Since its inception we have been aiming towards making V02 stereoscopic. It wasn’t long after a few discussions about equipment prices and technical challenges that we decided to drop this entire ‘3D’ thing and save ourselves a whole load of pain. Then we got back onto the idea, forgetting about this entire ‘pain’ thing and with both eyes converged we started to churn through a lot of possible solutions and throwing around a few comprimises.
This is what we want:
Stereoscopic Fulldome VJing.
Here’s what we have:
Stereoscopic = No problem.
Fulldome = No problem.
VJing = No problem.
Stereoscopic Fulldome = No Problem
Fulldome VJing = No Problem
Stereoscopic VJing = PROBLEM… and thus
Stereoscopic Fulldome VJing = PROBLEM
First of all it should be said that stereoscopic VJing IS possible, but not practical within the limitations we have set.
- Comfort. The stereoscopic quality is paramount to the wealth of experience. I’ve seen too many people dismissing stereoscopy as a valid creative tool based on a single bad experience from it to even consider presenting an audience with anything below perfect. Some people will accept a certain level of vertical disparities or divergence but these things can lead to a painful experience for many people.
- Resolution. The image we are projecting is big and wide and the viewer is also only a couple of meters away from the projection surface. Therefore the resolution feels much lower than it would if you were viewing it on a television screen. As a base level for a good experience we’ve set HD 1920 x 1080 as our lowest resolution.
- Multi Layered. This is most important for a good interactive experience. To be varied and responsive gives the viewer a real sense of creativity when playing with the beast. If it ain’t fun… It just ain’t fun. We’re not doing this for the high art Gods of apathy.
- Playback. The thing needs to playback at a decent rate. It can’t be too juddery with normal use or it will ruin the experience. Things will probably get a little slow when someone presses everything at the same time but then that’s what you expect to happen when you do that really.
- Brightness. It needs to be bright enough to illuminate even the darkest souls.
- Worthwhile. It needs to be worthwile doing. The 2D version of V02 would be so incredibly rich and colourful that it is only really worth doing if it is a genuinely good experience. Otherwise we can revert back to the 2D version.
- MONEY! Yes, money is an obvious limit. We don’t have a lot.
So what have we tried? Well, what haven’t we tried?… Seriously though, if you have any ideas, throw them in the comments section at the bottom of the page.
We’ve tried basically layering anaglyph footage in Modul8 or VDMX. A single video for each layer with an alpha channel means that any object will have black background in the areas the object is not seen in the other eye. Basically it has a black shadow either side of it. (limit 1)
We’ve tried layering anaglyph footage with additive blending modes. This produces luminance issues leading to ghosting artifacts in both eyes. (limit 1)
We’ve tried an over-under approach to the video frame and doing all mixing and compositing prior to applying an anaglyph effect in VDMX using a QuartzComposer plugin. This results in slight delays between left and right eyes causing vertical disparity issues (limit 1)
We’ve looked at piping the under-over image from Modul8 into VDMX using syphon to reduce the delay by handling the footage and anaglyph split separately. This didn’t fix anything. (limit 1)
We’ve tried piping the footage from modul8 which is handling the compositing of footage to MadMapper to split the top and bottom halves apart to VDMX, which adds the anaglyph effect. This also had no effect on the delay. (limit 1)
We’ve tried piping the footage from modul8 to VDMX by splitting the output into two separate sources and then applying the anaglyph effect in VDMX. this seems to work but is a bit juddery using the old Mac Book Pro we are testing with. (limit 4) This maybe fixed with a faster machine.
We’ve looked into piping the image from one machine to another using a capture card like a black magic. The card prices for capturing 4k are actually in the realms of affordability but HDMI Spec 2.0 isn’t available for Macs yet, so the maximum output would be 1920 x 1090 rather than the 1920 x 2160 we would require. (not possible)
We’ve looked into piping the footage into another machine at squashed HD under-over and then using stereoscopic player’s live input to render the anaglyph frame. This would make our effective resolution 1920 x 540, which is less than HD. (limit 2). Although this would also open up the possibility of piping the footage into a PC with a 3D graphics card capable of outputting active stereoscopic rather than anaglyph. This may then cause a breach of limit 5 and/or limit 7.
We can play only consecutive clips in stereoscopic no problem. (limit 3)
We can play only background elements in anaglyph as positively converged 3D and then have 2D elements appear on the screen plane. This will work for footage with an alpha channel but causes issues for anything with additive blending. We will also be limited to using greyscale tones as colour will breach limit 1. Nevertheless… We’ll be testing this out next week but I have a feeling we’ll be stepping quietly over limit 6 with this approach. Concerning this approach, Aron posted me and Robbie some queries yesterday that inspired the title of this post:
Good morning boys,
Pardon my naivety but I’m trying to get my head around this entire not working thing, there’s a large part of me tempted to leave you guys to it and trust in your wisdom but I wouldn’t feel right if I didn’t ask the obvious questions before spending lots of dollary-doos on something else that might not be the answer.
The anaglyph background thing not working with 2D elements overlayed: Is it to avoid the red and cyan not clashing with the new elements? For example, I’ve attached a child’s imagining of a 2D Superman punching a 2D giant banana over a 3D anaglyph Paris. The banana itself rests peacefully over the hustle and bustle of the Parisian streets where as Superman’s costume does cause a bit of eye wobble. Rather than grey-scale would just avoiding colours near red and/or cyan work?
I’ve also attached a picture of some boys in an anaglyph woods. The black and white boy does stand out better than his friends, so would going greyscale for the frontal features be so bad?
I know nothing about 3D, but I have to try God damn it. Please find attached a 2D Captain Kirk trying to understand an anaglyph plane engine.
Aron’s brilliance knows no bounds.