Current Server Time: 01:07 (Central Europe)

#apertus IRC Channel Logs

2015/01/05

Timezone: UTC


23:18
troy_s
Hi Bertl
23:19
Bertl
hey troy_s! how's going?
23:33
troy_s
Bertl: flipping to desktop
23:34
troy_s1
joined the channel
23:34
troy_s1
Bertl, Did you get the highlight recovery issue based on my post?
23:35
troy_s1
I honestly am unsure it is currently an issue. I'm sure you recall our profile generation issues that had problems at the higher end of the sensor due to some issues like this (and the sensor of course misbehaving)
23:39
Bertl
well, first, I don't think we can do much in this regard on the raw data (which we are mainly interested in, no?)
23:40
troy_s1
I'm with you.
23:40
troy_s1
The "highlight recovery" is really a problem as well.
23:40
Bertl
i.e. '255'(8bit) means '255' or over board :)
23:40
troy_s1
In that we really have invalid data. I appreciate wanting to glean the most information from the thing, but when we are missing a channel or two, we really don't have any information there.
23:40
troy_s1
Worse it is confusing the actual data quite significantly as being display referred. IE We have _no_ idea what the actual scene referred value is.
23:41
troy_s1
Yep.
23:41
Bertl
shifting data down into the 8bit range is very problematic IMHO, and extending the bit size is something we can do in post anyway
23:41
troy_s1
I wonder if there is a way that one could profile a sensor such that
23:41
troy_s1
based on surrounding photosite ratios
23:41
troy_s1
if we can "detect' if it is pinned
23:41
troy_s1
Any ideas there?
23:41
troy_s1
Or does that totally not make sense?
23:41
troy_s1
(As in use the voltage spill and such?)
23:42
Bertl
nah, I think the best approach is to sacrifice a single value
23:42
troy_s1
I completely agree with you 110% regarding the problematic aspect, and this is even vastly more massive for doing image manipulation. You actually have erroneous luminance values.
23:42
troy_s1
Well yes. Avoid the damn pinned sensels.
23:42
Bertl
i.e. the highest possible value and call that 'overexposed'
23:43
troy_s1
Yes. A debayered RGB triplet with a pinned channel value is technically undefined.
23:43
troy_s1
I appreciate the desire to recover the other two but damn.
23:43
troy_s1
It simply isn't a display referred ratio.
23:45
troy_s1
God OpenCine is a bit of a ...
23:45
Bertl
I'm still not sure what T244 is about, is it to visualize post processing over exposure correction on the monitoring channel?
23:45
lab-bot
T244: image pipeline: highlight recovery - http://lab.apertus.org/T244
23:45
troy_s1
I think it is a desire to try and make use of overexposed debayer
23:46
Bertl
but we don't debayer the raw data? :)
23:46
troy_s1
But I'm unsure everyone grasps the implications entirely.
23:46
troy_s1
Debayer / Bayer doesn't matter
23:47
Bertl
well, the raw data is the raw data, how to interpret is up to somebody else :)
23:47
troy_s1
The essence seems to be an attempt to recover the unfilled photosites
23:47
troy_s1
But in truth, the unfilled photosites around a filled set isn't data.
23:47
troy_s1
LOL
23:47
troy_s1
True. I'd hate to see this become some sort of "YES ZOMFG WE NEED THIS" and it ends up default.
23:48
troy_s1
But also, the Beta isn't going to have raw. So :P
23:48
Bertl
oh yes, it is going to have raw :)
23:49
Bertl
but I guess I understand now what T244 is all about, it is not intended for the raw output only for the 'cooked' output
23:50
troy_s1
How do you intend to do a debayer for view etc?
23:50
troy_s1
Just pin all of the channels?
23:50
troy_s1
And call it display referred white?
23:50
Bertl
that hasn't been decided yet, the first stage will be similar to the Alpha (poor man's debayer)
23:55
troy_s1
Even the poor man's needs to make a decision on the pinned bins no?
23:55
Bertl
no
23:56
Bertl
we currently just combine 4 sensel and call it a day
23:56
troy_s1
That ends up with completely batsh*t colors when they have pinned bins no?
23:57
Bertl
somewhat yes, that's why we have the second stage clipping active after color transformation
00:50
jucar
joined the channel
00:53
fsteinel
joined the channel
00:57
fsteinel_
left the channel
01:11
skinkie
left the channel
01:24
intracube
left the channel
05:59
troy_s1
left the channel
06:00
davidak
joined the channel
06:01
Bertl
off to bed now ... have a good one everyone!
06:01
Bertl
changed nick to: Bertl_zZ
06:34
davidak
left the channel
11:10
HAL-1999
joined the channel
11:11
HAL-1999
Hello guys. Anyone here?
11:15
HAL-1999
I'd like to try to contribute to the case design and other mechanical stuff if possible. What do I need to do to catch up with the current work?
11:16
HAL-1999
From the video I gather there is a manufacturing company that did the prototype. Are there any CAD files available?
11:18
mars_
the best way is to talk to se6astian|away or Bertl_zZ (both are away right now) and to take a look at http://lab.apertus.org/
11:21
HAL-1999
I was lurking there for the last two weeks but there are very few things about the mechanical aspect. I'd like to meet the team and start doing work.
12:16
intracube
joined the channel
12:33
philippej
joined the channel
13:03
se6astian|away
changed nick to: se6astian
13:03
se6astian
good day
13:03
se6astian
hi HAL-1999!
13:04
se6astian
We built the prototype ourselves
13:04
se6astian
laser cut acrylic glass enclosure :)
13:05
se6astian
The AXIOM Beta related CAD files are here: https://github.com/apertus-open-source-cinema/beta-hardware/tree/master/Enclosure/Version%20002
13:06
se6astian
currently designed in Inventor but I can export STP, IGES, etc.
13:14
se6astian
email me if you have more questions, apertus.org/contact
13:14
se6astian
gotta go to a CAD design meeting for the enclosure :)
13:14
se6astian
changed nick to: se6astian|away
13:24
HAL-1999
thaks se6astian I'll try to import these in solidworks.
13:24
HAL-1999
Was there any discussion about the modularity of the enclosure? This is kinda my expertise...
13:25
HAL-1999
Also if there is a meeting I'm interested.
13:34
Bertl_zZ
changed nick to: Bertl
13:35
Bertl
morning folks!
13:36
HAL-1999
hi
13:36
Bertl
HAL-1999: it is very likely the "local" meeting of the ergonomics group (or similar)
13:36
HAL-1999
oh.. ok.
13:36
Bertl
so unless you are in Austria (nearby Vienna) it won't be easy to be there
13:36
HAL-1999
I'm trying to get up to speed
13:37
Bertl
but I'm sure your input is very appreciated
13:37
HAL-1999
Well, I'm working on my own thing which also include a modular stacking design
13:38
HAL-1999
it has similar form factor too, but nothing I've put online yet
13:38
Bertl
so maybe get started with that and show your designs?
13:40
HAL-1999
Yes this will happen in time.
13:42
HAL-1999
The thing is, if my project and the Appertus camera happen to be similar enough mechanically there might be some benefit in defining a common form factor.
13:42
HAL-1999
This would be especially beneficial to science folks
13:42
HAL-1999
especially the scientific imaging people
13:43
Bertl
maybe
13:45
HAL-1999
I have no idea where you guys stand on this of course.
13:46
HAL-1999
I'd like to present some work which we can talk on.
13:46
Bertl
very much depends on your "project" and the common form factor, I guess ...
13:48
HAL-1999
There are many form factors out there, some of which are not even open source and the ones which are open source do not play very well with each other
13:49
HAL-1999
And all of them deal purely with PCB form factors. There is nothing much mechanicak out there
13:50
HAL-1999
Ecept the 19" rack, the optical bench assemblies and a few others.
13:50
HAL-1999
I am researching a way to bring all these together in an open source form factor
13:54
designbybeck
joined the channel
13:57
surami
joined the channel
13:57
HAL-1999
Here is some conceptual work done by NASA in 95
13:58
HAL-1999
http://trs-new.jpl.nasa.gov/dspace/bitstream/2014/30351/1/95-0940.pdf?origin=publicationDetail
13:58
HAL-1999
It shows some aspects for modular (stacked) electronic equipment
13:59
HAL-1999
Some of the ideas there have been implemented commercially
13:59
HAL-1999
I think we could do it better justice in the open source community
14:07
Bertl
looking forward to see how you map that to a camera
14:08
surami
left the channel
14:10
HAL-1999
I believe it is possible. In the prototype ypu already show a modular assembly with components stacked behind the sensor module.
14:10
HAL-1999
Not only possible but it could ve cheap to produce too
14:11
HAL-1999
There was something about you looking for a CNC mill?
14:12
Bertl
correct
14:12
HAL-1999
Have you looked into Deckel-MAHO machines? They are quite local to your area and should be available cheaply. Many people convert them to CNC and use them in production..
14:13
Bertl
we decided to go for the Tormach
14:13
Bertl
but you might want to add that with a link to the wiki for future reference
14:14
HAL-1999
Deckel FP1 would be perfectly suitable. They are excellent machines.
14:15
HAL-1999
Would do, thanks.
15:04
philippej
left the channel
15:17
se6astian|away
changed nick to: se6astian
15:42
Bertl
changed nick to: Bertl_oO
15:48
se6astian
HAL-1999: you still around?
15:48
HAL-1999
yes I'm here
15:48
HAL-1999
looking at the files
15:48
se6astian
ah great
15:49
se6astian
let me know if you have any questions
15:49
se6astian
the current concept is at the bottom of https://wiki.apertus.org/index.php?title=Ergonomics_Workgroup
15:49
se6astian
it plans the IO modules to be plugged in with two PCIe connectors
15:50
HAL-1999
I do have a few questions actually. There are two PCBs with a cut out. I suppose they stay besides the lens mount. What are they for?
15:51
se6astian
you mean the rounded centers?
15:52
se6astian
https://wiki.apertus.org/index.php?title=File:PCBModuleLeft01.jpg
15:53
HAL-1999
yes that one
16:00
se6astian
that is obsolete and not part of the current assembly anymore
16:01
HAL-1999
Ok, glad to hear that.
16:02
HAL-1999
What about the PCIe? Does that require a license?
16:12
HAL-1999
PCI-SIG says "Your company must be a member of this organization for you to gain access to the members area of this site."
16:13
HAL-1999
So I can't even download the spec to see if it requires a licence or not -_-
16:13
HAL-1999
I presume it requires at least a membership fee.
16:14
aombk
joined the channel
16:16
HAL-1999
which appears to be 3000$ annually
16:16
se6astian
we are not doing PCie
16:16
se6astian
we just use a PCIe connector
16:17
HAL-1999
Oh, ok sorry.
16:18
HAL-1999
What is the communications bus then?
16:19
se6astian
LVDS
16:19
se6astian
no bus
16:19
se6astian
just all spare IO lines routed to the PCie connector
16:20
HAL-1999
Hmm interesting
16:20
troy_s
se6astian: How goes the housing iterations at the meetings?
16:25
HAL-1999
Have you looked at PC/104e ? Their licencing is quite open as far as I remember and they provide a bus. Not sure if it is up to the task though, might be worth checking out.
16:32
HAL-1999
Correct name is PCIe/104 actually --- http://www.pc104.org/pci104e_specs.php
16:35
aombk
left the channel
16:44
davidak
joined the channel
17:11
troy_s
left the channel
17:12
mithro
left the channel
17:15
troy_s
joined the channel
17:17
mithro
joined the channel
17:18
se6astian
we did look at PC104 but see no immediate application here
17:18
se6astian
troy_s: good good, but not what you are expecting currently I guess
17:18
se6astian
we are building a 3d printed fixture for the very first prototype
17:18
se6astian
not really an "enclosure" yet
17:19
Bertl_oO
changed nick to: Bertl
17:19
Bertl
back now ...
17:20
HAL-1999
I only mention PC/104 because it has established component stacking.
17:21
HAL-1999
All similar form factors require a backplane or something similar to function, which limits the size.
17:21
HAL-1999
PC/104 can expand by stacking boards without a backplane.
17:38
lab-bot
sebastian-test committed rBH0c163dec7811: added updated cmosis datasheets (authored by sebastian-test).
18:08
troy_s
se6astian: I don't really have any expectations to be honest. I just was looking at the cupboard and trying to estimate where / why it goes that direction
18:11
troy_s
(By the way, I think the larger housing adapter is helluva clever if it manages to meet the needs of its existence.)
18:23
se6astian
changed nick to: se6astian|away
20:10
HAL-1999
left the channel
20:18
davidak
left the channel
20:19
se6astian|away
changed nick to: se6astian
20:26
davidak
joined the channel
20:37
se6astian
troy_s: the concepts are going in a lot of different directions, basically anyone having an idea can pursue it
20:37
se6astian
some concepts might never make it to actually being built
20:37
se6astian
but there is no dictation what people should/shouldnt do
20:38
troy_s
se6astian: I think you and I agree that the designs for the Cupboard aren't neutral (good thing â¢)
20:38
troy_s
se6astian: Cupboard is definitely moving forward I hope?
21:28
davidak
left the channel
21:37
se6astian
if christoph is motivated and receives good feedback I guess he will continue yes :)
22:01
designbybeck
left the channel
22:14
lab-bot
left the channel
22:16
lab-bot
joined the channel
22:20
lab-bot
left the channel
22:20
lab-bot
joined the channel
22:30
Topic
apertus° - open source cinema | www.apertus.org | join the apertus° Lab: http://lab.apertus.org/ | IRC Logs available at: http://irc.apertus.org
22:30
se6astian
has set the topic
22:38
lab-bot
left the channel
22:56
Topic
apertus° - open source cinema | www.apertus.org | join the apertus° Lab: http://lab.apertus.org/ | IRC Logs available at: http://irc.apertus.org
22:56
se6astian
has set the topic
22:57
Bertl
joined the channel
22:59
se6astian
joined the channel
22:59
philippej|away
joined the channel
22:59
philippej|away
changed nick to: philippej