3 ms·

From someone who has processed and edited a lot of RED footage, this is pretty much complete hogwash. RED cameras use a RAW codec called REDCODE, which records

From someone who has processed and edited a lot of RED footage, this is pretty much complete hogwash. RED cameras use a RAW codec called REDCODE, which records the raw sensor data with lossless compression. Almost every other camera records RGB data that has been through gamma correction, a colour curve, sharpening (ugh..), and a whole lot of other image processing. These other cameras therefore give you very little flexibility in post production for grading and so on. This gives the illusion that RED cameras have worse colour rendering - because it is not doing much image processing in camera, the un-processed image looks very flat - and it requires some skill to actually expose the image when you're shooting it, and properly process it in post production (this is a professional camera, remember). The amount of crap you see because of errors on set and in processing... But footage from somebody who knows how to shoot it properly is easily on par with the Alexa (which produces lovely skin tones but is very low resolution) and miles better than DSLR cameras with their aliasing and moire that comes from the line skipping and pixel binning that they need to do to get a decent frame rate out of a still camera sensor - not to mention the 8 bit low bitrate RGB codecs they use which are hopelessly lacking as an acquisition format. And the workflow is sooo hard - I mean, it's such a battle to get RED footage into Premiere for example (File -> Import -> OK). And only pretty much every single grading package, every single non-linear editor and compositing package supports RED footage...

davidjohnstone 15y ago
I'm pleased to be corrected by somebody who knows what they're talking about :-)
jermy 15y ago
As only a minor nitpick (my excuse is that this is my day job), you've got it the wrong way round - REDCODE is actually storing data as RGB, whereas 'every other camera' is almost certainly using a MPEG-based codec and storing it as YUV with chroma subsampling - somewhere between 4:1:1/4:2:0 to 4:2:2 depending on the camera in use. As with any lossless compression, it's a compromise between having almost all of the data shot to work with versus pretty much having to always do a full grade on any content you get and the additional online storage space needed in post-production.
- zokier 15y ago
  The Wikipedia page of REDCODE claims that it contains the Bayer-patterned subpixel data, so it's more comparable to RAW images than rendered RGB images. But according to that page, it's also lossy format, which makes it quite interesting technically.
- stephen_g 15y ago
  Are all the MPEG codecs YUV? I know MPEG2 is but I thought MPEG4/H.264 wasn't. Now that I think about it, you're probably right about that part (as for other codecs, I know the ProRes codec that the Alexa typically shoots (unless you use the ArriRAW recorder but hardly anyone does yet) definitely is YUV, but the HDCAM tapes that some of the more expensive Sony cameras shoot isn't). But the point still stands that RGB and YUV codecs aren't RAW like REDCODE is. REDCODE itself isn't exactly RGB. It actually is pretty interesting - it does store the red, green and blue pixels separably, but it's not usually classified as RGB since it stores two green channels for every red and blue channel (because of the way the bayer filter on the sensor is arranged - there's twice as many green pixels as red and blue), and because those channels haven't been demosaiced/debayered[1]. 1. http://en.wikipedia.org/wiki/Demosaicing http://en.wikipedia.org/wiki/Demosaicing

davidjohnstone 15y ago

I'm pleased to be corrected by somebody who knows what they're talking about :-)

jermy 15y ago

As only a minor nitpick (my excuse is that this is my day job), you've got it the wrong way round - REDCODE is actually storing data as RGB, whereas 'every other camera' is almost certainly using a MPEG-based codec and storing it as YUV with chroma subsampling - somewhere between 4:1:1/4:2:0 to 4:2:2 depending on the camera in use. As with any lossless compression, it's a compromise between having almost all of the data shot to work with versus pretty much having to always do a full grade on any content you get and the additional online storage space needed in post-production.

zokier 15y ago
The Wikipedia page of REDCODE claims that it contains the Bayer-patterned subpixel data, so it's more comparable to RAW images than rendered RGB images. But according to that page, it's also lossy format, which makes it quite interesting technically.
stephen_g 15y ago
Are all the MPEG codecs YUV? I know MPEG2 is but I thought MPEG4/H.264 wasn't. Now that I think about it, you're probably right about that part (as for other codecs, I know the ProRes codec that the Alexa typically shoots (unless you use the ArriRAW recorder but hardly anyone does yet) definitely is YUV, but the HDCAM tapes that some of the more expensive Sony cameras shoot isn't). But the point still stands that RGB and YUV codecs aren't RAW like REDCODE is. REDCODE itself isn't exactly RGB. It actually is pretty interesting - it does store the red, green and blue pixels separably, but it's not usually classified as RGB since it stores two green channels for every red and blue channel (because of the way the bayer filter on the sensor is arranged - there's twice as many green pixels as red and blue), and because those channels haven't been demosaiced/debayered[1]. 1. http://en.wikipedia.org/wiki/Demosaicing http://en.wikipedia.org/wiki/Demosaicing