Current Server Time: 22:17 (Central Europe)

#apertus IRC Channel Logs

2017/01/16

Timezone: UTC


00:00
Bertl
up to the point where one channel is overexposed, the linear data is perfect
00:00
Bertl
and after that point, we just don't have any data
00:00
troy_s
Except the moment one is full, you're into non-data.
00:00
arpu
left the channel
00:01
danieeel
joined the channel
00:01
Bertl
and to me that is exactly what is happening in your gif
00:01
troy_s
The more accurate technique is taking that "correct" data and decreasing saturation along the axis so you don't end up with whacked values like in the Canon example.
00:01
troy_s
Which is unlike some of the more sophisticated cameras which deliver a usable encode right off of the metal machine.
00:02
Bertl
so we are talking about encoding here not improving the raw data?
00:02
Bertl
i.e. what some encoders do wrong in some proprietary cameras?
00:03
troy_s
Correct. The encoding.
00:03
troy_s
(as in we should never get that yellow bowl.)
00:03
danieeeel
joined the channel
00:03
troy_s
(instead, the encoded values should be colorimetrically correct)
00:03
Bertl
well, we will still get this yellow bowl, if the red channel tops out while the blue and green is still in range
00:03
troy_s
Nope.
00:04
Bertl
yep
00:04
troy_s
Because you've properly saturated the other two wells before.
00:04
troy_s
To hold the chromaticity axis.
00:04
danieel
left the channel
00:04
Bertl
what you explain here is color guessing :)
00:04
troy_s
Nope.
00:05
Bertl
if you have data (i.e. are not saturated) then the data is fine without any LUT or modification
00:05
troy_s
Not really because you are then hard cutting.
00:05
Bertl
if you cross the treshold and one channel is saturated, you don't know 'how much' so you can't compensate
00:05
troy_s
Which is undesireable.
00:06
troy_s
The idea is that you don't need to guess after one sensel is pinned.
00:06
troy_s
As the moment one is pinned, your other two are pinned.
00:06
troy_s
So you are gracefully ascending up to non-data versus leaving the residue of non-data.
00:06
troy_s
(and leveraging the actual data to get there.)
00:06
Bertl
so you want to paper over it?
00:06
danieeel
left the channel
00:07
troy_s
It isn't papering. It's actually providing an extremely important transform.
00:07
Bertl
i.e. let everything go 'white' at the upper range?
00:07
troy_s
*and* delivering a much more aesthetic result.
00:07
troy_s
It isn't a hard cut.
00:07
Spirit532
is there a live demo of that?
00:07
troy_s
Remember you'd be gracefully desaturating (correctly colorimetrically)
00:07
Spirit532
as in, going up in saturation
00:08
troy_s
Spirit532: There's plenty in some of the work I've been doing on Filmic.
00:08
Bertl
while I agree that this is a good idea to avoid strange color effects, it will mess up the actual data
00:08
troy_s
Bertl: The actual data is messed up the other way.
00:08
troy_s
in that you aren't capturing a proper representation of the scene within the confines of a device referred encode.
00:10
Bertl
I have to disagree here, but I get now what you mean and I agree that it gives 'more aesthetic' results
00:10
troy_s
Bertl: The problem is that A) having a hard cut delivers incorrect results.
00:10
Bertl
and it should be easy to implement for a preview for example
00:10
troy_s
Let me dig up three examples.
00:11
ItsMeLenny
joined the channel
00:11
Bertl
the hard cut needs to reserve at least one value for 'overexposed' (aka non-data)
00:11
Bertl
still won't give you pretty pictures :)
00:13
troy_s
Bertl: The hard cut is the termination of your data.
00:13
Spirit532
Gotta go to bed. Gn.
00:13
Bertl
nn
00:13
troy_s
Bertl: So you are much more prudent to ramp up to it, preserving the xy axis, than the precise xy.
00:13
Bertl
yep, for pretty pictures I agree, but you are basically mutilating perfectly fine data for it
00:14
troy_s
It's not just pretty pictures.
00:14
Bertl
because the data you get in this 'compression' range will simply be wrong
00:14
troy_s
As you can simply push the saturation back out as required.
00:14
troy_s
Nope. It's the same.
00:14
troy_s
That's the point.
00:14
troy_s
You can in fact push out saturation in a colorimetrically correct fashion.
00:15
troy_s
So while you are compressing values, you are doing so in the proper manner, as opposed to relying on well saturation to do it inappropriately.
00:15
Bertl
let's make an extreme example here
00:15
Bertl
let's say you start compressing at 240 up to 255
00:16
Bertl
and let's also say the LUT is a simple 'reduce missing values by proportion'
00:16
Bertl
i.e. if you have 100/240/50 then the transform gives you 100/240/50
00:17
Spirit532
left the channel
00:17
Bertl
but if you have 100/248/50, then you half the distance for the other two channels
00:17
troy_s
Sure. A simple LERP.
00:17
Bertl
i.e. 100+155/2/248/50+255/2
00:18
Bertl
that will give you a simple 'fade to white' before overexposing
00:18
troy_s
Sure.
00:18
Bertl
that is what you mean as an example, right?
00:19
Bertl
of course, it will be a little more complex when two primaries reach the 240 value
00:20
Bertl
but, for the sake of the argument, let's assume we only have one primary in the critical range
00:20
troy_s
Yes.
00:21
Bertl
so we map 100/248/50 -> 177/248/177
00:21
Bertl
sorry 100/248/50 -> 187/248/177
00:22
danieeeel
changed nick to: danieel
00:22
Bertl
but we _know_ that the actual sensel information is linear and was correct (no primary is overexposed at 100/248/50)
00:23
Bertl
so, we now have the correct data 100/248/50 and incorrect data 187/248/177, which just happens to fade nicely to white when we near overexposure
00:23
Bertl
looks very pleasing, but is nevertheless wrong, no?
00:25
troy_s
It's certainly "wrong" but look at the alternate in terms of an image; the hard cut.
00:25
troy_s
Which from a photographic vantage is worse.
00:26
Bertl
I agree, but the question is, do you win anything by doing this 'compression' on the raw data?
00:26
Bertl
because you can always do this kind of operation on the raw data at any later point
00:26
Bertl
OTOH, you probably can't undo it, except for very specific cases
00:28
troy_s
My vantage would be from the end "what you see in the encoded data" it makes more sense to do it.
00:28
troy_s
(as in take a step out)
00:29
Bertl
as I said, for e.g. preview or similar I think it's an excellent idea
00:30
Bertl
for modifying raw data before recording, I think it's a very bad idea :)
00:36
Bertl
but I can also see why cameras with 'lossy compression' would go this route, because the 'cut off' poses a big problem for any (lossy) compressor
00:50
jucar
left the channel
01:21
jucar
joined the channel
01:22
jucar1
joined the channel
01:26
jucar
left the channel
02:03
Bertl
off to bed now ... have a good one everyone!
02:03
Bertl
changed nick to: Bertl_zZ
02:47
jucar1
left the channel
04:08
jucar
joined the channel
04:19
ItsMeLenny
left the channel
04:20
ItsMeLenny
joined the channel
04:44
jucar
left the channel
08:10
Spirit532
joined the channel
08:11
dimaursu16
joined the channel
08:28
ItsMeLenny
left the channel
08:29
ItsMeLenny
joined the channel
08:59
se6astian|away
changed nick to: se6astian
09:17
dimaursu16
left the channel
09:36
Bertl_zZ
changed nick to: Bertl
09:36
Bertl
morning folks!
10:07
Spirit532
left the channel
10:35
jucar
joined the channel
11:07
jucar
left the channel
11:25
Spirit532
joined the channel
11:55
simonrepp
joined the channel
11:57
simonrepp
helloo, i'm a bit ill currently, can anyone jump in to host the irc meeting today in the evening?
11:58
simonrepp
(i'll take care of the announcement however, if hosting is secured :D)
11:58
se6astian
will do
11:59
simonrepp
super \o/ many thanks
12:01
se6astian
you are welcome
12:04
simonrepp
is announced!
12:14
simonrepp
thanks again, i'm off! cheers
12:14
simonrepp
left the channel
12:29
Bertl
off for now ... bbl
12:29
Bertl
changed nick to: Bertl_oO
12:37
ItsMeLenny
left the channel
12:37
ItsMeLenny
joined the channel
12:53
Spirit532_
joined the channel
12:57
Spirit532
left the channel
13:42
Spirit532_
changed nick to: Spirit532
13:48
ItsMeLenny
left the channel
14:39
davidak
joined the channel
14:58
arpu
joined the channel
16:36
RexO
joined the channel
17:35
dimaursu16
joined the channel
17:35
dimaursu16
left the channel
17:35
dimaursu16
joined the channel
17:40
dimon_
joined the channel
17:40
dimaursu16
left the channel
17:42
davidak
left the channel
17:46
se6astian
now popular: 开源才是主流:Open-source Axiom Beta 4K Cinema Camera http://xilinx.eetrend.com/news/10884
17:47
se6astian
my favourite part is 此外,在短片中出现比较多的一个点是目前正在设想的摄像机外壳的一个透视图,如下图所示
17:56
davidak
joined the channel
17:57
davidak1
joined the channel
18:00
Bertl_oO
changed nick to: Bertl
18:00
se6astian
IRC meeting time!
18:01
se6astian
Simon is unfortunately not available today so I will try to follow in his footsteps
18:01
se6astian
as this method worked very well please PM me now if you want to give an update about your work area
18:01
se6astian
we will then give updates one after the other
18:02
RexOrCine
joined the channel
18:04
RexOrCine
Here.
18:05
se6astian
right, 4 speakers: bertl, davidak, RexOrCine and me
18:05
se6astian
Bertl please go ahead
18:05
Bertl
okay, tx
18:05
Bertl
I had some setbacks over the last few weeks but nothing critical (hardware problems, etc)
18:06
Bertl
Not on the Beta though, but with comodity PC hardware :)
18:06
Bertl
Nevertheless, I managed to build a new SD image for the Beta
18:06
Bertl
which contains an updated Arch Linux as well as an up-to-date Kernel
18:07
davidak
can you provide a download link for those who have a dev kit?
18:07
Bertl
The device tree was rebuilt from scratch to make it more general and support some future enhancements
18:08
Bertl
the image is already up, but I would suggest to wait a few more days as I don't have Beta hardware to test it on and Sebastian had some feedback
18:08
Bertl
so we will iron out potential issues over the week and then upload another image which will hopefully last for some time
18:09
Bertl
I also updated my build system to the latest Vivado, Diamond, iVerilog, KiCad, Eagle, etcc
18:09
Bertl
so that we can do some testing on those to prevent problems
18:10
Bertl
I almost finished the ESP32 shield (only power needs to be routed), so we will soon have a new shield to test and hopefully build :)
18:10
Bertl
and together with se6astian we did a session last week to help our assembly team to improve on hand soldering those parts who need it
18:11
Bertl
that's it from my side ... any questions?
18:11
davidak
how big is the assembly team?
18:12
Bertl
currently we have six people if I counted right
18:12
davidak
great
18:12
Bertl
but time coordination is still a problem
18:12
Bertl
might improve in the near future though
18:13
se6astian
many thanks, taking notes for meeting summary in parallel
18:14
se6astian
I count 7 people in the assembly team if we include me and bertl
18:15
Bertl
I didn't include us :)
18:15
RexOrCine
davidak The man in the photo that was tweeted earlier is Manfred I believe.
18:15
se6astian
yes
18:15
se6astian
ok davidak you turn please
18:15
Bertl
+r
18:16
davidak
we don't had any meetings with the website redesign team.
18:17
RexOrCine
Possibly because not much is going on at present.
18:17
se6astian
and I think most if not everything regarding content is actually decided/discussed already
18:17
RexOrCine
It's just getting someone to pull their finger out.
18:17
davidak
currently we see no progress with the website theme, so we called for some help
18:18
davidak
https://twitter.com/ApertusOSCinema/status/821018109624598529
18:18
davidak
se6astian: can you add the current result?
18:19
RexOrCine
This is always an issue, because this kind of thing takes time. Even when money is involved you typically have to pressure a dev.
18:19
Bertl
pressure doesn't work well on open source developers :)
18:19
davidak
normally you have deadlines and communicate progress
18:19
se6astian
current results: one agency and one person offered help via twitter
18:19
se6astian
one person via email after seeing the twitter post
18:20
se6astian
one person via facebook
18:20
RexOrCine
It may be that this guy I'm dealing with currently is familiar with drupel, I'll be mentioning it later.
18:20
RexOrCine
When there was bad feedback Re Drupel were any alternatives mentioned?
18:20
RexOrCine
Am I right in thinking that these people expect to be paid?
18:21
se6astian
currently it seems like 3 out of these 4 contacts expect this to be a paid gig
18:21
se6astian
I ll contact them all to clarify the situation though
18:22
se6astian
using another CMS is not really up for discussion at this point anymore
18:23
RexOrCine
The guy that was dabbling recently, is he a Uni guy?
18:23
se6astian
davidak anything else to report?
18:23
davidak
there was already some work on new flyers to print. the content is already pretty ready, but we need a designer to have a good first impression.
18:24
RexOrCine
I'd like to see those.
18:25
se6astian
davidak you mentioned the OS guy who offered paid graphics design help recently
18:25
davidak
my idea was that we could pay a designer to help us. like http://ura.design/ (helping open source projects, pay what you can basis) or ask here http://opensourcedesign.net/jobs/
18:25
RexOrCine
I take it you're not talking about the work that I did.
18:25
davidak
that needs to be discussed if we have the money and want to pay others to work for us
18:27
davidak
se6astian: can you add some more info about the flyer redesign?
18:27
se6astian
sure: there is a shared google doc where we collect references and current process with text, layout, graphical elements
18:28
se6astian
I suggested we discuss this topic of paying someone else here as first step and with the community/board as well
18:28
RexOrCine
So where are these new flyers being worked on?
18:28
se6astian
so far our policy has been that we do not pay anyone a "salary" in the project from crowd funding money
18:28
se6astian
including us here :)
18:28
se6astian
RexOrCine: you have access to this google doc, there is no other place
18:29
se6astian
so it would require a general consensus about what ground rules we want to set up for paying for services
18:29
RexOrCine
Ah, so the new flyers must be what I did. I will say, I don't think it's a good idea to settle on any concepts until we have a hi-key image of the AXIOM Beta.
18:29
se6astian
as this is a very difficult topic
18:29
se6astian
who gets paid, who doesnt get paid, why not me, why this guy, etc....
18:29
davidak
we could ask here if someone want to give some money for that. like a internal mini crowdfunding. i'm not sure what is a fair salery for such work
18:30
RexOrCine
Well I took 230 quid net in 2016 so I'm out.
18:31
Bertl
I think we can pay a 'service' as apertus (from the donations, etc) if we all agree that we need it done urgently or by a third party
18:31
RexOrCine
Personally I don't think shelling out money is required.
18:31
Bertl
I think it would be a good idea though to check alternatives first
18:32
Bertl
but that's just me, that is something the board/community has to decide
18:32
RexOrCine
Has davidak got access to this google doc with the flyers?
18:32
davidak
yes, i already commented
18:33
davidak
and did a review
18:33
se6astian
its definitely a topic we will not resolve today, but we can set an outline of how to arrive at such decision
18:33
RexOrCine
Ah... The reverse sides or internals can be worked on. It's a case of what text, images and or diagrams are included. When it comes to the front a picture of the Beta on a white background is important.
18:33
se6astian
eg. what requirements must be met
18:34
RexOrCine
... and we're waiting for Max for that.
18:35
davidak
i think later, when we sell products we will have more money to fund further development and get help from professionals and pay our contributors
18:36
RexOrCine
When we did the project overview... (remember the article that went to print) would there be any updates required in that document in your estimation se6astian / Bertl?
18:36
davidak
and fund other people who also help improve open source (software, hardware, ...) :)
18:36
RexOrCine
ie... would anything need to be changed / included?
18:37
se6astian
lets finish the money topic first
18:37
davidak
that was everything from my side to report.
18:37
Bertl
thanks davidak!
18:37
se6astian
I outlined a few points as proposal for the spending procedure
18:37
RexOrCine
The reason I ask is because, when it comes to deciding upon what's included on the inside / reverse of a flyer, it might be worth our while pulling text from that docuement.
18:38
se6astian
What are the requirements that we pay a third party for a service with donation/crowdfunding money:
18:38
se6astian
-) The service must be required urgently
18:38
se6astian
-) There must be no alternative
18:38
se6astian
-) Reasonable effort to find someone do it pro-bono must have been made
18:38
se6astian
-) The association board needs to consent before spending any money
18:38
se6astian
anything to change, remove, add?
18:38
Bertl
I think that we should not use crowfunding money for anything not related to the funding
18:39
Bertl
which IMHO rules out flyers for example
18:39
se6astian
thanks davidak for your points
18:39
se6astian
Bertl: good point, adding
18:39
davidak
se6astian: the requirements sounds good
18:40
RexOrCine
-) Any decisions should be based around what money is available and from which channels any money might be channelled from.
18:41
se6astian
of course - we cant spend money that we dont have obviously
18:41
se6astian
ok RexOrCine your turn now
18:42
RexOrCine
I'd be keen to know if any updates are required in the article document.
18:43
RexOrCine
Se6asian and I have been working on the forum. Which can be accessed here: https://forum.apertus.org/index.php
18:44
RexOrCine
There are still some design issues, but it's safe to register if you're reading and haven't already done so.
18:44
se6astian
is it ok if we invite the team first to test/use it
18:44
se6astian
then axiom-dev, backers, community
18:44
se6astian
and then general public
18:44
Bertl
and it seems to be HTML confirm! yay! \o/
18:44
Bertl
*conform
18:45
se6astian
with one week between each stage?
18:45
davidak1
fine for me
18:45
RexOrCine
In the interim between website redesign we're going to need to link toward the forum from the site homepage.
18:45
Bertl
but the CSS needs some work, 67 errors
18:45
se6astian
RexOrCine: please send me the URL of the article/document to review directly afterwards
18:45
RexOrCine
Ten4
18:46
se6astian
do you want to invite the team or should I do it?
18:47
RexOrCine
I'll let you do that se6astian as your methods are established.
18:47
se6astian
ok will do
18:48
se6astian
anything else from your side?
18:49
RexOrCine
That's pretty much it. I'll be liaising with respect to bugs over the course of the next two weeks I'd imagine.
18:50
se6astian
great
18:50
se6astian
ok quick update from me:
18:50
se6astian
-) http://notes.apertus.org/ is still not operational
18:50
se6astian
I found nobody who could help us fix it so far
18:50
se6astian
documentation is sparse,
18:51
se6astian
-) on the Pick and place front
18:51
se6astian
I placed a set of different size components pretty successfully recently
18:51
se6astian
you received the results via email
18:51
se6astian
I am now in the process of replacing the liteplacer top and bottom vision cameras
18:52
se6astian
bottom vision camera arrived already and the 3d printer is printing parts for the holder right now
18:52
se6astian
hexagonal led lights (http://malte-randt.de/hexagonal-led-light/) are ordered
18:52
se6astian
top vision camera inbound, holder already delivered
18:53
se6astian
I also created a 3d printable semi manual tape feeder
18:53
se6astian
https://cad.onshape.com/documents/09670830aaa709a325eec93c/w/1010c19833535d6d8bf3472f/e/9442d1d49c729f510bf71026
18:53
se6astian
first tests are very promising
18:54
se6astian
still waiting on cnc milled tray base to really use it
18:54
se6astian
I also made some progress with openpnp and liteplacer documentation:
18:54
se6astian
https://wiki.apertus.org/index.php/Liteplacer_PnP_Machine_Notes
18:55
se6astian
optical homing, nozzle changer, etc.
18:55
Bertl
I only see a strip of tape at onshape :)
18:55
se6astian
you need to switch to the main assembly tab at bottom
18:55
se6astian
and pushed some small open openpnp code contributions, displaying which nozzle is currently attached, crosshair improvements, etc.
18:56
se6astian
regarding the new firmware image creation I added my wishes/todos here
18:56
se6astian
https://wiki.apertus.org/index.php/AXIOM_Beta/AXIOM_Beta_Software/Firmware
18:56
se6astian
an indian university student contacted us to work on cross compiling the beta internal software automaticaly
18:57
se6astian
I have not heard anything back though
18:57
se6astian
will need to follow up
18:57
se6astian
Phil Kerrs AXIOM Remote Plus testing progress is collected here: https://wiki.apertus.org/index.php/AXIOM_Remote_Plus
18:58
se6astian
currently the last remaining step is to figure out the serial USB connection between remote and beta after the devicetree changes the latest firmware image brought
18:58
se6astian
herbert will help me in that regard soon
18:58
se6astian
on the hardware assembly front we added a new guy to the team last week
18:58
se6astian
I trained him locally and he already assembled two mainboards then
18:59
se6astian
he also attended to hand soldering documentation so it was half training half documentation
18:59
se6astian
on the wiki we now also have a price estimate that outlines how much it costs to build a Beta yourself
18:59
se6astian
https://wiki.apertus.org/index.php/AXIOM_Beta/AXIOM_Beta_Prices
18:59
se6astian
its not 100% complete yet though
18:59
se6astian
but growing
19:00
se6astian
thats it from my side
19:01
Bertl
thanks!
19:02
se6astian
ok exactly 1h meeting time is already pretty long, so lets close todays meeting if there are no further general remarks/topics
19:03
davidak1
thanks everyone
19:03
se6astian
next meeting time?
19:03
se6astian
2weeks monday 19:00 again?
19:03
se6astian
I see I have an appointment on the 30th at 19:00
19:04
se6astian
a "material regulars table"
19:04
se6astian
dealing with new materials and compounds for producing products, architecture, etc.
19:04
se6astian
but its an offline meeting
19:05
se6astian
Tue 31st of january?
19:05
davidak1
ok for me
19:05
Bertl
fine with me
19:05
se6astian
great
19:05
se6astian
noted
19:06
se6astian
ok, thanks everyone for attending, meeting closed
19:07
RexOrCine
Good stuff
19:24
davidak1
i'm curious if simonrepp has already contacted marcin from opensourceecology back. i really like his vision and see great progress on his projects. we can probaply learn a lot from him and get helpful feedback on our designs. for example he does google hangouts with people to discuss https://youtu.be/QXoGIEG6-Xs
19:24
davidak1
(do you got the whole message including the link?)
19:25
Bertl
yup
19:27
Bertl
very strange and somewhat confusing ... :)
19:31
davidak1
if you don't know them, this is their current project: http://openbuildinginstitute.org/
19:32
davidak1
everything open source
19:33
davidak1
also a wider vision for a whole open source ecology http://opensourceecology.org/wiki/Main_Page
19:33
davidak1
they are building a 3D printer for example
19:33
Bertl
yeah, I know the open source ecology
19:34
davidak1
and everything else to build a civilization :)
19:35
davidak1
his TED Talk is very inspiring (and short) http://www.ted.com/talks/marcin_jakubowski?utm_source=tedcomshare&utm_medium=referral&utm_campaign=tedspread
19:37
davidak1
he posts a lot of raw videos of the progress on youtube: https://www.youtube.com/user/marcinose/videos
20:36
davidak
left the channel
21:01
dimon_
left the channel
21:56
RexO
Posting that here, it's pretty much lost, on the forum it might get some traction.
21:57
RexO
left the channel
22:06
Spirit532
left the channel
22:21
se6astian
changed nick to: se6astian|away
23:29
ItsMeLenny
joined the channel