BBC Virtual Studio 1993-1999

JR

jrothwell97 14 April 2007 6:33pm

AFAIK the background was not blue-or-green screen, but the insets were obviously achieved by CSO.

What I suspect is that the background of the three globes existed in reality. This is only speculation, but I think that the insets were achieved by using an empty shot of the studio.

The image of the presenter sitting in his chair goes to a computer, which already has in its memory a shot of the studio, in the same configuration and with the camera in the same position, but without the newsreader, the chair or the desk.
The computer searches for parts of the live image that are the same as in the pre-existing empty shot, and keys these out (essentially cutting out these parts of the image).
The computer overlays the live shot over the empty shot. It then adds the inset over the empty shot.

This is probably overcomplicating things, and it would probably need incredible computing power to do this frame-by-frame 25 times a second for half an hour, but it could have been done that way. Either that or the background was just a green screen.

The desk configuration probably remained the same throughout programmes, but it was probably possible to key out people who weren't being talked to/interviewed at the time by simply using an existing shot of the studio (as described above, but with the empty shot over the live shot).

This would require pre-configured camera angles though with no opportunity for even the slightest deviation.

SE

seamus 14 April 2007 7:47pm

StuartPlymouth posted:

BBC WORLD posted:

Does anyone have a foto of how the dry set looked? (i.e. the actual desk and backdrop, without the virtual additions)

A link to the BBC Engineering Journal is on page one of this thread, but there seems to be a problem with the link. Fortunately I had copied the document (as you do) and here are the pics:

http://www.rp-networkservices.com/tvforum/uploads/realnews.gif

http://www.rp-networkservices.com/tvforum/uploads/realnews2.gif

Sorry about the quality, apparantly the documents was a very old paper version which had recently been scanned.

I think it actually looks alot better like that. I don't know why they had to have the whole big virtual thing, when the real shot looked alot better, albeit a bit cramped. I remember the set, but just barely. The ominous music used to creep me out a bit

.

JH

Jonathan H 14 April 2007 10:09pm

CharlieMouse posted:

A clever system, that worked seamlessly.

It looked great, but it really wasn't a "clever" system. It was very simple!

CharlieMouse posted:

You'd be hard-pressed to duplicate it, these days, with automated gizmos and server-based doodahs.

That really isn't true at all. Almost any VR system in use today (even the bad ones) would be able to recreate this look very easily, AND make it so that it was genuinely virtual!

JH

Jonathan H 14 April 2007 10:12pm

jrothwell97 posted:

AFAIK the background was not blue-or-green screen, but the insets were obviously achieved by CSO.

What I suspect is that the background of the three globes existed in reality. This is only speculation, but I think that the insets were achieved by using an empty shot of the studio.

The image of the presenter sitting in his chair goes to a computer, which already has in its memory a shot of the studio, in the same configuration and with the camera in the same position, but without the newsreader, the chair or the desk.
The computer searches for parts of the live image that are the same as in the pre-existing empty shot, and keys these out (essentially cutting out these parts of the image).
The computer overlays the live shot over the empty shot. It then adds the inset over the empty shot.

This is probably overcomplicating things, and it would probably need incredible computing power to do this frame-by-frame 25 times a second for half an hour, but it could have been done that way. Either that or the background was just a green screen.

I don't know where to start! You're missing some fundamentals of how keying worked then and now. And yes, you're massively overcomplicating it!

CM

CharlieMouse 14 April 2007 10:48pm

Jonathan H posted:

CharlieMouse posted:

A clever system, that worked seamlessly.

It looked great, but it really wasn't a "clever" system. It was very simple!

CharlieMouse posted:

You'd be hard-pressed to duplicate it, these days, with automated gizmos and server-based doodahs.

That really isn't true at all. Almost any VR system in use today (even the bad ones) would be able to recreate this look very easily, AND make it so that it was genuinely virtual!

I meant "clever" in the sense of "clever people using kit in an innovative and effective way." Agreed, the system itself was entirely dumb, and really a triumph for smoke & mirrors, making people believe all that "virtual reality" hype.

And as far as the recreating today is concerned, yes you're probably right, in terms of modern live CGI rendering. But thinking in terms of the control of CRV machines, and DVEs, and manually timed lighting changes, and CSO from (in some cases) rather non-blue backgrounds, which this system employed, I don't think we'd be able to do it any more. (Not, of course, that we'd want to!)

JH

Jonathan H 14 April 2007 11:12pm

CharlieMouse posted:

Jonathan H posted:

CharlieMouse posted:

A clever system, that worked seamlessly.

It looked great, but it really wasn't a "clever" system. It was very simple!

CharlieMouse posted:

You'd be hard-pressed to duplicate it, these days, with automated gizmos and server-based doodahs.

That really isn't true at all. Almost any VR system in use today (even the bad ones) would be able to recreate this look very easily, AND make it so that it was genuinely virtual!

I meant "clever" in the sense of "clever people using kit in an innovative and effective way." Agreed, the system itself was entirely dumb, and really a triumph for smoke & mirrors, making people believe all that "virtual reality" hype.

And as far as the recreating today is concerned, yes you're probably right, in terms of modern live CGI rendering. But thinking in terms of the control of CRV machines, and DVEs, and manually timed lighting changes, and CSO from (in some cases) rather non-blue backgrounds, which this system employed, I don't think we'd be able to do it any more. (Not, of course, that we'd want to!)

Some or all of that stuff happens every day on ITV News!

CM

CharlieMouse 14 April 2007 11:30pm

Jonathan H posted:

CharlieMouse posted:

Jonathan H posted:

CharlieMouse posted:

A clever system, that worked seamlessly.

It looked great, but it really wasn't a "clever" system. It was very simple!

CharlieMouse posted:

You'd be hard-pressed to duplicate it, these days, with automated gizmos and server-based doodahs.

That really isn't true at all. Almost any VR system in use today (even the bad ones) would be able to recreate this look very easily, AND make it so that it was genuinely virtual!

I meant "clever" in the sense of "clever people using kit in an innovative and effective way." Agreed, the system itself was entirely dumb, and really a triumph for smoke & mirrors, making people believe all that "virtual reality" hype.

And as far as the recreating today is concerned, yes you're probably right, in terms of modern live CGI rendering. But thinking in terms of the control of CRV machines, and DVEs, and manually timed lighting changes, and CSO from (in some cases) rather non-blue backgrounds, which this system employed, I don't think we'd be able to do it any more. (Not, of course, that we'd want to!)

Some or all of that stuff happens every day on ITV News!

Lucky ITV news, if they still have the skills base to program CRV machines. Unlucky ITV News, if they're still using CRV machines!

More seriously, I'm hinting at the fact that a lot of de-skilling has gone (certainly in areas I have some acquaintance with.) Colour matching a live camera to a recorded CRV would be impractical, and essentially impossible, in a gallery such as News 24, since there is nobody dedicated to the job, and nobody has time to do it. That's no criticism of the people concerned, but it is a criticism of the system under which they work.

ST

Stuart 15 April 2007 1:16pm

Jonathan H posted:

jrothwell97 posted:

AFAIK the background was not blue-or-green screen, but the insets were obviously achieved by CSO.

What I suspect is that the background of the three globes existed in reality. This is only speculation, but I think that the insets were achieved by using an empty shot of the studio.

The image of the presenter sitting in his chair goes to a computer, which already has in its memory a shot of the studio, in the same configuration and with the camera in the same position, but without the newsreader, the chair or the desk.
The computer searches for parts of the live image that are the same as in the pre-existing empty shot, and keys these out (essentially cutting out these parts of the image).
The computer overlays the live shot over the empty shot. It then adds the inset over the empty shot.

This is probably overcomplicating things, and it would probably need incredible computing power to do this frame-by-frame 25 times a second for half an hour, but it could have been done that way. Either that or the background was just a green screen.

I don't know where to start! You're missing some fundamentals of how keying worked then and now. And yes, you're massively overcomplicating it!

Indeed, from that Engineering Bulletin it wasn't half as complicated, and sometimes went wrong, causing people's shoulders and/or arms to disappear behind the overlaid graphic. This is the relevant bit of the bulletin (pic and explanation):

http://www.rp-networkservices.com/tvforum/uploads/newsinsert.jpg

The pic on the left shows presenter with graphic insert. On the right is the mask signal required to control the keying. The grey area allows CSO keying where the presenter's shoulder is likely to overlap the graphic. The black area forces the graphic through, regardless of the backing, and the white area forces the presenter and backing through.

Obviously this is the reason why there were blue elements in the backing.

MD

mdtauk 15 April 2007 2:15pm

NG

noggin Founding member 15 April 2007 6:21pm

jrothwell97 posted:

AFAIK the background was not blue-or-green screen, but the insets were obviously achieved by CSO.

What I suspect is that the background of the three globes existed in reality. This is only speculation, but I think that the insets were achieved by using an empty shot of the studio.

The image of the presenter sitting in his chair goes to a computer, which already has in its memory a shot of the studio, in the same configuration and with the camera in the same position, but without the newsreader, the chair or the desk.
The computer searches for parts of the live image that are the same as in the pre-existing empty shot, and keys these out (essentially cutting out these parts of the image).
The computer overlays the live shot over the empty shot. It then adds the inset over the empty shot.

This is probably overcomplicating things, and it would probably need incredible computing power to do this frame-by-frame 25 times a second for half an hour, but it could have been done that way. Either that or the background was just a green screen.

The desk configuration probably remained the same throughout programmes, but it was probably possible to key out people who weren't being talked to/interviewed at the time by simply using an existing shot of the studio (as described above, but with the empty shot over the live shot).

This would require pre-configured camera angles though with no opportunity for even the slightest deviation.

Sorry - far simpler than that.

Straightforward CSO technology was used with a simple twist. The positioning of the real globe on the real background and the positioning of the camera shooting the presenter were arranged so that the presenter's shoulder co-incided with a blue, rather than yellow, bit of globe. There was therefore a CSO key available for the area where you needed to key the foreground bit of the presenter over a graphic.

Using a clever shaped mask, you were able to arrange for the CSO only to be used in the areas it had to be (where the background element was helpfully blue!), meaning the areas of the camera shot where no keying for presenter/inset separation was required could be any colour you liked (in this case yellow) Each (or most) cameras appeared on the vision mixer in the news gallery already keyed - so that each camera could be set-up and hot-cut between (within reason)

Effectively the mask that accompanied each graphic/camera shot either forced the background graphic, the foreground camera, or the CSOed combination to be used. I've used the black/grey/white keying elsewhere - it can be a useful technique for carrying two keys in one signal.

Clever bit of lateral thinking.

The graphic design of the CRV rendered graphic elements, the DVE tracked to the CRV for animation, the DVE "bevelled edge glass" wipes, and the design of the inset graphics, were cleverly harmonised, so that lots of different bits of kit and techniques could be used to achieve something that appeared to be a single, simple look.

HA

harshy Founding member 15 April 2007 7:34pm

noggin posted:

jrothwell97 posted:

AFAIK the background was not blue-or-green screen, but the insets were obviously achieved by CSO.

What I suspect is that the background of the three globes existed in reality. This is only speculation, but I think that the insets were achieved by using an empty shot of the studio.

The image of the presenter sitting in his chair goes to a computer, which already has in its memory a shot of the studio, in the same configuration and with the camera in the same position, but without the newsreader, the chair or the desk.
The computer searches for parts of the live image that are the same as in the pre-existing empty shot, and keys these out (essentially cutting out these parts of the image).
The computer overlays the live shot over the empty shot. It then adds the inset over the empty shot.

This is probably overcomplicating things, and it would probably need incredible computing power to do this frame-by-frame 25 times a second for half an hour, but it could have been done that way. Either that or the background was just a green screen.

The desk configuration probably remained the same throughout programmes, but it was probably possible to key out people who weren't being talked to/interviewed at the time by simply using an existing shot of the studio (as described above, but with the empty shot over the live shot).

This would require pre-configured camera angles though with no opportunity for even the slightest deviation.

Sorry - far simpler than that.

Straightforward CSO technology was used with a simple twist. The positioning of the real globe on the real background and the positioning of the camera shooting the presenter were arranged so that the presenter's shoulder co-incided with a blue, rather than yellow, bit of globe. There was therefore a CSO key available for the area where you needed to key the foreground bit of the presenter over a graphic.

Using a clever shaped mask, you were able to arrange for the CSO only to be used in the areas it had to be (where the background element was helpfully blue!), meaning the areas of the camera shot where no keying for presenter/inset separation was required could be any colour you liked (in this case yellow) Each (or most) cameras appeared on the vision mixer in the news gallery already keyed - so that each camera could be set-up and hot-cut between (within reason)

Effectively the mask that accompanied each graphic/camera shot either forced the background graphic, the foreground camera, or the CSOed combination to be used. I've used the black/grey/white keying elsewhere - it can be a useful technique for carrying two keys in one signal.

Clever bit of lateral thinking.

The graphic design of the CRV rendered graphic elements, the DVE tracked to the CRV for animation, the DVE "bevelled edge glass" wipes, and the design of the inset graphics, were cleverly harmonised, so that lots of different bits of kit and techniques could be used to achieve something that appeared to be a single, simple look.

It's got to be said the viewers at home including me all thought it used virtual technology and even today looks mighty impressive, I loved the orchestral music of the time, didn't BBC World use a virtual set at this time, I think their virtual desk was even massive!

JR

jrothwell97 15 April 2007 9:02pm

So if the presenter moved his shoulder to the right, it would disappear?

(April 2007)