Current Server Time: 13:35 (Central Europe)

#apertus IRC Channel Logs

2013/07/12

Timezone: UTC


23:21
jucar2
left the channel
23:24
jucar1
joined the channel
00:38
jucar2
joined the channel
00:40
jucar1
left the channel
00:49
Bertl
dmj_nova: how's going?
02:02
jucar1
joined the channel
02:04
jucar2
left the channel
02:46
jucar1
left the channel
02:47
jucar1
joined the channel
02:54
Bertl
okay, I uploaded version 0.8 of the CMV12k adapter, this should have everything required so please start checking the schematic and pcb for errors
02:55
Bertl
I probably will add an optional power supply later, in case the zedboard is not able to power the adapter properly
03:08
jucar1
left the channel
03:42
jucar1
joined the channel
03:46
jucar1
left the channel
04:04
jucar1
joined the channel
06:38
s3bastian
joined the channel
09:06
s3bastian
left the channel
09:16
s3bastian
joined the channel
09:40
dmj_nova
kicad is what we're using for PCB design, yes?
09:44
s3bastian
Bertl, is using Eagle
09:45
s3bastian
but KiCad can read the schematic and board files from Eagle as far as I understand
09:45
s3bastian
or the conversation/exchange is rather easy
10:01
FergusL
hi fellow apertusians
10:01
dmj_nova
hi FergusL
10:01
FergusL
hi dmj_nova, I think we know each other from some other channels :)
10:01
dmj_nova
FergusL: I was just about to ask you the same thing
10:02
dmj_nova
But I can't remember which
10:02
FergusL
Novacut and blender related channels most likely
10:02
dmj_nova
FergusL: That would definitely be it
10:02
FergusL
yes !
10:05
dmj_nova
FergusL: I'm trying to remember, are you part of the blender team?
10:05
FergusL
oh, no ! not at all
10:06
dmj_nov
found a conversation about rsync in the logs but can't think what prompted it
10:07
FergusL
I'm simply part of a spontaneous and informal micro-team of blender user in the cinema-field with which I developped some py scripts, it's all work in progress
10:07
dmj_nova
Ah I see
10:07
dmj_nova
what sort of scripts?
10:08
FergusL
hm... I probably asked about ways to backup camera shots
10:08
dmj_nova
Ah, I see
10:08
dmj_nova
Yeah, the Axiom is pretty exciting as far as dmedia is concerned
10:09
FergusL
oh, basic stuff, a hack that kind of emulates the "dual monitor" of usual NLEs, a script that auto-refreshes clips from their files and the biggest one, currently unfinished is basically adding EDL export
10:10
dmj_nova
ah, do you know much about EDL format?
10:10
FergusL
sure ! yes !
10:10
dmj_nova
EDL export is on Novacut's list of things to add eventually
10:11
FergusL
hm, I know some of it yes, but not much when it comes to transitions, overlays
10:11
dmj_nova
Of course we currently don't do transitions at all
10:15
FergusL
well, that's not a bad thing
10:15
FergusL
I mean, not to have it
10:16
FergusL
have you had the chance to encounter troy_s around this server ?
10:17
dmj_nova
name sounds familiar
10:17
dmj_nova
(yes
10:20
dmj_nova
we had a nice long conversation actually
10:20
FergusL
he knows a lot about the pro cinema fieald
10:20
FergusL
be it in editing workflow or in color science
10:22
dmj_nova
indeed
10:22
FergusL
btw about EDL, is Novacut still Python ? if yes, then part of the code is already here
10:22
dmj_nova
yes, novacut is in python3
10:23
FergusL
I really think EDL is a key point, and more generally offline editing is
10:27
christiankrupa
joined the channel
10:28
dmj_nova
FergusL: yes, and moreover, I would like to see more applications build upon dmedia
10:28
FergusL
hi here christiankrupa :)
10:28
christiankrupa
hey FergusL
10:29
FergusL
dmj_nova: is dmedia python too ?
10:29
dmj_nova
yep
10:31
FergusL
cool !
10:32
FergusL
does dmedia do/will do re-encoding ?
10:34
dmj_nova
FergusL: Not at the moment
10:35
dmj_nova
Though there is a jobs system which would allow us to to reencoding in the background and tune that behavior
10:37
dmj_nova
of course Novacut works perfectly with DSLR footage, so there hasn't been much need for re-encoding as yet
10:38
FergusL
it's smooth with 1080p footage from dslrs ?
10:39
dmj_nova
on my development machine, yep!
10:40
FergusL
is it particularly powerful or something ?
10:41
dmj_nova
dual core sandybridge?
10:41
FergusL
not particularly ! that's cool then
10:41
FergusL
I mean, that Novacut can do that
10:41
FergusL
ffmpeg in Blender struggles to do so
10:42
dmj_nova
you can make it not be smooth if you make firefox eat all of your ram, but that's not Novacut's fault
10:42
dmj_nova
FergusL: I suspect blender is doing some extra kind of compositing or something that we aren't
10:42
dmj_nova
all we have to do is playback 1080P video
10:43
dmj_nova
It's roughly the same feat as playing DSLR footage smoothly in chromium
10:44
FergusL
yes but then there still are cuts in the edit etc... it needs to switch between files
10:44
dmj_nova
FergusL: Oh, we have a little html cheat for that
10:45
dmj_nova
it works far better than it should
10:45
FergusL
yes I remember ! we talked about it
10:46
dmj_nova
Yeah, the switching clips is very much a problem that not all video editing software has solved
10:46
FergusL
absolutely
10:46
FergusL
and that's where the quality of the internal data model comes in, and with it caches and more ride along
10:49
dmj_nova
in our testing, the fluidity of the switchover between clips is very important for judging the pacing and feel of an edit
10:49
FergusL
yes, it boils down to frame synchronisation too
10:50
dmj_nova
if you don't have both, you're guessing at what you've actually edited
10:51
FergusL
yes !
10:52
dmj_nova
I once used novacut to discover that another program had inserted 13 black frames between various cuts in a 70 slice edit
10:52
dmj_nova
this was entirely the fault of sloppy use of time
10:53
FergusL
haha yes
10:53
FergusL
(sometimes it's also just the editor, by the time I used Premiere Pro, I remember sometimes messing and entire edit by slightly sliding shots)
10:53
FergusL
messing a whole edit*
10:54
dmj_nova
I do quite like how Blender uses frames rather than time
10:55
FergusL
yes, to me as a young user it's fine
10:55
FergusL
but some pros with a longer experience like to have timecode everywhere
10:56
FergusL
this is something else I did in Blender ! adding a tc display in the header
10:57
dmj_nova
well, there's a difference I think between giving the user timecode for reference, and using it as your data model
10:57
FergusL
ha !
10:57
FergusL
that ! yes
10:58
FergusL
are NLEs using tc in the data model ? I never wondered such thing but that's interesting
11:00
dmj_nova
I've seen NLEs use nanosecond timestamps in the data model
11:01
dmj_nova
nanoseconds are simultaneously too precise and not precise enough
11:03
FergusL
uh, nanoseconds !
11:05
dmj_nova
FergusL: It *sorta* makes sense if your data model is obviously basically a serialized gstreamer pipeline
11:06
dmj_nova
A bad idea, but one can understand the logic
11:06
FergusL
at the lowest level, gstreamer certainly has to be at nanosecond precision but...
11:34
dmj_nova
but you only convert to timestamps at the very final moment
11:35
dmj_nova
it's easy to go from samples and frames to perfect timestamps
11:35
dmj_nova
the other way is really, really hard
12:12
Bertl
good morning everyone!
12:13
Bertl
ad eagle vs kicad, I had a look at kicad (once again :) and it has significantly improved since last time I looked, but it is still missing a lot of features, nevertheless, I plan to properly convert everything from eagle to kicad
12:14
Bertl
the reason to stick with eagle for now was that I'm a lot faster (atm) with eagle and we wanted to get something usable asap
12:15
Bertl
and as FergusL found out the hard way, recent eagle versions use xml so there is no danger of not being able to read/convert the data
12:17
Bertl
the light version of eagle is available as freeware for linux, mac and windows, so it should be no problem for review either
12:25
Bertl
as I said, please review the design, I'm open for suggestions and I definitely appreciate any bug/potential-problem-reports
12:25
Bertl
http://vserver.13thfloor.at/Stuff/AXIOM/qad-center-v0.8.sch
12:25
Bertl
http://vserver.13thfloor.at/Stuff/AXIOM/qad-center-v0.8.brd
12:26
Bertl
I'm off for now, have to get rid of some bulk trash in the garden.
12:46
s3bastian
Bertl, let me know when you are back
12:46
s3bastian
I am now fully focused on PCB reviewing and would like to discuss what would make sense
12:46
s3bastian
for example comparing pinout of the cmv12000 datasheet to the schematic
12:53
Bertl
not back but I'll be around every now and then ... yes, pinouts should be verified for all components
12:53
Bertl
voltage levels for all involved pins
12:53
Bertl
proper connectivity on ground/power planes
12:53
s3bastian
ok will start with cmv12000 pinout
12:54
Bertl
also the physical layout, footprints and 3D space
13:34
s3bastian
1. why did you split the 1.8V supply pins on the cm12000 into two groups?
13:35
s3bastian
2. why does group 1 have 6 capacitors linking to ground and the second group 7 capacitors
13:37
s3bastian
I can't see valuable of any part (ohm values for resitors, etc.) do I need to change some view settings or is that intentional?
13:43
s3bastian
3. same split for 3.3V pins (E2 and G4) on separate 3.3V symbol - just cosmetics though
14:18
jucar2
joined the channel
14:20
jucar1
left the channel
14:21
intracube
joined the channel
14:24
Bertl
the 1.8V is split up for routing reasons, and because it might help with getting a stable supply, but basically to simplify routing
14:24
Bertl
ad 1) just because there are a different number of power pins in each group
14:24
Bertl
ah, that was ad 2)
14:25
Bertl
3) will have to wait, it shouldn't be separate nets
14:27
Bertl
if you mean the schematic, then that's fine, not all +3V3 have a visual connection, but they are the same net (use the all seeing eye from the sidepane and click on it :)
14:29
s3bastian
yes its just schematics for now ;)
14:29
s3bastian
sounds good
15:03
s3bastian
pin out of FMC and CMV12000 mostly checked
15:03
s3bastian
schematics page 1 http://picpaste.com/pics/schematic-review-page-1-V-01-sRbv1ced.1373641372.jpg
15:04
s3bastian
schematics page 3
15:04
s3bastian
http://picpaste.com/pics/schematic-review-page-3-V-01-BKpit8fc.1373641385.jpg
15:04
s3bastian
no errors found
15:05
s3bastian
just added a text note to the sch file about cmv12000 running 32 channel two sided readout mode
15:05
s3bastian
https://github.com/apertus-open-source-cinema/alpha-hardware/blob/master/SFE-PCB/qad%20center%20-%20eagle/qad-center-v0.8.sch
16:46
Bertl
so, back now ...
16:48
Bertl
ad values for parts: yes, they are still missing, but I'll add them via script
16:48
Bertl
the decoupling capacitors are all 100nF
16:49
Bertl
the resistors I have to calculate, but that doesn't matter for the layout
17:02
FergusL
(hint for ops : this channel has no topic)
17:03
s3bastian
ok great
17:04
s3bastian
ad topic, I think thats OK we all know where we are and why we are here ;)
17:15
Bertl
well, it won't hurt to have a topic though
17:16
Bertl
for example: "Axiom Alpha Now!"
17:20
s3bastian
why not :)
17:25
s3bastian
apparently I am not the channel owner ;)
17:27
s3bastian
changed nick to: Se6astian
17:28
ChanServ
has set the topic
17:28
Se6astian
hurray
17:28
Bertl
severe case of split personality :)
17:29
Se6astian
all these letters and numbers make my head swirl :)
17:30
Bertl
now get us a real topic!
17:30
ChanServ
has set the topic
17:31
Bertl
arrrr! but you're missing one :)
17:31
ChanServ
has set the topic
17:32
Bertl
much better! :)
17:32
Se6astian
so official!
18:24
Bertl
okay, uploaded version 0.8.1 which has most of the non 100n values, the reduced pmod connector (and therefore a larger power supply area), and all the fixes/enhancements incorporated which were suggested so far :)
18:27
Se6astian
perfect, can you commit it to GIT?
19:03
Bertl
nice one! :)
19:08
dmj_nova
left the channel
19:08
dmj_nova
joined the channel
19:45
s3bastian
joined the channel
19:47
Se6astian
left the channel
19:47
s3bastian
changed nick to: Se6astian
19:54
intracube
left the channel
20:03
Bertl
I've uploaded scands of the FMC connectors if somebody is interested ... they can befound here: http://vserver.13thfloor.at/Stuff/AXIOM/fmc_scan_600dpi.png http://vserver.13thfloor.at/Stuff/AXIOM/fmc_scan_1200dpi.png
20:14
FergusL
is there a size reference ?
20:22
intracube
joined the channel
20:37
dmj_nova
left the channel
20:37
dmj_nova
joined the channel
21:58
Se6astian
good night!
21:58
Se6astian
left the channel