[Open Manufacturing] Video is not the point
Bryan Bishop
I've noticed a recent outbreak of video-centricism. While videos are
useful, I'm not sure it should be a main focus, for the reasons that I
explained more than a week ago. Nathan, what do you think about this?
I note that you've been mentioning videos frequently lately.
http://groups.google.com/group/openmanufacturing/msg/97a69098d77b34e9
On Feb 12, 11:10 pm, Bryan Bishop wrote:
> On Tue, Feb 10, 2009 at 2:18 AM, Nick Taylor wrote:
> > Imagine how difficult it would be to convey that information textually? All
> > that ambiguity etc.
>
> At the same time, imagine the difficulty in conveying that information
> visually. on the other hand, data structures, while not text, are a
> way of formatting text and codifying a system of the representation of
> types of information, even executional information for running on
> hardware (like programs for a computer). So I'm not sure video is the
> answer either- imagine if each one of your 14 trillion cells had to
> watch millions of videos before they could make their own parts and
> components :-). Yikes.
>
> > It would be quite cool to compile a whole Civilisation-Seed of these videos
> > - and though assuming that people have the technology is something of a
> > leap of faith... maybe it's less of a leap of faith than assuming that they
> > have the brains to understand the text version rather than the video
> > version. Because I know I struggle with text.
>
> Yes, text is pretty terrible. Video is pretty bad too. I am not sure
> that a videotapped civilization seed would be useful though. It would
> be better than what we presently have (comparatively little), but
> you'll have to convince me.
>
> > It would be an interesting project - I'd be up for any programming that
> > needs doing etc.
>
> Really? How about you and I talk sometime- either on the IRC channel
> (#hplusroadmap on irc.freenode.net) or over the phone, or over some IM
> protocol of your choice. There is a lot of programming work that I can
> give you related to these projects- it's kind of why I'm here.
>
> > As a total aside, this civilisation seed reminds me a bit of the occult
> > thing - where if you make your own instruments (daggers etc) you're
> > supposed to start from zero - mine and smelt your own metal rather than
> > buying it, and if you can make your own tools to make better tools etc...
> > then so much the better.
>
> What? Do you have some references on this "occult practice"?
>
> > I find this idea quite fascinating - an genotype for an entire civilisation
> > - which includes every point in the path from caveman to five minutes
> > ago.
>
> What's interesting about biological genomes is that not only do they
> encode information, but it's also functional or executional ("function
> is structure")- making for very high signal to noise ratio, which we
> aren't going to be able to see in text and video simply because these
> symbols that we communicate with are hardly 'grounded' to the reality
> of the fabrication technologies.
Nick Taylor
>
> I've noticed a recent outbreak of video-centricism. While videos are
> useful, I'm not sure it should be a main focus, for the reasons that I
> explained more than a week ago. Nathan, what do you think about this?
> I note that you've been mentioning videos frequently lately.
>
> http://groups.google.com/group/openmanufacturing/msg/97a69098d77b34e9
It's memes vs microformats.
They're not necessarily incompatible I don't think... but if you want an
idea or technology to propagate through humans rapidly, signs are that
video is the best way of doing this. The "I could do that too" factor is
so much stronger.
Building up to video/animation from a codified base probably isn't going
to happen... at least partly because so much of the innovation is being
communicated purely at the level of video. It's become a kind of oral
tradition again.
Someone adapts, puts it up on youtube... someone adapts that, puts that
up on youtube... which isn't to say there isn't loads of wordy stuff
going on as well, but I think the vitality of the revolutions we're in
at the moment are currently being expressed in video.
Bryan Bishop
> They're not necessarily incompatible I don't think... but if you want an
> idea or technology to propagate through humans rapidly, signs are that
> video is the best way of doing this. The "I could do that too" factor is
> so much stronger.
How about this. Once we have that technology, you go make all the
videos you want of it. Meanwhile, that technology is by far not going
to be 'video'. There's already multiple encoding formats if you want
to go focus on that, like MPEG, AVI, etc. MPEG is like JPEG- it's
lossy compression on hardware information, it's not going to be good
for engineering.
Paul D. Fernhout
That's a funny way to put it about how-to videos: "it's lossy compression on
hardware information"; funny, but true. :-)
Inspired by your suggestion, I agree that if we have manufacturing
information in a format which is machine readable, with metadata and a high
level of detail in 3D geometry of components, then we can in theory produce
instructional videos by recording simulated assembly sequences (including
with simulated people doing the simulated assembly). Although it won't look
as real anytime in the next few years, it might be good enough considering
you could get exactly the instruction video you wanted on demand, even if no
one made one before. For OSCOMAK, I had thought of producing paper books
with assembly directions on demand, but never before thought about
instructional videos:
http://www.kurtz-fernhout.com/oscomak/goals.htm
"One process to be included is a way to convert the high-tech computerized
library to a low-tech paper one as desired. Key to the whole endeavor will
be to present everything in a how-to fashion."
By the way, here is a web site where you can make your own videos by just
typing text, just as a sort-of proof of concept:
http://www.xtranormal.com/
"Text-to-Movie(TM) If you can type, you can make movies."
I just made one where an actor said "hello world" and pointed. So, it is
possible to generate videos from textual metadata quite easily now, even on
remote websites.
Obviously, it might take something more elaborate to make a video about,
say, a virtual actor replacing an automotive transmission, let alone a
virtual actor going through all the steps of Gingery's metalworking shop
from scrap. But it would seem quite possible eventually, with the right data
formats and simulation framework, even with just today's desktop computers
(and certainly easy on computers of twenty years from now). And you could
put in probabilities of things going wrong, too, like when the molten metal
gets poured on the floor, kind of like a OM version of the comedy series
"Home Improvement" where every repair usually goes wrong. :-) I'd suggest,
once our simulations get to the point where we're modeling accidents and
parts that fail and are made wrong, we'll have made great progress. :-) In
general, having data in a format that you can both use it for
simulation/visualization and automated assembly would make a lot of things
possible.
--Paul Fernhout
Vinay Gupta
Video is the only way Hexayurts replicate other than by direct
personal instruction. No written instructions or photosequences got
people over the "can I do that myself?" hurdle.
Written instructions also necessary, often used as checklists.
Vinay
--
Vinay Gupta
Free Science and Engineering in the Global Public Interest
http://guptaoption.com/map - social project connection map
http://hexayurt.com - free/open next generation human sheltering
http://hexayurt.com/plan - the whole systems, big picture vision
Gizmo Project VOIP : (USA) 775-743-1851
Skype/Gizmo/Gtalk/AIM: hexayurt
Twitter: @hexayurt http://twitter.com/hexayurt
UK Cell : +44 (0) 0795 425 3533 / USA VOIP (+1) 775-743-1851
"If it doesn't fit, force it."
ben lipkowitz
machines what to do. can we leave it at that?
Bryan Bishop
>
> Video is the only way Hexayurts replicate other than by direct
> personal instruction. No written instructions or photosequences got
> people over the "can I do that myself?" hurdle.
Probably because you didn't use a standard format for writing out the
information. It's in an image, of all things:
http://www.appropedia.org/Hexayurt_schematics
How is my instruction generator supposed to work on random images? :-(
Nick Taylor
I think we're talking at cross-purposes here.
Who is your audience?
Nick
Bryan Bishop
> Bryan Bishop wrote:
>> On Sun, Feb 22, 2009 at 5:02 PM, Nick Taylor wrote:
>>> They're not necessarily incompatible I don't think... but if you want an
>>> idea or technology to propagate through humans rapidly, signs are that
>>> video is the best way of doing this. The "I could do that too" factor is
>>> so much stronger.
>>
>> How about this. Once we have that technology, you go make all the
>> videos you want of it. Meanwhile, that technology is by far not going
>> to be 'video'. There's already multiple encoding formats if you want
>> to go focus on that, like MPEG, AVI, etc. MPEG is like JPEG- it's
>> lossy compression on hardware information, it's not going to be good
>> for engineering.
>
> information in a format which is machine readable, with metadata and a high
> level of detail in 3D geometry of components, then we can in theory produce
> instructional videos by recording simulated assembly sequences (including
It would also be useful to be able to pause and re-position the camera
to see what motions look like, or how two parts are really supposed to
snap together or otherwise connect, and so on.
> with simulated people doing the simulated assembly). Although it won't look
> as real anytime in the next few years, it might be good enough considering
I wonder how real it would have to look though. Reflections aren't
necessary for rendering, and shadows should not be a factor anyway,
right?
> you could get exactly the instruction video you wanted on demand, even if no
> one made one before.
Yes. It's one of those cases where technical experts know more than
they know others don't know, so they wouldn't think to make those
videos in the first place.
> Obviously, it might take something more elaborate to make a video about,
> say, a virtual actor replacing an automotive transmission, let alone a
> virtual actor going through all the steps of Gingery's metalworking shop
> from scrap. But it would seem quite possible eventually, with the right data
> formats and simulation framework, even with just today's desktop computers
Well, I always thought that this was basically what you meant when you
(or any of the others of us) were talking about writing a simulator.
> (and certainly easy on computers of twenty years from now). And you could
> put in probabilities of things going wrong, too, like when the molten metal
> gets poured on the floor, kind of like a OM version of the comedy series
> "Home Improvement" where every repair usually goes wrong. :-) I'd suggest,
The other week I proposed to a friend the idea of "Home Improvement"
but based off of an automation job shop instead. The idea is that if
you're going to repair a machine and make it "more powerful", you
should make it *significantly* more powerful. ;-)
> once our simulations get to the point where we're modeling accidents and
> parts that fail and are made wrong, we'll have made great progress. :-) In
> general, having data in a format that you can both use it for
> simulation/visualization and automated assembly would make a lot of things
> possible.
Workin' on it. Trying to keep MPEG out of it though. :-/
Paul D. Fernhout
>> Obviously, it might take something more elaborate to make a video about,
>> say, a virtual actor replacing an automotive transmission, let alone a
>> virtual actor going through all the steps of Gingery's metalworking shop
>> from scrap. But it would seem quite possible eventually, with the right data
>> formats and simulation framework, even with just today's desktop computers
>
> Well, I always thought that this was basically what you meant when you
> (or any of the others of us) were talking about writing a simulator.
I saw a simulator more for designing things and virtually testing them as-if
mechanically, as well as looking at systemic failure modes etc. of entire
assemblages or structures or habitats, with the assumption that most
assembly would be automated somehow or just be done by following textual
directions maybe with pictures (or even like today's Lego guides for
assembling models, which have little text, just a detailed sequence of
incremental pictures). I really had not thought much about simulation of the
people around the machinery in any detailed industrial engineering ergonomic
way, like simulating how far an arm can reach inside a housing to tighten a
bolt with a wrench (although I've thought about simulating people abstractly
in terms of work-hours or air/water/food consumption etc.). The
instructional video aspect is a different angle from what I have been
thinking of, as would be, if generalized, human-factors testing from that
same infrastructure, with a 3D human model interacting in a "crash test
dummy" sort-of way with the system.
You're also right that if you have the simulation of an assembly process, it
would often be helpful to look at things from different angles. You can do
than now with LEGO's Digital Designer, viewing your model assembly from an
arbitrary direction:
"LEGO Digital Designer."
http://ldd.lego.com/
There you can build models, generate instructions for them as HTML pages,
and even use that model to place an order with LEGO for all the parts in
your design (it gives you pricing data). The designs can be shared through
that site. We have bought the parts for one design made by someone else, and
bought one design we made ourselves. To an extent, that software does some
simulation in the sense that it does collision detection for bricks
(although they might do that in some more rote way). You can "explode" your
models too, so again, a sort-of simulation aspect. But I don't think in
either case it is anything like a real physics engine.
Anyway, LEGO's Digital Designer is another example of how open manufacturing
(in terms of sharing designs for 3D objects) is becoming mainstream, but
sometimes in unusual and unexpected ways.
--Paul Fernhout
Nathan Cravens
Nick Taylor
>> Video is the only way Hexayurts replicate other than by direct
>> personal instruction. No written instructions or photosequences got
>> people over the "can I do that myself?" hurdle.
>
> Probably because you didn't use a standard format for writing out the
> information. It's in an image, of all things:
>
> http://www.appropedia.org/Hexayurt_schematics
>
> How is my instruction generator supposed to work on random images? :-(
yea - so as I said, we're talking at cross-purposes.
Who's your audience?
n
ben lipkowitz
> Let's call this the CAD instructional. Here's an example:
>
> Design-in-focus:
> Desktop computer installation:
> Text, Audio Command: "Plug mouse into jack": Visual: Mouse input appears
> being plugged into the appropriate jack repeatedly until "Next" on the
> display is clicked, confirming the process understood and complete.
> Text, Audio Command: "Plug keyboard into jack": Visual: Similar to the mouse
> animation. "Next" on the display is clicked, confirming the process
> understood and complete.
> And so on...
This just flopped onto my screen:
http://blog.reprap.org/2009/02/pull-yourself-together-bot.html
in which the reprap project rediscovers grid-beam, or: a working example
of interactive instructions.
short summary: you hook a stepper motor to some threaded rod; the computer
then instructs you to put a nut on the rod and turns it a certain number
of turns to achieve the desired frame dimensions.
Nick Taylor
> http://blog.reprap.org/2009/02/pull-yourself-together-bot.html
>
> in which the reprap project rediscovers grid-beam, or: a working example
> of interactive instructions.
>
> short summary: you hook a stepper motor to some threaded rod; the computer
> then instructs you to put a nut on the rod and turns it a certain number
> of turns to achieve the desired frame dimensions.
Yea - one of the things that's interesting about that is that... well,
in the words of Zach...
"
2. the instructions are the design!
the first 20 lines of the gcode are instructions such as the number of
nuts you'll need, the number and lengths of the various rods you'll
need, as well as various bits of information such as volume diagonal
lengths, etc. not only that, but as its an interactive process, it makes
assembly a snap: you simply follow the step-by-step instructions and the
various parts go together in front of your eyes. no more digging through
complicated build instructions. just connect the dots!
"
Which ought to appeal to Bryan, though obviously if you want this idea
to rapidly replicate in people's minds, the process will need to be (and
will inevitably be) videoed :)
n