Monthly Archives: September 2011

BOUJOU 3D MOTION TRACKING TUTORIAL

http://lesterbanks.com/2010/06/boujou-3d-motion-tracking-tutorial/

A video tutorial where Leo Covarrubias goes over how to do a basic but accurate 3D Motion Track in Boujou. Some of the topics covered are, radial camera distortion compensation, tips for quality footage, exporting a track scene to Autodesk Maya. This is a great introduction to Boujou and its interface and will let you integrate 3d objects and characters into live action footage easily.

Apple iPhone 4 camera specs

On 2010, June 7 His Steveness has introduced the Apple iPhone 4 which features a new camera module. Two actually but I am going to talk about the main camera (on its rear side) only. And only about its still photography capabilities (it records 720p 30fps HD video as well).

It is typically hard to figure out the exact camera specification for cell phones and to compare them with a P&S camera. E.g., the HTC EVO 4G is said to have an 8 megapixel camera but little else is known. Images shot with the EVO all lack the significant EXIF information!

However, I think I managed to extract the full iPhone 4 camera specs now.

First, the pixels are 1.75 µm (according to Job’s keynote) and we may assume that this is the pixel pitch. Sample images are 2592×1936 pixels (5.0 MP, 4.02:3). This makes a 4.54 x 3.39 mm^2 chip (5.67 mm diagonal) which corresponds to a 35mm film camera crop factor of 7.64.

In a P&S, this would be called a 1/3.2″ sensor and P&S sensors range between 1/3.8″ and 1/1.5″ (crop factors between 8.6 and 3.9). So, it is an imaging sensor at the lower end of the P&S range, but it is in the P&S range and with a reasonable number of pixels and a back-illuminated sensor like the Sony Exmor R sensor. Here is a sample image from Apple’s web site:


(click on the image for the sample in original size; hosted on apple.com)

Gizmodo.de has published the full EXIF data for exactly this sample photo (it is magically absent from the web sample though …):

Aperture: 2.97
Color space: Uncalibrated
Date and time: 2010/05/07 15:14
Exposure: Auto exposure
Exposure time: 1/887
Flash: Flash did not fire, auto mode
FNumber: 2.8
Focal length: 3.85
ISO speed rating: 80
Metering mode: Spot
Sharpness: Hard
White balance: Auto

So, the lens is a 3.85 mm f/2.8 lens.

This corresponds to 35mm film camera equivalent lens of 29.4mm f/21.4.

I should note that even an aperture this small is not restricted by diffraction yet (limit at f/2.8 is at 54% of pixel width). This is a good value. But one really shouldn’t stop down beyond f/4 and it seems the iPhone 4 has an f/3 fixed aperture lens anyway.

Summary

iPhone 4 main camera:

  • 5.0 Mpixels (2592 x 1936)
  • 1/3.2″ back-illuminated CMOS sensor
  • 4:3 aspect ratio
  • 35 mm film camera crop factor: 7.64
  • Low ISO 80 (or better)
  • 3.85 mm lens focal length
  • f/2.8 lens aperture
  • Autofocus: tap to focus

Equivalent 35mm film camera and lens:

  • 30 mm f/22

So, don’t throw away your FourThird, APS-C or full frame SLR ;)

Besides the deep depth of field, an f/21.4 aperture at ISO 80 produces as much noise as an f/8 lens at ISO 570 (for a given shutter speed). Where 35mm film camera f/8 and APS-C f/5.6 produce the same level of noise.

Conclusion

The iPhone 4 seems to have a camera which should deliver an image quality on par with the lower segment of P&S cameras. It’s field of view (no zoom!) corresponds to a 30mm lens (on a 35mm film camera) and it captures as much light (produces as much noise) as a system camera with a lens aperture between f/11 (FourThird) and f/22 (35mm film).

POSTED BY FALK LUMO AT 01:56

http://falklumo.blogspot.com/2010/06/apple-iphone-4-camera-specs.html

SCIgen, An automatic computer science paper generator.

SCIgen is “An automatic CS Paper Generator”. It can generate a random computer science paper in 10 seconds.

 

Please have a look this site.

http://pdos.csail.mit.edu/scigen/

 

The authors said that the primary objective of this work is for fun however, I think there are important implications beyond this.

A random generated paper was accepted by WMSCI 2005!! and this conference disappears at this moment.