The unique thing about this is that it uses HD video, and not one, but two monitors, plus a sheet of beam splitter glass to create a reflection that gets folded back in to the image.
It's a delicate art to operate the device, an interplay between the camera and monitors, the position of the monitors, and the monitor control dials (hue, saturation, brightness and contrast). Doing controlled feedback like this requires these control dials, but most HD TVs and monitors don't have analog knobs like old CRT TVs did, making it difficult to create controlled feedback in HD. [...]
All the images in this video are created by video feedback only - no computers are involved. The upper and lower monitors both display the same thing - the image from the camera, which is looking at the upper monitor. This creates a video feedback loop (much like a microphone next to a speaker creates an audio feedback loop).
The Light Herder
Tags: art, blobs, glitch, mpegs, retrocomputing, xscreensaver
21 Responses:
I would _so_ buy tickets to see that being played live.
Preferably with some wierd ass noise art band improvising along to it.
I'm pleased it didn't go away with analogue television.
Well they used a Thinking Fellers Union Local 282 track for the upper right video. Not exactly noise but hella alt.
It looks like this technique leverages heterodyning between the frame rates of the hdmi monitor and the camera. I’m sure they’re both within spec but just a little off to make things interesting. Further analog tweaks could be done with careful application of a chill spray on the electronics to slightly alter the frame rate.
There's no way that altering the clock via temperature would result in a different HDMI frame rate, it's a digital transport. If it went too far out of spec you'd just get a blue screen and a multi-second EDID re-negotiation.
This guy has done amazing work but the downside to doing feedback with digital sensors, transport, and screens is that each of those stages introduces some integer-greater-than-one number of frames of delay. Video feedback using analog sensors and tube screens had a much tighter timing loop and looked way more organic.
Yeah, I don't know what temperature would have to do with anything. I'm actually introducing delay on purpose using a Blackmagic converter box. The processor on the box can be turned off resulting in much less delay. There are several HDMI to SDI and SDI to HDMI converter boxes that introduce a very small amount of delay.
I could create more traditional looking feedback with this setup, but that is not the look I'm going for. As not looking organic - I think this stuff actually looks much more organic than any other video feedback I've ever seen. For an example, take a look at https://vimeo.com/446318435. I've posted this in another comment, but if you'd like to see what's going on with the device, check out https://vimeo.com/508776650.
Yes, those for sure look organic in that they look like crazy deep sea creatures or something! But it's also very sharp and... precise looking, I guess? Whereas tube-based feedback just felt more analog, less CGI-adjascent -- that's really what I meant by organic. Not to disparage your work at all, I think it's incredible! But I also have a soft spot for the noisy mess of the tubes.
Last month I set up a video feedback instagram trap at the club -- nothing complicated, just a giant LCD TV and an HDMI camera pointed at it -- and I was frustrated at the long delay due to the various HDMI components in the path. In an early iteration I tried using a vintage-2004 NTSC camcorder, through a composite-to-HDMI converter, but that didn't bring anything interesting to the party, it just made it all blurrier.
Uh, yeah, it's sharp...and precise looking. That's what the project is all about. You like tube-based feedback because it "looks more analog" (whatever that means), great. Not the point of this project.
That being said - I do appreciate you posting this and that you like the work...
As for tube-based feedback - I think the only thing different you'd see from chip-based camera feedback is a bit smearing of the image, as tube cameras are famous for. I don't think there would be any difference as far as delay inherent in the fact that it has a tube. Maybe some cameras have more of a delay in their HDMI outputs than others, or maybe HDMI outputs have more of a delay in general than cameras with analog outputs (tube or chip based) - this I don't know. But, I do think your video feedback wall that people can move around in front of is pretty cool, and the delay is an enhancement, not a determent.
*detriment
Any HDMI device that does A/D, D/A, or frame processing is going to necessarily introduce 1 or 2 frames of delay, because it has to buffer the whole frame before it can manipulate it. So a camera, an effects box, and a TV and you've already racked up 1/10th to 1/5th of a second of debt. If there's an actual computer involved, now you add on whatever buffering is introduced by HDMI-to-Lightning or whatever, plus whatever buffer is in the app and/or its underlying ffmpeg invocation. Whereas I think that in the NTSC world, the delay of the whole system all the way through was probably 1/60th second, one interlaced field -- the feedback was truly racing the beam.
The real-world consequence of these unavoidable delays is that, for example, in my video wall, that 1/10th to 1/5th second delay turns into inches of displacement: since I don't have my TV on a gimbal like yours, I end up with distinct stripes instead of blooming smears.
One isn't necessarily better or worse (I mean, the mission of any feedback toy is "let's push this thing past its limits and see what happens"), but having things traverse the digital domain for sure introduces constraints that didn't exist in the analog domain.
And of course there are upsides too, like, finding a 75" CRT with a thousand scanlines is uh.... probably not even a thing.
Glad you appreciated the use of Thinking Fellers Union. If you'd like to see what's going on with the device, see the schematic, and what's up with the delay, check out https://vimeo.com/508776650.
Oh, also, it's not just Thinking Fellers - I mixed something else in starting at 4:20 :-)
Something else you might be noticing in addition to the delay in the system is that these videos are recorded at 60fps and I'm often editing and exporting them in 24fps or 30fps.
I kind of that more film-y look as opposed to the more video-y look. Here's one video in both 24fps (https://vimeo.com/446317570) and the original 60fps (https://vimeo.com/446318435). I go back and forth on which I like better.
I'd definitely like to create video with this at a live music venue with a video feed going to the HDMI input. Here's a bit of experimenting I did with a previously recorded music performance as input. This was very early on in the device's creation, and so much more could be done with video input than is done here: https://youtu.be/U00VMVkA99A
Ohhh, nice. Thanks gir the link.
This takes be back so far I’m showing my age here. I used to see these guys perform live with a real time video artist doing a combination of analog video effects using cameras (on the band or the crowd) as well as Commodire 64 3D animations. Usually in dingy little venues to a few dozen people, at least before this song got big:
https://youtu.be/MVMiFfdcKJM
Wonderful stuff, automating feedback. I think you could produce a signficantly cheaper but only slightly worse version of this with raspberry pi high quality pi outputting directly to HDMI, and also controlling stepper motors for the rotations and other movements. That would allow recording a performance and replaying it physically. It's not really computer-free at that point, but the DSLR cameras he uses with HDMI outputs aren't, either.
I appreciate your idea about automating, but it would be extremely involved, since many of the monitor's hue/sat/color/contrast knobs are tweaked ever-so-slightly during the creation of images, plus the height of each monitor, the switching of inputs on the two switchers with the foot pedals, not only the rotation of the rod, but distance pushed in or pulled out, etc. Also, I think the thing about this is that it's done live, and never quite the same way twice.
If you wanted to try anyway, maybe look at 3D printer or CNC mill driver boards controlling stepper motors?
Half stepping a 200 pole stepper gives you slightly less than 1degree steps, microstepping can get you down under 0.01deg per step.
(And if you use common 3D printer stepper motors, toothed pulleys and belt is easily available. Laser cut acrylic toothed pulleys to fit on whatever pot shafts you have would be easy and cheap. You could also choose to gear down from the stepper pulley so half stepping and 4:1 pulley ratios would give you 1/4-ish degree resolution on the pot shaft.)
Or, you know - you do you, and I’ll stop with the unsolicited advice. :-)
so... when's xscreensaver coming out? :)
In this house we click previouslies.
I don’t consider myself a pervert, but I instantly thought this was some kind of bondage device when seeing the video preview.