[Tig] Silver Screen Butterfly Test

Richard Kirk richard at filmlight.ltd.uk
Fri Aug 6 15:33:55 BST 2010

Hi, Tigs.

I had a post under the recent Silver Screen Nightmares thread where I 
suggested we might try dong a butterfly test to see if there are any 
nasty psychophysical effects that might make silver screen images look 
different to unity gain screen images.

I have just done a silver screen butterfly experiment at MPC London. 
They had a Harkness 240 silver screen and a regular unity gain. By 
rolling the screen half-way up we could get the top half unity gain, and 
the bottom half silver. I made a set of test images flipped about the 
middle. Then, using the screen save and wipe modes in the Truelight 
viewer, I was able to apply a different cube to each half.

We started off with a straight projection on both halves. I had to 
reduce the brightness on the silver screen by about 10%. It is not clear 
where the 'hot spot' was but simple geometry puts it in the top half. 
The brightness match depended on where you sat, so we picked on the 
front row in the grading theatre. We stuck in a tiny white point shift 
(just over 1.0 deltaE). The darks looked a bit darker on the silver 
screen (we may get less flare from ambient light when the screen gain 
goes the other way) which was compensated for by setting the contrast to 
1.004. These shifts would be down in the noise if we were comparing the 
projector to film.

Do we see any of these psychophysical effects when looking at a silver 
screen? I had to adjust the luminance down by only 10% - that was 
probably because the silver was below the white, and our chosen observer 
point could not see the hot spot. The other edits were tiny, and would 
be down in the noise if we were comparing the projector to film.

We then drove the projector as for 3D ande put on the RealD glasses. The 
silver screen image looked green. If you put your head on the side it 
looked pink. If you took of the glasses and rotated them, the image went 
from green to pink: at about 45 degrees the glasses gave the right 
balance. This colour shift changed sightly with seating position. 
Meanwhile, the image on the white screen was constant.

If the RealD glasses give the right colour balance at 45 degrees, why 
don't they make the glasses with the polarizer rotated by 45 degrees? I 
guess they don't because most people watch the screen with their heads 
mainly vertical, but may nod from side to side a bit. In this case, 
picking the point where the colour shift passes through a maximum will 
give a minimum colour shift if you tilt your head by a few degrees.

The final sanity check was to match the projection on a sliver screen to 
a CRT, and then switch to 3D. The obvious differences were in 
brightness. If you adjusted the brightness by either making the CRT 
dimmer, or allowing the 3D image to clip then the visual differences 
were small. If you had the lower light level (3.5 ft-L or 5 ft-L with 
glasses on instead of 16 ft-L on a white screen) then some colours 
looked different, and you saw less of the shadows. If you turned the 
Truelight flare correction to about 5%, then this restored some of the 
shadow details without messing the saturated colours too much.


If you drop the light level from normal cinema levels to typical 3D 
levels then the loss in brightness will cause us to see colours 
differently. The neutral balance is unchanged but colour shifts are 
complicated: I don't know of a good model. The tone curve is changed, 
but a typical flare correction of about 3% can restore the visibility of 
shadow detail: in this case the 'flare' is perhaps measurement noise in 
your retina rather than actual light spill.

If you swap your white screen for a silver screen with modest gain, the 
average light level for the best visual match does not change by much. 
We are probably integrating the screen gain over the whole image, and as 
the screen is not making or absorbing light, the average screen gain 
will lie closer to 1.0 than the peak gain figure. There seem to be very 
few psychophysical side-effects from looking at a silver screen rather 
than a white one.

There is no substitute for the light we have lost with 3D. However, the 
other changes are quite cheering. If we adjust the white point until we 
get a visual match, then the silver screen does not introduce any other 
significant problems. You can increase the shadow contrast as for a 
flare correction, and this can correct the neutral tone curve. If you 
know the 3D projection is going to be dim, it might make sense to grade 
the 3D version of the film, or tweak the tone curve of the projector 

The 3D light levels are still somewhat pants, but things could be a lot 

Richard Kirk

FilmLight Ltd.  	Tel: +44 (0)20 7292 0400 or 0409 224 (direct)
Artists House,		Fax: +44 (0)20 7292 0401
14-15 Manette Street
London W1D 4AP

More information about the Tig mailing list