(1) Plug the power (black), synch (gray) and RGB cable (red, green, blue) from the camera to a ceiling socket shown in the right above image.
(2) There are 14 ceiling sockets installed on the ceiling and
the wall
of the studio. These sockets are labelled 1-14 which furtherly
corresponding to the patch connections in control room. So if you
connect a camera onto a socket, please lable the camera using the
socket number.
(3) Note that these numbered
sockets normally have a fixed connection to the mount panel of the
grabbing rack installed in the contol room. Signals coming from the
mount panel are ROUTED using a switch matrix, the so
called AVROUTER, and then patched into the 14
grabbing computers, which correspond to 14 A/D sampling cards installed.
(4) Note that only one chanel of the 14 signals can be embedded with
time stampers (embedders). This unique signal can also be routed onto a
grabber computer.
(5) We can use the program "avrouter" to route the SDI output from the
embedders. As shown in the figure bellow, in order to set the camera
number (source) in the program "avrouter" corresponds to A/D number on
the patch panel, they (camera signal and A/D signal) should be
connected as shown in the above figure, using the connecters (red,
green and blue connecters). In this way, each socket corresponds to the
same camera number, i.e socket 1 is camera 1, and so on.
(6) So, if you are using the
program "avrouter", for instance, you can route Camera_5 (ceiling
socket_5, actually) to the grab computer CVPSD_05. Of course, Camera_5
(ceiling socket_5) can also be additionally routed to another grab
computer, for instance, CVPSD_02, as shown in the figure bellow.
(7) You can run "avrouter" from any console (for instance: "%rxterm ernie")- if the alias does not track the binary. (The error msg would be "/opt/bin/avrouter: line 2: /opt/PDsoft/RAVL/bin/vscp: No such file or directory" and " /opt/bin/avrouter: line 2: exec: /opt/PDsoft/RAVL/bin/vscp: cannot execute: No such file or directory".) Personally, I would recommend to use the binary located at "/vol/vssp/localsoft/RavlBin/linux-suse9.3/30Jul06/", which is stable and reliable. You might need to set the display variable by using "%setenv DISPLAY ernie:0.0", to run the program "avrouter".
(8) The sources (left) of "avrouter": The destinations (right) of "avrouter":
Camera_1,...,Camera_14:the 14 camera sockets mounted on the ceiling and the wall of the sutdio. CVPSD_01,...,CVPSD_14:the 14 grabbing computers. Pronto:?. SonyD1:?. AudioRack:?. WorkStation_1,...,WorkStation_5:?. Robot_x55, Robot_x56:?. SonyD1Monitor:?. HDTV_Downconverter?. Engineering:?. ColourBars:?.
(9) VSCP:AVRouter Operations:
The VMLs' infrastructure was designed to address the need of flexibility but at the same time ease of use.The core of the system is the 64 in 64 out digital SDI switcher. In day-to-day use this is controlled by the avrouter software,although in exceptional circumstances manual switching can be used via a panel located in room 43AB05
A GUI program activated by typing avrouter into any
Linux machine was given above.
An OPERATION EXAMPLE, from Camera_1 to Storm_input_1, is given bellow:
Click on the Source dropdown box.
Select the location from which the video / audio are coming
Click on the Destination dropdown box.
Select the location to which the video / audio are going
Click Add
Scroll down the source/destination window to check that the
connection has been made
For those who would like more detailed information on this
software see the source code
To view the camera output on a monitor you need to route the SDI stream (sources) from the embedder to an output socket (destinations, for instance, workstation1-5) in the lab and connect the output of the workstation sockets to a monitor. Using the program "avrouter", maxium five SDI streams can be displyed on 5 monitors, independently. Of course, one SDI stream can be switched onto multiple monitors, flexibily.
As shown in the above figure, the outputs are named WorkStation and labelled with a corresponding number. Select the camera number as a source in "avrouter" and the workstation output as a destination (Remember to click the "Add" button to activate the routing matrix). Then connect the output of the workstation sock to the SDI input on the monitor. Make sure that C/SDI and LINE/RGB are selected on the front panel of the monitor.
To route an output to the monitor on the control desk select
Desk_SDMonitor as the destination in avrouter.
Colour Balance |
|
The colour response of the cameras must be matched to either compare colour between different views (assuming the object is perfectly diffuse and has no specularities) or to combine colours sampled from different images without seeing any offsets in the colour levels between the cameras. White-balancing adjusts the white-point of the camera so that with a given lighting colour temperature, a neutral object will appear white in the image. Balancing simply adjusts the RGB levels for a white object under the lighting so that they are equal. White-balancing requires a perfectly diffuse, neutral object that fills the majority of the camera image. The reverse side of the A0 paper roll can be used for this as it is close to neutral. Spray with the "Flawfinder Developer Spray" to give a diffuse surface. Lift and gain provide a linear transformation that matches the response of the cameras by tieing the black level and maximum white level in the scene to a particular point. Again this requires perfectly diffuse black and white targets. For the black level, a felt lined boxed with an opening at the top can be used or simply by switching off all lighting in the lab. For the white level a diffuse 18% grey card is available, corresponding to a 50% signal rather than a 100% maximum signal. The 18% grey chart is held by Andrew Birt. NOTE: Camera zoom will affect the light entering the camera. To white-balance a camera you will most likely have to zoom on a white target positioned at the centre of your capture volume. Do this before you finish the camera setup, then set the lift gain to match levels between cameras with the final zoom for your configuration. |
Ensure that the SD camera is set to Auto White Balance (AWB).
Turn down the aperture on the camera to the point where the scene is not clipping. If the aperature is too great the CCD will clip and the object will always appear white as the R,G,B values are all at the top of the dynamic range. Check the response on the oscilloscope, select the SD monitor from [INPUT], select [WFM], deselect [LINE SEL] and check that the scene is not clipping i.e all below 700mV.
With the menu page off and the camera imaging the neutral object select [White] on the back of the camera. The image should display WHITE OK when complete and the camera is white balanced.
The SD gain is set on page 1 of the camera menu and black level lift is defined by the pedestal on page 2 of the menu.
To set the lift, bring up the SD image on the oscilloscope and select [WFM]. Make sure the baseline is at 0mV and switch on filtering. Switch off all lights in the lab and set the pedestal to lift the signal above the baseline, e.g 0, then reduce the pedestal in - 5/10 unit steps until the signal just reaches the baseline. If you continue reducing the lift the black levels are crushed and will always sit on the baseline. Use [LINE SELECT] to strobe out a single black line if you look at the scope with the SD menu up.
To set the gain, place the 18% grey card at the centre of the volume
with all lights switched on. Use [LINE SELECT] to strobe out a single
line across the chart and set the camera aperture so that the signal
hits the 50% mark on the waveform monitor. If you cannot get a 50%
level even with the aperture fully open, you will need to add some
digital gain.
We can end up using 600mV for green on HD and 700mV on SD while setting up the signal level by adjusting the aperture settings of different cameras. This is because, practically, we are not using SD for colour and we wanted better blue segmentation in the low quality SD images. We use such settings to captuer the data "Roxanna".
Different capturing circumstances can have different settings. Note that singals will be clipped if it is above 700mv. So if we are going to capture close ups, for instance, a typical capther circumstance shown in the figure bellow, we can set the signal level of both HD cameras and SD cameras reach 700mv.
A studio setup for grabbing close ups and head-shoulder images
700mv settings of the green matt (nearly saturated) captured using a HD camera
Black lift of the HD camera
700mv settings for the green matt (nearly saturated) captured using a SD camera
Black lift of the SD camera
In order to guarantee the coincidence of the pixel values between images captured using HD cameras and SD cameras, we highly recommend you to double check the pixel values within individual images using GIMP. Select interested areas using "magic wand", then go to "Dialogues->Histogram" to calculate the mean value of the selected area. The difference between the mean values of different images should not differ too much.