<cr1901_modern>
lain: Do Intel USB controllers have the behavior described in the highlighted message to your knowledge?
<cr1901_modern>
Since you know a thing or two about Intel :P
digshadow has quit [Read error: Connection reset by peer]
<lain>
oh, I didn't have scripts enabled on that site so I didn't see the highlight
* lain
reads
<lain>
yeah I have no idea
<lain>
intel doesn't make their own usb3 controllers afaik
<lain>
I'm pretty sure it's a third party IP core
digshadow has joined ##openfpga
<cr1901_modern>
My question is why would a device scan SDRAM in the first place for xfer requests? Would you, ya know, send commands to the device itself?
<cr1901_modern>
wouldn't* you
<lain>
I don't know anything about how usb controllers in PCs work tbh
<lain>
not a clue lol
<lain>
that sounds goofy though, if true
<cr1901_modern>
I have no idea how something goes from "CPU register to PCIe device", tbh. I just know there's a lot of layers on PC motherboards
<felix_>
iirc intel xhci is synopsys dwc3
<lain>
guessing they comm via memory-mapped registers but I haven't actually looked at the xhci spec
<cr1901_modern>
lain: Thanks for doing the research for me :3
kuldeep has quit [Remote host closed the connection]
kuldeep has joined ##openfpga
m_w has quit [Quit: leaving]
pie_ has quit [Changing host]
pie_ has joined ##openfpga
promach has joined ##openfpga
<azonenberg>
So it looks like i should be able to buy a secondhand (windows XP era) 1 GHz DSO from ebay for under $3K... I don't have the cash now, but is that a reasonable price? Anybody here have experience with secondhand test equipment?
<azonenberg>
Just thinking over options for long-term as i build out my lab
<azonenberg>
I see ancient CRT-based ones for a lot less than that but i'm looking for something at least somewhat modern
<azonenberg>
Other option would be to hold off until i can afford $12K for a new keysight et al
DocScrutinizer05 has quit [Disconnected by services]
DocScrutinizer05 has joined ##openfpga
digshadow has quit [Remote host closed the connection]
digshadow has joined ##openfpga
<davidc__>
azonenberg: seems a bit pricy
<davidc__>
azonenberg: Also, do you want it for anything other than eye diagrams (and equivalent?)
<azonenberg>
I want to be able to sniff normal signals too
<azonenberg>
i.e. non-sampling
<azonenberg>
But i could probably get by with a sampling scope if i could get one cheaper
<azonenberg>
i rarely have to view non-periodic *analog* waveforms
<azonenberg>
i either need fast digital one-shot or periodic analog
<azonenberg>
davidc__: but i dont think you can get sampling scopes cheap either :p
Hootch has joined ##openfpga
<azonenberg>
davidc__: i can get used ones for way cheaper but they're in horrible condition or ancient
<davidc__>
azonenberg: I meant more in terms of memory depth; unless you're just doing SI work, most of the older scopes only have record lengths that are short enough to be irritating
<azonenberg>
Oh
<azonenberg>
I mostly want it for SI
<azonenberg>
For deep captures i'll be using the starshipraider :p
<davidc__>
(some actually had combined ADCs/memory ICs, or the crazy capacitive samplers)
<azonenberg>
Which will give me 500 Mbps (I could probably build a 1 GSa/s front end board for it that is less tolerant to overvoltage)
<azonenberg>
I have some 6" SMA cables coming in tomorrow which should tidy the setup up, the 12"s are overkill for this
<davidc__>
yeah, fair enough. Eh, would probably do the job.
<davidc__>
re: probing, I find myself using the 1k/50ohm trick for most of my high speed probing anyhow
<azonenberg>
"1k/50 ohm" ?
<davidc__>
easy to make up with an SMD resistor and some mini-coax
<azonenberg>
So i have yet to figure out how starshipraider will actually be getting signals from the DUT to the comparator
<azonenberg>
my thought is to have one front-end with flying leads for low speed then a solder-down active probe for high speed
<davidc__>
azonenberg: set scope to 50ohm input mode, probe the board with a micro-coax with a 1k ohm resistor between the trace and the coax center pin
<azonenberg>
but details TBD
<davidc__>
ground as near as possible
<davidc__>
1k input impedance as seen at the circuit, no transmission line issues, at the expense of increased noise
<davidc__>
azonenberg: you know about the crazy lossy-coax they use for high impedance probes, right?
<azonenberg>
No?
<azonenberg>
also interesting
<azonenberg>
I may want to make up some of those for starshipraider
<davidc__>
azonenberg: a crappy 1k/50ohm probe worked great up to 500mhz (BW limit of the scope I was using), and seemed to be just fine
<azonenberg>
excellent
<davidc__>
and it was made out of whatever crap I had at the bench, without particular care
<davidc__>
It performed better then the passive probes with the scope (which is to be expected, they're only specced to 350mhz)
scrts has quit [Ping timeout: 248 seconds]
<azonenberg>
Lol
<azonenberg>
well for the short term since i dont have a magic money tree to buy a new DSO
<davidc__>
Re: lossy coax; I need to get to sleep, but gist of it is that you can't build 1mohm characteristic impedance coax, and since thats the termination inside the scope in high impedance mode, you'll have reflections on the coax
<azonenberg>
I think i'm going to build a PRBS generator out of my other fpga devkit
<azonenberg>
and rather than measuring the eye, just measure the BER in loopback mode
<azonenberg>
if i get an acceptably low BER, IDGAF what the eye looks like :P
<davidc__>
You could build a DIY ETS ;)
<azonenberg>
ets?
<davidc__>
equivalent time sampler
<azonenberg>
Oh
<azonenberg>
um, yeah - that require more analog-fu than i have time for (to do full eye measurements)
<azonenberg>
And more importantly i dont entirely trust that it wouldnt introduce noise too
<azonenberg>
or jitter, or other things that woulud confound the measurement
<azonenberg>
honestly i dont need eye compliance for anything
<azonenberg>
If i send a bit and get the same bit back with failure probability below say 10^-10 I'm good
<azonenberg>
(better would be nice but we'll see how far i can push it)
<azonenberg>
It should be straightforward to build a design in the Zybo that generates a LVCMOS33 PRBS31 pattern, feeds it out through a SMA cable to the DUT
<Spida>
azonenberg: did you finish the TDR? Does it work?
<azonenberg>
Basically i PoC'd the architecture, it functions as a TDR, but the first board spin is a) not super useful and b) not fixable
<azonenberg>
i used a programmable gain amp at the front end for measuring the reflections and at max gain it's not strong enough to get a decent ENOB
<azonenberg>
but this isnt the amp's fault
<azonenberg>
It's primarily due to the RJ45 I used being a magjack without decent high frequency performance
<azonenberg>
it's low-pass filtering my pulses and reflections so i lose all of the high frequency information
<azonenberg>
and i wasnt able to find a non-magjack with a compatible pinout
<azonenberg>
so without really extensive rework i wouldn't be able to get useful work done with it
<azonenberg>
I may or may not eventually respin with these things fixed, either in a quad RJ45 configuration or a single channel SMA that I can hook to whatever
<azonenberg>
Might actually be possible to make a TDR AFE that hooks to a starshipraider and replaces one of the 8-channel probe modules
<azonenberg>
with very little additional cost i could add an i2c eeprom to each card for discovery that lets me tell what the interface card is
<azonenberg>
Might need that anyway for other purposes, like if i make a line card that's 1 GSa/s over a restricted voltage range vs 500 MSa/s over a wide range
massi has joined ##openfpga
scrts has joined ##openfpga
Bike has quit [Quit: leaving]
promach has quit [Ping timeout: 276 seconds]
promach has joined ##openfpga
kuldeep has quit [Read error: Connection reset by peer]
kuldeep_ has joined ##openfpga
kuldeep_ has quit [Quit: Leaving]
pie_ has quit [Ping timeout: 240 seconds]
Bluebell_ has joined ##openfpga
pie_ has joined ##openfpga
pie_ has quit [Ping timeout: 240 seconds]
pie_ has joined ##openfpga
pie_ has quit [Ping timeout: 256 seconds]
m_t has joined ##openfpga
kuldeep has joined ##openfpga
whitequark has quit [Ping timeout: 245 seconds]
whitequark has joined ##openfpga
Bluebell_ has quit [Ping timeout: 264 seconds]
pie_ has joined ##openfpga
pie_ has quit [Changing host]
pie_ has joined ##openfpga
bibor has quit [Read error: Connection reset by peer]
bibor has joined ##openfpga
bibor has quit [Read error: Connection reset by peer]
bibor has joined ##openfpga
Hootch has quit [Read error: Connection reset by peer]
Bike has joined ##openfpga
eduardo_ has joined ##openfpga
eduardo has quit [Ping timeout: 240 seconds]
digshadow has quit [Quit: Leaving.]
pie_ has quit [Ping timeout: 245 seconds]
amclain has joined ##openfpga
Hootch has joined ##openfpga
massi has quit [Remote host closed the connection]
pie_ has joined ##openfpga
pie_ has quit [Changing host]
pie_ has joined ##openfpga
m_t has quit [Quit: Leaving]
digshadow has joined ##openfpga
maaku has joined ##openfpga
indy has quit [Ping timeout: 245 seconds]
indy has joined ##openfpga
openfpga-bb has quit [Ping timeout: 240 seconds]
azonenberg has quit [Ping timeout: 256 seconds]
azonenberg_work has quit [Ping timeout: 258 seconds]
openfpga-bb has joined ##openfpga
openfpga-bb has quit [Ping timeout: 255 seconds]
azonenberg_work has joined ##openfpga
openfpga-bb has joined ##openfpga
azonenberg has joined ##openfpga
digshadow has quit [Quit: Leaving.]
<balrog>
cyrozap, pointfree, how's the PSoC stuff going?
qu1j0t3 has quit [Quit: WeeChat 0.4.3]
qu1j0t3 has joined ##openfpga
Hootch has quit [Quit: Leaving]
<pointfree>
balrog, cyrozap: I'm finishing up a big HV tile map in Inkscape. I'm currently labeling VS coordinates and their bits/registers on the SVG image.
<pointfree>
aka:
<pointfree>
(1,0), (1,1)
<pointfree>
(0,0), (0,1),
<pointfree>
The HV map contains (right and left) HV blocks from UDB's:
<pointfree>
VS VS
<pointfree>
HS--HVB--HC--HVA--HS
<pointfree>
HS--HVA--HC--HVB--HS
<pointfree>
VS VS
<pointfree>
The pattern repeats so you can see why the UDB array is 4x6 (not entirely arbitrary). DSI registers are the same as ROUTE registers, different only in that they are for the upper and lower rows of UDB's while the ROUTE registers are for inner two rows of UDB's. Even PI (Port Inteface) tiles start at the same offset for DSI and ROUTE. Very clean.
<pointfree>
Although, It looks like there is some more flipping and inverting action between blocks of 4 HV's.
<pointfree>
In case my ascii art got screwed up, the HC tiles plug in between the HV quadrants (in the middle between HVA and HVB).
<pointfree>
...and the ##openfpga channel logger reordered my message and truncated it or my client didn't wait between each line :(
<pie_>
pointfree, are you putting this stuff up somewhere?
<rqou>
hmm, so apparently the chipwhisperer ships out of halifax and supposedly will be here tomorrow
<rqou>
that's why shipping cost me so much
<nats`>
I'm still hesitating buying one
<nats`>
I would like to support the software dev which is a really fucking great piece of tool
<rqou>
dumb question: do all sma edge connectors have (approximately) the same dimensions?
<pointfree>
pie_: The HS tile map and UDB map are on the openfpga wiki. Some stuff is on my federated wiki, which unfortunately doesn't work great on mobile btw http://www.gelforth.org/ I have some routing formulas in darcs repositories. They are written in Forth, but I promised to post infix formulas to the wiki. Most stuff is only on my laptop at the moment.
<balrog>
rqou: you ordered a ChipWhisperer-Lite?
<rqou>
yeah, the two part version is both much more expensive and costs much more in shipping
<pie_>
oh i was just totally missing context on what the project actually is
<nats`>
rqou do you plan to play directly with it ?
<rqou>
i have some projects in mind for it
<nats`>
**it'll stay ona bench :p**
<rqou>
but they're both "sooper sekrit"
<nats`>
:!D
<pointfree>
pie_: I don't think PSoC Creator changes any seed value between builds. I will have to explore the PSoC Creator cli options and flags some more http://hub.darcs.net/pointfree/psoc-cli-tools/
<balrog>
pointfree: hmmmm
<pie_>
then im confused
<balrog>
do you think it would be possible to wrap these tools within wine and make them work with iceStudio? :)
<pie_>
argh im getting sidetracked again xD
* pie_
goes back to studying
<nats`>
icestudio the python stuff ?
<nats`>
my port is a fail
<nats`>
I stopped after spending a day on it for a simple verilog code
<nats`>
it's way to new and not mature to be usefull
<rqou>
another dumb question: how is a $2 sma connector different from a $5/$10/whatever sma connector?
<nats`>
bandwidth mainly
<rqou>
most of them all claim to be rated dc-18ghz
<nats`>
you have 26GHz sma and 1GHz sma :D
<nats`>
yep they claim a lot but there are a lot of factor
<nats`>
look at insertion loss/attenuation/number of insertion
<nats`>
etc
<rqou>
alright, so i'll buy the molex one for the main chipwhisperer board and the shitty one for the DUTs
<rqou>
:P
<nats`>
:)
<pointfree>
balrog: These cli tools are already running with wine. You could probably write your components in Verilog on Linux. The C code is already no problem on Linux.
<pointfree>
balrog: The problem is the top level "design file" of a project is not understood and Verilog seems to be autogenerated from it. However, I'm not sure that this is the case with components you have written yourself in Verilog.
digshadow has joined ##openfpga
<rqou>
hmm apparently digikey doesn't even stock crappy sma cables
<azonenberg>
i've probably spent $100+ on RF cables, terminators, attenuators, etc over the last few weeks but it's worth it, they'll hold their usefulness as my lab upgrades to faster equipment
<rqou>
yeah, that's basically the cheapest one digikey has
<rqou>
no chinesium u.fl to rp-sma stuff :P
<mtp>
u.fl is the worst
<nats`>
u.fl rated for 10 insertion :D
<rqou>
yeah i know that :P
<nats`>
USING UFL TOOL !
<rqou>
there's a dedicated tool?
<nats`>
0.5 insertion by hand :D
<nats`>
yep
<rqou>
i didn't even know there's a tool
<rqou>
:P
<rqou>
i don't think i've ever heard of anyone having it
<nats`>
I was browsing on amazon to find stuff about RF or digital design
<lain>
nats`: if you have any interest in optics (keep in mind that optics of course also covers EM, since light is just really high frequency EM), Optics by Eugene Hecht had a new revision last year and it's fantastic
<nats`>
I'm adding it on the list :)
<rqou>
huh didn't know about the new edition
<rqou>
i have the 4th edition
<rqou>
(i still suck at optics/EM :P )
<lain>
yeah I pre-ordered 5th edition
<lain>
got it about a year ago
<lain>
still haven't finished it, very close to the end though :P
<nats`>
not so expensive
<rqou>
i didn't read through it because it was a course textbook :P
<lain>
I don't have any formal training in EM, but that book made it totally accessible
<rqou>
so i only looked at the parts we covered in class
<lain>
all the other books I see are like
<rqou>
i have no formal training in EM or optics
<lain>
"well here's the equations, you can figure out the rest"
<rqou>
beyond the one class
<lain>
and I'm like
<lain>
no I want words :<
<lain>
and pictures
<lain>
pictures are great.
<rqou>
you won't like berkeley very much :P :P
<nats`>
yep that's why I have a bunch of old books where they explain a lot
<lain>
lots of academia is math math math
<lain>
which is one of the reasons I dropped out
<azonenberg>
i want a panavise too
<rqou>
more seriously, we do have words/pictures if you attend lecture
<rqou>
but berkeley is very math heavy
<lain>
eh, school just isn't for me
<rqou>
we're pretty infamous for "got EE degree, can't even solder"
<lain>
haha
<lain>
rqou: sadly true of many EE degrees
<nats`>
someone can come here and push the run/stop button -_-
<lain>
I've only heard of one uni that actually talks about pcb concerns, like fabrication, design, etc
<lain>
I've met a lot of EEs that get jobs and they're like "the heck is CAD?"
<azonenberg>
o_O
<lain>
they've never touched cad in school
<rqou>
the professor teaching the intro circuits class here decided that "got degree, can't solder" wasn't every acceptable
<lain>
never done a pcb, never done any real design
<azonenberg>
lain: you want spicy food? just squirt a little dab of OC spray into a corner of the pot
<azonenberg>
mix it up
<rqou>
so he added a pcb project
<lain>
maybe some stuff with 7400 series logic on a breadboard or etc
<azonenberg>
rqou: and everybody failed? :p
<lain>
rqou: nice
<rqou>
however, there wasn't much about how to do it other than "ask the TA in lab"
<azonenberg>
there are people graduating our CS program who can barely code
<azonenberg>
they know theory but no applications
<rqou>
so you end up with some people like me that know what they're doing
<rqou>
and some people that just did "line up all the resistors and all the capacitors in rows and try to connect them all together"
<rqou>
and then wonder why it doesn't work very well
<lain>
and my main complaint with "all theory no practice" is that if they don't practice, they forget all that theory :P
<rqou>
project was an instrumentation amplifier btw
<lain>
haha
<azonenberg>
lain: yeah
<nats`>
ok problem solved take a screw driver in the mouth while holding probe and push the button :D
<azonenberg>
You need to do both
<lain>
rqou: and the thing is, I bet they could understand fully why it didn't work if it was presented correctly, because they probably do know the theory... it's just they have no way to relate theory to practice
<rqou>
yeah probably
<lain>
the two things are just totally disjoint in today's universities
<rqou>
apparently ME at berkeley is even worse
<rqou>
we have a full-blown machine shop, and how many students know how to use it?
<rqou>
a negligible amount
<lain>
I was /really/ impressed to see a friend of mine doing a CS degree, and one of the classes was all about exploits. they had to reverse-engineer professor-provided binaries, find the exploits, write an exploit, understand what went wrong, and, given partial source, rewrite it securely
<lain>
notre dame uni (right up the street from me) has a pretty awesome silicon design program from what I've seen - I think they even have a semiconductor fab on campus
<lain>
or at least it's some kind of nanofabrication facility
<lain>
with bunny suits and all that
<rqou>
we have two of those
<lain>
and I've seen a class where they design a chip
<lain>
rqou: WTF "On the first day of class, course professor Doug Tygar handed out a syllabus that said students would receive failing grades and be removed from the course for a single tardy or absence or for any use of electronics in class."
maaku has quit [Read error: Connection reset by peer]
<lain>
this is not how to handle this
<lain>
wtf
<azonenberg>
lain: o_O
<azonenberg>
how are you supposed to be a CS student without a computer
<lain>
I went to a local branch of a university here in town (IUSB - Indiana University South Bend) and they over-enrolled the fuck out of CS101
<lain>
and the prof was real laid back about it
<lain>
he's like yeah well, we're going to do class as usual, it'll just be a little warm in here with all the extra bodies :P
<lain>
because he knows the drop rate is like 70% anyway
<lain>
lots of people go into CS thinking that designing video games is as easy as playing them
maaku has joined ##openfpga
<lain>
by the end of first semester it becomes clear it isn't, and most change majors :P
<rqou>
one semester our intro cs class got overenrolled to the point that it was (on paper) larger than the largest lecture hall on campus
<lain>
lol
<rqou>
but lectures were webcast, so it was managable
<lain>
one uni I attended... their engineering program, the entire first year, was "you cannot be an engineer, change majors"
<lain>
they would just repeat this in different ways
<lain>
they had people come in saying they wish they had changed to chemistry or etc, not engineering, so unhappy, etc
<lain>
it was bullshit tactics to try to get people to drop early, because "realizing too late that you won't be an engineer is a very expensive mistake"
<azonenberg>
Ok sooooo
<nats`>
I love french universities :D
<azonenberg>
My laptop is leaking X handles again
<lain>
so I dropped out and 7 years later that school's IT department bought like a dozen of a product I created myself :P
<lain>
poetic justice or something
<azonenberg>
"Maximum number of clients reached"
<azonenberg>
Any ideas on how to figure out what is leaking?
<rqou>
someone (iirc bunnie?) pointed out that CS is basically the only field where they assume you already know how to do it before you get into the major
<rqou>
i didn't even know x had a maximum handle count :P
<lain>
azonenberg: lol
<rqou>
i thought that was exclusively a windows problem :P
<nats`>
it must have a maximum :p
<nats`>
can't be infinite :D
<rqou>
i meant a maximum that is less than the total memory size of the computer
<lain>
just start swapping out to amazon s3
<lain>
infinite enough™
<rqou>
windows has a maximum because gui objects are kernel objects
<rqou>
because windows (win32) is architected brilliantly
<nats`>
IIRC that's no more true
<rqou>
i thought most of it is still in the kernel? i know they finally took out font rendering
<nats`>
maybe it was a long time ago for me
<nats`>
I stopped VXing and reversing in late 2000's
<nats`>
:D
<nats`>
something like 2008/2009
<rqou>
interestingly enough windows kernel memory pages can be paged out
<rqou>
afaik linux/mac can't do that
<nats`>
someone has a good doc about internal working of DDR ram ?
<nats`>
basically I can't find anythin explaining bank system etc....
<rqou>
micron's datasheet/appnotes? :P
<lain>
I always end up looking up the bank/rank crap
<nats`>
it's really confusing
<lain>
iirc rank is how many chips make up a full 64bit wide bus, so like if you have x8 chips, you need 8 chips to make a x64 bus
<nats`>
like if they choose name exactly to be confusing
<lain>
so that's 8-rank
<lain>
and then bank is the number of chip selects
<lain>
so banks are overlapping, like if you had 16 x8 chips, you could have 2 banks of 8 rank each or something like that, I think
<nats`>
I thought bank was in the chip
<azonenberg>
yeah banks are within the chip
<lain>
ah
<lain>
I probably got it all wrong then :D
<lain>
like I said, I always forget
<azonenberg>
ranks are i think the number of chips sharing one DQ bus
<lain>
I mess with it often and I've never managed to remember any of it
<pointfree>
Hello digshadow, it's looking like my talk will need to ripen a little more. I think it should be about methods as well as PSoC stuff so it's interesting to more people.
<pointfree>
Although, cyrozap, balrog, and others, I did just automatically synthesize a route from: hseg --> hv --> vseg
<azonenberg>
so a 4-rank dimm would have 32 x8 chips, 16 x16, 8 x32, etc
<rqou>
i'm pretty sure webgl --> TZEE is almost certainly possible
<azonenberg>
every ~2 weeks of uptime i stop being able to open new X apps
<nats`>
maybe it's really X server itself
mifune has quit [Ping timeout: 255 seconds]
<azonenberg>
Good question
<azonenberg>
I just decided that this time i was going to put some work into tracking it down
<azonenberg>
rather than blindly restarting X
<azonenberg>
Looking at xlsclients... let's see
<lain>
inb4 "there's 5000 remote users connected to my laptop!"
<azonenberg>
i didnt restart my terminal emulator
<lain>
:P
<azonenberg>
or xfdesktop or xfsettingsd
<azonenberg>
not the terminal
<azonenberg>
i had a background Thunar process but that only had one handle
<azonenberg>
hmmm
<azonenberg>
not xfce4-power-manager
<azonenberg>
or vmware-tray
<azonenberg>
i'm about out of ideas
<azonenberg>
i have five handles free now, which is a start
<azonenberg>
by killing random processes and not restarting
<azonenberg>
But that tells me zilch about the leak b/c i think i've killed everything :p
<lain>
except the hidden leaky backdoor
<lain>
which doesn't show up in `ps'
<nats`>
I bet it's a X server fault :p
<azonenberg>
nats`: very likely
<azonenberg>
lain: lol
<azonenberg>
if i'm pwned, this is the only evidence they left
<nats`>
a shitty one :D
<nats`>
they could have made a mermaid song in your speaker :p
m_w_ has joined ##openfpga
magulo_ has joined ##openfpga
seu_ has joined ##openfpga
davidc___ has joined ##openfpga
pointfree_ has joined ##openfpga
<lain>
how do they paint those anime figures so intricately
balrog_ has joined ##openfpga
defparam_ has joined ##openfpga
m_w has quit [*.net *.split]
Bike has quit [*.net *.split]
davidc__ has quit [*.net *.split]
balrog has quit [*.net *.split]
magulo has quit [*.net *.split]
seu has quit [*.net *.split]
pointfree has quit [*.net *.split]
defparam has quit [*.net *.split]
reportingsjr has quit [*.net *.split]
magulo_ is now known as magulo
pointfree_ is now known as pointfree
Bike has joined ##openfpga
balrog_ is now known as balrog
<lain>
huh.
m_w_ has quit [Quit: Leaving]
<lain>
they're 3d printed with laser litho, by the look of this behind-the-scenes video... and then they do silicone molds from the 3d printed part, and cast the actual figure components in those molds
<lain>
makes sense
<lain>
painted by hand with airbrush
<lain>
no wonder they cost a fortune :P
reportingsjr has joined ##openfpga
<azonenberg>
lain: but there are so many well-paid software devs and engineers among the ranks of otakus
<azonenberg>
that there's a thriving market who can afford them