Colour space on Sony SLT Alpha 33L DSLR

[andrewgerm]

Good day all
We are using a second hand Sony SLT Alpha A33L DSLR camera to shoot AVCHD video. Soon, we will be working on a project with another team, and as part of pre-production, they want to know the colour space of the camera (i.e. video).

The manual, and Sony website, are not very forthcoming with regards to video, but do mention that stills can be sRGB or AdobeRGB.

Can anyone shed any light on what the video colour space would be?

Much appreaciated

[jagabo]

Since it shoots AVCHD it's almost certainly rec.709 YUV with 4:2:0 chroma subsampling. MediaInfo may be able to tell you.

[andrewgerm]

Thank you Jagabo
Just downloaded MediaInfo it, so what should I look for?
I'm using a Vista 64 system.

From what I can see, it is 4:2:0, YUV with High@L4.0 as the Format Profile.

[intracube]

It's also worth mentioning that a lot of consumer cameras record video that doesn't follow the Rec.709 standard - using full swing 0-255 instead of 16-235 range. Without user intervention the shadows can get crushed and the highlights clipped, so watch out for this.

Any decent software should show a histogram with the legal ranges marked out.

[jagabo]

It's also worth mentioning that a lot of consumer cameras record video that doesn't follow the Rec.709 standard - using full swing 0-255 instead of 16-235 range.

I've only seen that with cameras that record in MJPEG where it's normal for luma to range from 0 to 255.

From what I can see, it is 4:2:0, YUV with High@L4.0 as the Format Profile.

I don't know of any program which will tell you more.

[intracube]

It's also worth mentioning that a lot of consumer cameras record video that doesn't follow the Rec.709 standard - using full swing 0-255 instead of 16-235 range.

I've only seen that with cameras that record in MJPEG where it's normal for luma to range from 0 to 255.

I've noticed this on quite a few AVCHD cameras.

For example, D800 Video 5 (original MOV):
http://www.dpreview.com/reviews/nikon-d800-d800e/30

or this D600 video:

(even after it's been re-encoded by YT)

I think I've interpreted the results correctly. I don't know of a way to generate an accurate histogram* on linux (like AviSynth for Windows can do), so I use mplayer -vo gl:levelconv=2 video.mp4 to show the full range instead of 16-235, and visually look for more picture in the shadows/hightlights. Recent versions of ffmpeg have the lutyuv filter that can be used to remap the range.

*EDIT: and by histogram I mean the levels visualisation...

[jagabo]

You're right. I downloaded a few of the videos in that review and they use the PC.601 matrix (frame size reduced):



You can see excursions below Y=16 (it's hard to see in these small images with little bright content but there are excursions above Y=235 too) and that blacks are crushed by incorrect conversion to RGB using the rec.601 matrix.

Below is the image after correcting for PC to TV levels (ColorYUV(levels="PC->TV") in AviSynth):




MediaInfo shows the use of the BT.601 matrix (usually used for standard definition), not BT.709 (usually used for high definition):

Video
ID                                       : 1
Format                                   : AVC
Format/Info                              : Advanced Video Codec
Format profile                           : High@L4.0
Format settings, CABAC                   : Yes
Format settings, ReFrames                : 2 frames
Codec ID                                 : avc1
Codec ID/Info                            : Advanced Video Coding
Duration                                 : 31s 832ms
Bit rate mode                            : Variable
Bit rate                                 : 22.1 Mbps
Width                                    : 1 920 pixels
Height                                   : 1 080 pixels
Display aspect ratio                     : 16:9
Frame rate mode                          : Constant
Frame rate                               : 29.970 fps
Color space                              : YUV
Chroma subsampling                       : 4:2:0
Bit depth                                : 8 bits
Scan type                                : Progressive
Bits/(Pixel*Frame)                       : 0.356
Stream size                              : 83.9 MiB (93%)
Language                                 : English
Encoded date                             : UTC 2012-03-17 17:00:09
Tagged date                              : UTC 2012-03-17 17:00:09
Color primaries                          : BT.709
Matrix coefficients                      : BT.601
Transfer characteristics                 : BT.470 System M

Andrewgerm, if you upload a short sample I'll take a look at it. Use a clip with a mix of bright and dark content. Something with a known color would be good too -- especially reds (where the difference between the 601 and 709 matrices is most obvious).

[andrewgerm]

Thank you everyone for the information, and quick assitance.

The responses have gone way beyond my question, so a big thank you!