’ve now got the ability to change a photo once it’s taken and have a photo that simply couldn’t have been taken by a traditional camera.
The result is a series of 2-8 images, depending on what’s in view, each at five megapixels at a different focus that’ll be aligned and forged together to make one image.
By tapping on any part of this final image you’ll be able to shift the focus to that particular part. The great thing here is that you can achieve a variety of different shots from just the one photo."
Refocus app can also put the whole scene into focus at the same time to provide extra clarity:
Engadget says that Refocus is based on Scalado technology that Nokia acquired a year ago.
Hi Vladimir, I just wanted to comment on the header: Refocusing using depth that is acquired by an array camera is something that was invented by LinX Imaging in 2010 and ironically presented publicly on this blog much before others like Pelican Imaging mentioned the terms Refocusing on an array camera. (http://image-sensors-world.blogspot.co.il/2012/07/linx-imagings-multi-aperture-camera.html)
ReplyDeleteFor reference Pelican filed a patent on refocusing too:
https://www.google.com/patents/US8542933?dq=pelican+imaging+refocus&hl=en&sa=X&ei=JYaEUoriHrSg0wXs4IDAAw&ved=0CD4Q6AEwAQ
But fortunately for us we were first:
http://www.google.com/patents/WO2012057623A1
Aside from that, technically speaking Lytro should have the credit for inventing the refocus idea.
Nokia's app is a very different thing which is cool since it doesn't need any special camera.
The downside of using an AF to do the refocusing is that the blur is "real" so the depth of focus depends on the focal length of the lens which is short in mobile.
Artificial blur like what LinX invented allows you to apply stronger blur and emulate a lens with long focal length.
The challenge of doing this well is that the depth maps from a small array camera is not very accurate and is spatially not consistent with the edges of the image. This issue is dealt with using sophisticated algorithms.
Ziv
Actually, I think that you will find the patent you are mentioning relates to a file format and how you encode and decode light field image data – specifically with reference to synthesizing images from different viewpoints. Pelican Imaging's patents appear to first discuss dynamically refocusing images captured by an array camera in patent applications starting in 2008.
Deletewell, Ren Ng published his PhD thesis back in 2005, with half of the content about Refocusing.
ReplyDeleteYes true but all that was on light filed cameras and not array. There is a HUGE difference on the method of doing refocusing.
Deletein some of the DSC's it's done by taking multiple images and moving the camera slightly. One could imagine such app to be used within the mobile industry overcoming the miniature size of an array camera for mobile phone and the by Ziv mentioned limitations to depth measuring
ReplyDelete