Test #2 – Gallery Test – The Results

The tests on the whole went well. Here are my observations categorized as described in the previous post (planning).

Operational Issues


The 20m USB extension was long enough to connect the furthest webcam to the dev machine in the control room. Now I need 15 m and 10 m extensions to wire up the remaining 2 webcams. MIDI and power routing was all fine. The mini-display port to VGA adapter also worked fine. The venue has plenty of VGA cables. Audio via the mini-jack taken directly as an analogue out  from the audio machine was very noisy. I had anticipated this and had planned to bring a digital mixer (to which the audio machine can connect optically) and jack to XLR cables to connect directly to the venue’s XLR stereo input. However, I didn’t bring up the mixer for the test due to lack of space and hoping it might be unnessary. It seems that it will need to come up for the duration of the exhibition. In an ideal world I would connect the audio machine directly to the venue sound system via optical and thus avoid  having to find room for another item of kit and associated cables.


With the help of Richard Fair (the venue manager), I tried many permutations of lighting pattern and curtain placement to achieve optimum conditions. It seems that the best arrangement is to have the curtains drawn (to prevent light flooding in from the entrance) and the recessed ceiling lights up to about three quarter strength. There are a number of suspended LED lights which provided a small level of additional side fill but I found these distracting. I tried floor lights but also found these too much of a distraction to the eye. The recessed ceiling lights (presumably halogen bulbs) are capable of whitewashing the entire screen if set too high, so it’s a matter of balancing the need for overhead light, critical to the luminance difference technique (discussed previously), with the improved visibility of the screen in subdued light. More tweaking will be required in the final set-up but I am fairly confident that an adequate balance can be struck. One downside with the ceiling lights is that they appear in the captured image (ie they are ‘seen’ by the webcam) but this is unavoidable given that the ceiling is low and the lights are fixed with no provision for diffusers or similar to be set.

Webcam placement

The optimum placement was approximately 80cm above the floor (mounted on a mic stand). Any higher and the webcam captures the projector light itself, which must be avoided. One downside is a slight image distortion (acceptable to my mind) but more off-putting is the fact that when you stand directly in front of the webcam at the ‘sweet spot’ (the closest point at which the whole body is seen) you appear to be looking up as the webcam is capturing from below the line of sight. This won’t matter so much if the final image is non-realistic and the eye detail is effectively obscured. More on final image quality below. With Richard’s help, I also tried the webcam at near-ceiling height looking down. Again there was image distortion but no ceiling lights and critically no projector beam incursion. My feeling was that the image distortion was too great – in effect a position of 80cm above the floor is closer to average eye level than the venue ceiling height (approximately 3 metres). Webcam mounting is a new topic for my to-do list. I had thought of using 3 identical mic stands but theses are quite clunky being designed to mount heavier items. I’m hoping to either fabricate or adapt a stand that provides stability at 80cm while presenting less surface area and therefore less distraction eg a thin rod mounted on a flat shoe rather than a tripod. My first line of enquiry will be hifi speaker stands as I’m sure I have seen something along these lines.

Resolution and Framerate

The maximum resolution detected by the dev iMac when connected to the venue system was 1920 x 1080 and this seems to be the optimum size. I did try lower resolutions  scaled up to fit the screens but the image quality seemed to drop (unsurprisingly). One downside was framerate which dropped significantly to around 10fps even with a simple webcam test. The framerate did not improve noticeably by reducing resolution and is probably an intrinsic attribute of the set-up. This is frustrating as I have spent much effort trying to maintain 15fps and above in the work to date. I can at least re-calibrate the work to fit around a reduced framerate ie hopefully improving image quality.


There is a semi-circular pattern in the carpet and it so happens that the outer edge of this pattern coincides with the ‘sweet spot’. My plan is to run hi-vis gaffer tape along this line and add an arrow for the centre point of each screen ie the point directly in front of the relevant webcam. The webcams themselves will have boxes marked around them on the floor. These measures are informative rather than preventative and I hope that the majority of visitors to the exhibition will understand and abide by the demarcation. There will normally be an invigilator in the room, so human intervention in the case of visitors openly flouting the demarcation to the detriment of the piece will hopefully be achievable in as  polite a way as possible.

Critical Techniques

Live / Quartz Composer sync via MIDI

The technique essentially worked – Ableton Live was able to instruct Quartz Composer to change between vignettes while all the time accepting and responding to incoming MIDI information generated by viewer presence. This was a 1 :1 experiment (audio to visual) whereas the final will be a 1:3 and this will require a little more thought and experimentation to perfect.

Luminance frame difference as a basis for identifying presence

This key technique worked although will certainly benefit from further lighting optimisation.

Webcam focus

The Agent v6 webcam features a manual focus ring which is one of the reasons that I have chosen to use it for this project. However, it was difficult to be precise about focal range and I’m now thinking about producing a focus chart that I can place at the sweet spot in front of each webcam in order to adjust focus more efficiently

Aesthetic Results

The Slit-Scan piece didn’t work as well as the February Slack Space test (performed in front of a large-scale art work) due to the mundane nature of the background ie the venue entrance shown behind the viewer. I did try radical adjustment to the lighting  and even the addition of a footlight pointing back at the viewer, but none of these options worked. The answer may be to come up with a non-realistic version of the Slit-Scan piece – but this approach has the potential to become confusing to the viewer. I will experiment further but am prepared to discard this technique from the final selection if necessary.

The Rutt Ettra-based test (which looks amazing at close range on a computer monitor) just didn’t work at scale. The scanlines are too fine and the overall impression is confused.  It is good to know this now rather than later. I will probbaly discard this type of technique from the final selection.

Both the Neon Dance and Sparks pieces worked well if only so after substantial tweaking, largely to accommodate venue lighting and playback framerate conditions. This really is good news as these are the 2 pieces I have worked up the most and had the most hopes for. I’m sure I can optimise the aesthetics of both of these pieces with further tweaking in situ during set-up. The last major concern is that they may not work as well on the NUA machines to be used for the final piece which are slightly less powerful than the dev machine. I will be able to access these machines from June the 13th onward.

Here is fairly low-res recording of me calibrating Neon Dance and Sparks during the test.

Next Steps

Are many! I have 3 weeks to turn around the exhibition and am currently scheduling my time. Tomorrow I intend to write a press release and kick-start my promotion plan – more on this as it happens.

Leave a comment

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s