#apertus IRC Channel Logs

when somebody thinks, this feature would be cool, here have a camera to show you how it does or doesn't work

like I said, I'd like to help if I can

I wish I had a vector to fish around for some grant money

(as well as getting the software part working)

the development/build system for the camera software, yes

regarding post processing you probably know better than I do :)

great!

nope

oh!

basically the camera has wifi, as does the remote

well, we plan to have ethernet, probably wifi as well

it mounts directly to the module mount on the camera

and we definitely plan to allow modifying all aspects via network

yeah, a iOS app could achieve the same thing without proprietary (expensive) hardware

wifi ought to be part of the central body

ethernet makes sense as a module

yeah

guesst: probably depends on the digitizer

but you could build a controller very cheaply with Arduino and wifi

oh? Interesting

first, it is not easy to find a wifi module/stack which doesn't require proprietary firmware blobs

right

Doesn't seem to be a problem with the Red cameras

plus you have to pair the devices with each other before you can remotely control

never heard of ssl?

why it has to be so open

yes, there are many options, just not convinced that (open) wifi should be one of them :)

there is no difference between an open/closed wifi... the application has to be secure

you can as well as break an eth cable and steal control

does it matter if a blob is part of module (pre-flashed) or not part of it (ram-only module) ?

the problem with most devices is that they do require a proprietary driver, and that is not really an option for us

note that bluetooth might be a viable alternative

then the world will be a better place I guess? :)

if you put it in a metal box, yes

everything sucks inside a metal box

others got antennas

unless you are running an ENG style camera and need the crazy range

rather an antenna than lost control..

yeah I figured you'd play that card :D

guesst: have you even done a cinema style shoot with a crew?

you don't want little bits sticking off the camera, trust me

there are already going to be wires everywhere

if you did have a small antenna it would have to be recessed, protected somehow

zigbee would be awesome

but what if Zigbee on the camera and a Zigbee-wifi bridge? would that still be a no no for you guys?

there is no no-no per se, even if we decide not to go this or that route for whatever reason, you still can add that part yourself (or find somebody to do it for you)

usually information which is accompanied by expressions like 'because I say so' or 'that's how it is supposed to be' get a lower rating than those presented with a striking argument :)

"build your own damn camera

hahaha :D

would not be much easier to make for them a new sensor front, they were opensource and open for such a mod..

the problem is more that andrey doesn't really want to share development stuff

(for example, you won't we able to get access to the latest designs, may it be schematic or layout with elphel)

that is alays the problem with and/or explanation for not being open :)

both, yes

btw: Bertl there are now a number of wifi modules with open drivers, some are still closed but they shouldn't be problematic to avoid.

We use it it peer computers for collaborative editing and asset management.

Bertl: I don't think finding wifi with open drivers will be much of an issue

maybe you can make a suggestion for a module which can be easily integrated and has open drivers then?

I'm not entirely certain as to the easily integrated part

anything which connects to a microcontroller or operates like an ethernet phy would be fine though

broadcom has binary crap unless that's changed recently

http://hackaday.com/2013/08/09/tiny-wifi-modules-again/

http://en.wikipedia.org/wiki/Comparison_of_open_source_wireless_drivers

that might have changed over the past few years though

but it general, the question is what you want to do via wifi

Bertl: Could be lots of things

remote operation, transmission to video village, peering with other devices

plus, I'd recommend integrating a wifi AC chip

realtime video transmission would need to be lower quality (generate jpegs and send those), so it's whatever your encoding is

but you can totally send HD video over wifi, bandwidth just affects the bitrate

as far as sending full-quality, the camera could catch up between takes

such a thing would let the director see the live camera output

Though that would be more cumbersome

Another dongle to mount on your camera

well, we can't integrate everything into the camera, otherwise it will require four people to carry it, not speaking of the batteries :)

true, though I doubt a wifi chip will change much

Wifi could also be on a separate board attached to the main bus

sorry, not to me -- I don't see why you would want to torture your SSDs with wifi :)

so the camera and your PC can manage your footage behind the scenes without you fussing

do a quick calculation what amount of data we are talking about and how long the synchronization via wifi would take

also calculate what amount of efford would go into handling partial syncs and proper recombination

Bertl: no effort at all

btw, is this a real world scenario at all, i.e. that you want to process your raw footage on site? wouldn't it make much more sense to record everything and then do the time consuming post processing somewhere else?

Bertl: so there are stages: recording, ingest, logging, light grade, editing, final grading...

ingest is importing the footage from the camera's recording media to your asset management system

depending on the production, at least recording and ingest are done on set

It used to be that you began the logging process with the "dailies" after the film lab got done processing the footage from the previous day's shoot, but with digital you can start whenever the files are on a computer

assuming your 'computer' can handle the bandwidth, the SSDs can be read with 500MB/s and more, that means roughly 30 minutes for a single SSD with 1TB

assuming that you will barely get 25MB/s over wifi, this means roughly 10 hours for a single drive over wifi

I don't think only 25MB/s is a safe assumption with wifi anymore

Bertl: so wifi AC can achieve close to 7gbps theoretically and maybe 1.7-2.5 gbps in practice

which one from the list was the AC chipset with open drivers and firmware?

troy_s: that's true for *most* filmmakers

there are some exceptions

And even temp edits aren't editorial.

Robert Rodriguez does

troy_s: fair enough

Bertl: of course with modern chips, it seems you will likely get under 100 MB/s in actual TCP throughput

Normally the DIT is pulling the full raws and doing the one light.

And the resultant LUT would be baked into editorial, so yes, they would need the frames.

dmj_nova1: single stream maximum of 802.11ac is 866Mbit/s under optimal conditions

The DIT would do a dupe off of the original disk.

they multiplied a number already factored in

looking at actual benchmarks, 250-600 mb/s looks typical

i.e. just change the SSD pack in the first pause and you'll be done before the next recording starts

8 to 1 shooting ratio?

I have yet to see wifi transfer.

SDI everywhere. Teradek or such for wireless display.

Either a R3D rocket scenario or the Sony reader etx.

troy_s: what recieves the teradek signal at video village?

Personally I would think it is all moot by and large given the target audience. Likely say, dual SSDs on board? Then all one would need is an eSATA or USB3 or Tbolt drive dock.

precisely

I have no clue about Bolt protocols.

dmj_nova1: maybe consider a proper directional microwave link?

you could get really good rates with that and even supply the camera with power or recharge during breaks :)

Microwaves still freak me out. There are some xmitters that easily exceed regulations.

But for a get-it-done goal driven mindset, SSDs or dual SSD striping or whatever would seem fine.

bad

Huh?

If all you can muster is a stripe, so be it.

Striped frames.

troy_s: oh

Nothing wrong with striping though.

dmj_nova1: what kind of data loss do you see/expect with striping?

Bertl: RAID array striping easily results in brokenness

As far as alternating frames accross drives, yeah, we should

I have only seen RAID striping suck with hardware RAID, and that is always a snapped controller doing its own thing.

At the data volumes we're looking at, we may want file-based striping and mirroring

Because if you're shooting terabytes a day, you will get bad copies on a daily basis

All hypothetical.

Having a camera that can shoot five minutes is still infinitely far away. ;)

Bertl: :)

yes, it is a 'perfect' sensor, but it isn't mounted yet

I will try to get an OCIO LUT set done.

dmj_nova1: you should change the brand of your SSDs :)

if you get an unrecoverable read error every 3TB :)

I would think the project would be celebrating a read error.

It would just silently give you data different than was stored

We should exert an effort to stay in reality

over the last 3 years, there have been about 12 replacements

Which currently is a "camera" that is mounted to a room and can only shoot still frames.

but there hasn't been a single uncorrectable read error yet .. so do the math :)

dmj_nova1: Personally? I use mag disks and SSDs.

I think rexbron does a similar thing.

Mags are cheap and reasonably brisk.

btw, in what way would 'silently giving wrong data' be different with a single disk vs striped?

Yes

yes, those are magnetic disks

Cost effective

Bertl: 'silently giving wrong data' is the same for single vs striped

in parity raid, it's a killer because it can seriously fuck more data up in the name of "correcting" the data

You boys are being silly zippers.

No camera yet remember?

I don't like to rely on anything that propagates hardware failure

striping is fine as long as it's not at the partition level

I definitely would stripe the entire disk :)

To be fair... HFS propagates corrupted data and it is the most in-vogue OS out there right now.

who the fuck uses HFS?

well, OS X and Windows do have shit file systems

oh "in vogue OS" not "in vogue filesystem"

if Bertl gave me a 150 fps 12 stop prototype that fragged frames every 3 tb I would be ecstatic actually...

btw, my whole thing about file systems and wifi transmission is that I think it's possible to do better than the status quo given the right foundation

I agree that it is a good thing to explore possibilities

I also agree that both wifi and storage is not our primary concern for the next stage (after the crowd funding)

we will be busy identifying the proper baseboard and sensor interfaces and hopefully figure out a way to make it extensible

Ax?

Axiom

Project Next always appears more appealing than Project Must Finish

haha

ask away :P

dmj_nova1: What the _fsck_ is going on with Novacut?

DMedia confuses the shit out of me.

Night all. Be well.

evening!

will be there any 4k video footage from the prototype soon?

you are doing a 4k camera and you can not show footage?

at least they got a brand

it is a limitation of the zedboard's I/O

i suggest to get some of those

you can borrow one from xilinx dist

consider it, gotta go

4k. 4k. 4k.

yeah, 4k looks completely different than 2k :)

Using the mythical numbers of the sunglasses-cum-camera vendor, the Panavision Genesis mark one was a 6k camera.

(With perhaps an exception for rexbron)

time for bed

good night!