Recreating the look of tube cameras

[Claire Buckley]

Originally Posted by Robin Davies-Rollinson

I was a BBC lighting cameraman during the seventies and have to cringe when I see some of my footage on dramas today on archive programmes.. Funny that at the time, the film inserts never looked as bad as they do now...!

In the 70s inserts were at 16mm played in on a Rank Cintel Telecine (where I was) via studio then on to VT (2inch quad). Problem was the small amounts of playback mechanical stability in the telecine gate resulted in flutter-like jitter which, inserted into video, stood out noticeably. Similar too the noise (film grain) in some scenes.

In response to the OP: BBC are running a series called The Hour (on tonight I think in the UK) - I think they've recreated some plumb tube shots. Crushed highlights, large depth of field and a washed out looking chromaticity.

Gary N is correct, the "look" was fairly flat and "hot" due to the lighting. Keys were predominantly 10Ks. If I can remember levels of 150-180 ft/candles (old term).

EMI 2001s - crikey, that takes me back...

:)

[Claire Buckley]

Update: The Hour is set in the 50s so it won't be plumbicons then - my error. I thought it was a drama in the 70s. Sorry.

:)

[Brian Drysdale]

More likely image orthicon in the 1950s.

You must've been working in the big TV studios for the 10ks, the smaller TV studios would've been more 2ks, with the odd 5k.

I guess with a f2.8 a f4 split at 150-180 ft/candles, the EMI 2001 would've had an ISO of about 100.

[Claire Buckley]

Yep, that's right Brian - I believe at the time it had two of the biggest LE (Light Entertainment) Studios in Europe.

Not sure how this converts into DSLR speak? In racks you just pushed and pulled the iris quadrant paddle gizmo along with the black level rotary twiddle thing - using the push to monitor dooby. Let the camera guy frame and focus - team work!

:)

[Seymour Clufley]

David must be right about the lighting not contributing to the tube camera "look". Obviously it contributes to studio footage (overhead lighting of actors and sets) but since this doesn't apply to exterior filming, which still has "the tube look", there's obviously something else at play.

Likewise DOF can be discounted.

As for comet trails, I'm not bothered about them. They're a decoration that can be added afterwards (or so I'm reliably informed!) by hand.

Highlights overloading into pure white is not something I'm bothered about either. There are pretty artefacts and annoying artefacts!

Which leaves us with the basic, generic look of something shot on a high quality tube camera. (I don't want it to look like old camcorder footage!) Perhaps, after all this discussion, the basic look could be achieved simply with grading?

This is why I asked at the start how people would describe tube camera footage. It may be better to look at the end results rather than the building blocks that go into making it.

• It's not saturation. Some tube footage is very vivid.
• It's not a lack of darkness. Some tube footage is very dark
• It's not grain or noise. Some tube footage is very clean
• It's not depth of field, for reasons described by other people. Indeed, a lot of footage I've been watching shows a pretty wide depth of field.

So what does that leave us with? I still don't know where to begin!

As I said before, I've replicated the process in After Effects. R,G and B layers (shrunk to 25% then enlarged again, to eliminate detail) then the luminance layer on top (not resized, so full sharpness). It doesn't have "the look". I've experimented with tweaking it - overlay a black layer with Soft Light blending, same with grey, etc. - but the tube look just isn't emerging!

[Seymour Clufley]

Originally Posted by David Heath

And now I find the electronic exteriors far preferable to those done on film - the film ones just look dull.

Agreed. I much prefer video to film in those 70s/80s programmes. For example, in Blake's 7, exteriors are fine on film because you expect it, but occasionally there is studio stuff shot on film and it looks really nasty.

However, the Doctor Who Restoration Team have achieved great results by scanning in 70s 16mm film on modern telecines. It does look a lot better.

[Bob Hart]

A grubby method of getting "trails" in post is to stack the same clip in several layers on the timeline with each clip one frame late and reduce the transparency of each overlaid clip to a fraction of 100% which you calculate from the number of clips you decide to stack, eg., 5 clips, the four ovelaid clips each at 20% transparency.

You have to play with the levels in effects on each of the clips. It is really timeconsuming. The following clip is not a tube camera emulation.

http://www.youtube.com/watch?v=VUStQzrirZA

Whilst for this clip, I used the delayed layer method for video noise reduction of an old VHS copy of a VHS recording out of a deceased estate, if you look closely, you can see the smear effect on highlights in the water movements. To reproduce the smoothness of interlace cameras, my person preference would be with my EX1, to shoot at 50P 1280 x 720, light for contrast, let highlights blow out, then in post, use frame blend, 5 layer stacks and contrast effects to get the laggy look (trails), export that clip, then re-import it, change the speed and select frame blend "on" to restore the normal frame rate.


A hypothetical.

In absence of using a real studio tube camera of the times, another very awkward way to treat modern footage would be to reproduce it to a modern camera through a night-vision intensifier from a high defninition display, Electophysics Corp Astroscope likely the best.

Night-vision is the last remains of tube based image technology but likely to fade away as solid state devices have now emerged. It has the blowout and overload characteristics although there is some modern improvement to minimise this.

You need to be careful to keep light levels really low with very heavy ND filtering, infrared filtering to keep out the infrared light to preserve more sharpness ( techie thing of long explanatio ) and lens iris settings.

Overloading the tube will damage it.

Record the playback of the clip three times, each with only one colour channel permitted by using filter gels between the camera lens.

Each clip will be monochrome green which is the colour of night-vision tube display to the camera. Desaturate each of the recordings to remove the green, then apply colour tint in effects to restore to each its colour information. Stack the layers without any delay between them. Apply fractional transparency to the two overlaid clips. Levels and contrast effects would be needed.

No warranty of satisfactory performance is made with express or implied - hypothetical only.

I have delay stacked monochrome night-vision footage to reduce display tube noise and the effect is very much like the original tube-camera smear.

[Brian Drysdale]

Originally Posted by Seymour Clufley

However, the Doctor Who Restoration Team have achieved great results by scanning in 70s 16mm film on modern telecines. It does look a lot better.

Well shot 1970's 16mm film has a lot of information that the old telecines couldn't extract.

The EMI 2001 had an interesting method of creating its image.The luminance tube was used more for sharpness and fine picture detail, rather than the luminance.

EMI 2001 - Wikipedia, the free encyclopedia

The look can vary from programme to programme, LE tending to be more vivid looking than the serious dramas.

[Robin Davies-Rollinson]

The main "look" for me from old tube cameras is the ringing on edges around subjects. I seem to remember it being white / black depending on the scan direction, so the left edge - say - might be white and the right one black.

[Seymour Clufley]

Originally Posted by Robin Davies-Rollinson

The main "look" for me from old tube cameras is the ringing on edges around subjects.

I've heard about this before, and I think I can identify it in, say, this image.

But can you tell me where the ringing is in this image?

[Bob Hart]

My recall of "ringing" was in B/W television, not colour but here in Western Australia, we were late bloomers for colour TV, 1975 ?? where I was living at the time. As for Blake's 7 and the other good productions of the era, I would not have noticed any ringing or other problems as I was too much into the storystuff in them.

[David Heath]

Originally Posted by Seymour Clufley

I've heard about this before, and I think I can identify it in, say, this image.

No - what you're seeing there is a registration error, a discrepancy between the way the scans of the 3 (or 4 ) tubes are aligned.

But can you tell me where the ringing is in this image?

It's a bit too small an image to see much online, but it doesn't really appear to show very much. The effect Robin refers to is on such as a vertical edge between such as a nearly white block (say 90 IRE) and a dark grey one - you'd typically see (left to right) plain 90% white, brief 100% white line, followed by brief 0% black line, then plain 10% dark grey. The effect is as if a line had been drawn round high contrast objects.

But it was adjustable, and on location drama deliberately taken to quite low settings. It would typically be set to higher levels on such as LE or sport to give a more "punchy" look.

Originally Posted by Seymour Clufley

* It's not saturation. Some tube footage is very vivid.
* It's not a lack of darkness. Some tube footage is very dark
* It's not grain or noise. Some tube footage is very clean
* It's not depth of field, for reasons described by other people. Indeed, a lot of footage I've been watching shows a pretty wide depth of field.

So what does that leave us with? I still don't know where to begin!

Have you tried what I suggested quite a long time ago? Split the luminance and chrominace images (to give Y,U,V) then apply a soften filter to the chrominance (U and V), and recombine. The effect is a still sharp luminance image, but with the colour looking a bit as if painted in, a little like a childs colouring book. Just don't overdo it.

[Seymour Clufley]

Originally Posted by David Heath

Have you tried what I suggested quite a long time ago? Split the luminance and chrominace images (to give Y,U,V) then apply a soften filter to the chrominance (U and V), and recombine.

Yes, I did this in After Effects using R,G,B,Y. Some test images were posted a few days ago now.

However, reading your suggestion again, I thought about doing it with the YUV model as you state. (Thanks for reminding me about YUV!)

AfterEffects can't seem to work in YUV - at least, not to do what we need - so I wrote a program instead. Here are some results:

Oliver Stone

Bill Nighy

In each case, the images are...

LEFT: raw image
MIDDLE: YUV remixed (as David describes)
RIGHT: YUV remixed + chrominance noise added

To be honest I'm quite pleased with these results. It's surprising how much more "convincing" it is with the noise added. Clearly, though, the noise is only "first version" and can be improved. The dots need to be smaller, for starters.

Does anyone have any other ideas? Should the noise be less colourful? Should it be confined to (or exaggerated in) dark areas?

[Brian Drysdale]

I'd keep any noise to the dark areas, it tends to look digital and I don't think the video engineers of the period would've accepted that level of noise on faces.The noise would've been random rather than fixed pattern.

[Seymour Clufley]

Brian, the noise in my example is completely random. Can you be more specific? Does it look like it's fixed pattern?