Archive for February, 2012

Matt Roberts (MBR’s) Automatic Color (Chart) Corrector

Tuesday, February 14th, 2012

How about an After Effects plugin for automatically grading any footage featuring a Gretag Macbeth color chart in-shot (e.g. at the beginning and/or end of shot)?  Matt Roberts’ new plugin, still “steaming off the press”, works in Premiere as well as After Effects, and has been tested in CS5 and CS5.5.  You simply pause on a frame featuring a color chart in-shot, place corner locators to identify that chart, and ‘Go”.  It not only fixes white balance but also adjusts for saturation and compensates for certain kinds of “color twisting” defects such as can occur in cameras.    Subsequent “expert tweaks” can then be made if preferred, e.g. 20% saturation reduction for “film look”.  The free version works in 8 bits, the paid (£50) one (in the process of being made available on ShareIt.com) works in 32 bits, multithreaded etc.  To find out more and to download it: http://www.mattroberts.org/MBR_Color_Corrector/.

Example: Canon 7D Video Footage:

Canon 7D before correctionCanon 7D after correction

So what’s the point of this plugin?  Greater quality, reliability and productivity, as compared to traditional color correctors, as explained below.

Those with an eye for accurate color reproduction from video footage will be familiar with traditional tools such as 3-way color correctors and meters such as waveform monitors and vectorscopes.   All proper Non Linear Editing systems (NLE’s) have these.  Generally-speaking such tools work well, but sometimes in practice the situation can become confused when for example a subject’s “white”(assumed) shirt is in fact off-white, or when tinted light mirror-reflects off skin or results from camera filters.  Easy to understand in retrospect, but initially can cause “running round in circles” of interative adjustment and re-checking.  Furthermore, some cameras have peaks, pits, twists and ambiguities (e.g. infra-red) in their colour response that many such correctors cannot correct in a straightforward manner.   Not only can time be wasted but it is quite possible to end up with an image that “looks” right to most people but which in fact has done something inexcusable such as altering the very precise color of a corporate logo.

One way to reduce the potential fo such confusion is to incorporate a color chart in shot.  Various types exist, including Gretag Macbeth (GM) and Chroma Du Monde (CDM).  The GM card, while primarily targeted at photography, is also in widespread use for video.   That chart consists of a matrix of colored squares, one row of which represents (steps on) a grey-scale.  It also includes some near-primary colours and some approximate skin colours of a few types.  The simplest use of such a chart would be to use the grey-scale row for white balancing and the other colours for “by eye” grading/tweaking.  The more experienced will probably make use of vectorscopes etc. but that can still be a nuisanceful if not cumbersome process.

Enter Matt Roberts’ Automatic Color Corrector.  We tried it out on some footage from his own Canon 7D and from my Sony EX3, the latter fitted with a slightly green-tinted infra-red filter, on a snowy day.  We even tried it on an image (featuring such a chart, as well as a model with lots of fleshtones) on Canon’s website ?URL? for their C300 camcorder.  In all cases, the correction was achieved in seconds.  We were particularly confused as to why Canon’s web-image image was so off-colour, but it certainly was, and the Corrector fixed it.

Once again, the link: http://www.mattroberts.org/MBR_Color_Corrector/

MXF and its Variants

Friday, February 10th, 2012

Previous coverage of this topic: http://blog.davidesp.com/archives/400

Further info, including a great technical breakdown of the format itself: http://www.mog-solutions.com/img_upload/PDF/MOG_SOLUTIONS_operational_patterns_NAB2007.pdf, discovered via Adobe Premiere CS5.5 Help on “Importing assets from tapeless formats”.

TimeSeries Analysis in R

Wednesday, February 1st, 2012

When the time comes, this looks like a good tutorial / experience:

Image Analysis & Processing in R

Wednesday, February 1st, 2012