Current Server Time: 18:00 (Central Europe)

#apertus IRC Channel Logs

2014/09/11

Timezone: UTC


01:00
Seku
ive been slitently following it since the first sensor test... liked it D
01:01
Seku
well, the public test
01:03
Seku
doesnt mean much tho
01:08
intracube
blender's compositor might be an option for some colour work
01:08
intracube
there's also an integrated NLE, but mileage may vary with that :)
01:11
ludidonkihot
left the channel
01:14
tyrone_
left the channel
01:16
Samuel
So if I get it right, with an axiom, I need: a rig, a HDMi recorder, external audio?
01:19
Bertl
with the basic AXIOM Beta as planned, yes, at least for recording raw movies at high resolution or high framerate
01:20
Bertl
you won't need the external audio or HDMI recorder for e.g. still photography :)
01:21
Samuel
Bertl: for still photography, how would the pics be stored ?
01:22
intracube
Bertl: was any stabilisation used on the aerial footage?
01:22
Bertl
on one of the three microSD cards, or sent via ethernet or on an USB stick
01:22
Samuel
Bertl: OK, thank you
01:23
Samuel
'nite all
01:23
Samuel
left the channel
01:23
Bertl
intracube: as far as I know the cable buggy has some kind of stabilization
01:24
intracube
it's really nicely done
01:24
Bertl
according to se6astian, it was really hard work
01:25
Bertl
somehow the USB remote interface didn't work as expected, so he had to make adjustments to the camera standing on the tower reaching out for the axiom USB/Ethernet port :)
01:26
Bertl
https://twitter.com/ApertusOSCinema/status/469491718279421953
01:27
intracube
oh, so it wasn't a quadcopter/hexacopter that was used?
01:28
Bertl
no, a cable based system
01:28
intracube
aah, I see
01:34
intracube
https://twitter.com/ApertusOSCinema/status/470173623379329024/photo/1
01:35
Bertl
yup
01:36
Bertl
the funny detail was, the Atomos had real problems recording under those conditions, while the Alpha did just fine :)
01:36
Bertl
we got a huge number of partial files and corrupted data on the filesystem (probably due to vibrations)
01:37
intracube
what was the data recorded onto?
01:38
Bertl
one of the non-ssd medias used in the atomos, basically a small laptop hdd in an enclosure
01:39
intracube
yeah, I can imagine a regular HDD might have problems
01:40
Bertl
interesting, so far, by way the most contributions to AXIOM Beta come from the UK
01:40
intracube
especially if the throughput is near the limit of the drive
01:41
Bertl
while in the number of actual visits the USA is leading 3:1 against all other
01:42
Bertl
(according to the indiegogo dashboard)
01:44
intracube
interesting
01:44
intracube
considering the UK has 1/6th the population
01:45
troy_s
intracube: greets
01:46
Bertl
yeah, we have way less visits from the UK as well, so the contribution per visit ratio is like 7-10 times higher for the UK than for the US :)
01:46
intracube
I guess part of the USA are still at work. westcoast, at any rate
01:46
intracube
hey troy_s :)
01:46
intracube
see what happens in the early hours EU time
01:49
intracube
I'm seeing some slight gauss effect on the test footage if I push the colour correcting hard
01:51
Bertl
Gauss is everywhere :)
01:54
intracube
heh
01:55
troy_s
intracube: Resolve?
01:55
intracube
troy_s: hmm?
01:56
troy_s
What are you testing in?
01:56
intracube
oh, blender
01:57
troy_s
intracube: The DNxHD is loading?
01:57
intracube
ya
01:57
troy_s
intracube: With 601 matrices. :(
01:58
intracube
oh, well I'm eyeballing everything
02:01
troy_s
intracube: in the compositor?
02:01
intracube
yep
02:18
aombk
iots nice you added 2nd batch
02:19
aombk
but move it under 1st batch
02:19
aombk
nobody wants mft!
02:37
Bertl
unfortunately that's not possible with indiegogo
02:38
Bertl
all you can do is feature one, which will then become the first entry (as far as I know)
02:39
Bertl
but I'm off to bed now anyway, se6astian will be up again in a few hours and maybe figure out a way
02:39
Bertl
thanks to everyone so far!
02:40
Bertl
changed nick to: Bertl_zZ
03:44
alesage
left the channel
03:45
alesage
joined the channel
03:51
anton
left the channel
04:18
dmjweb
left the channel
04:25
aombk
left the channel
04:48
Morethink
joined the channel
05:12
ItsMeLenny
joined the channel
05:48
jucar
left the channel
06:07
jucar
joined the channel
07:19
alexML
try removing the colon from "AXIOM Beta: Super35 2nd Batch" - it should fix the sorting issue
07:25
macbookprocmc
joined the channel
07:30
derWalter
joined the channel
07:54
macbookprocmc
left the channel
08:03
se6astian|away
changed nick to: se6astian
08:20
se6astian
good morning
08:23
derWalter
morning guys, just reading the irc log from yesterday and i came up with an idea, i just want to share with you: i was thinking of what i rly miss right now on the axiom beta, internal recording and a screen. internal recording can be fixxed for quite cheap till the release i think and the screen problem reminded me on this: http://www.heise.de/newsticker/meldung/VR-wird-groesser-als-Doom-Spiele-Pionier-John-Carmack-im-Gespraech-2
08:25
derWalter
carmack is using regular cellphones for the VR, as they mostly have stunning displaying capabillities (OLED fullhd). i dont
08:31
derWalter
and with wlan AC, bluetooth 4.0 one can get up to 800 / 700 mbits, a obligatory onboard hardware x264 decoder should allow for a lot of possibilities to make that true
08:32
derWalter
so if i can record internally to ssd/2,5" hdd and use my phone as monitor/controller (touchscreen) than i am rdy to go!
08:35
se6astian
perfect, but we first need to focus on the camera core before we build everything around it :)
08:37
philippej
joined the channel
08:41
macbookprocmc
joined the channel
08:41
macbookprocmc
changed nick to: WAYOFF
08:43
philippej
left the channel
08:46
philippej
joined the channel
08:52
philippej
Hello everyone !
08:59
tyrone_
joined the channel
08:59
tyrone_
morning
09:09
Michal__
joined the channel
09:09
Michal__
left the channel
09:26
seku_
joined the channel
09:26
seku_
moarnings
09:26
se6astian
hello there
09:31
seku_
hi se6astian, good going with the funding and the cam. really looking forward to it
09:39
seku_
reading up older irc logs to get to know things better
09:44
ApertusWeb3
joined the channel
09:46
philippetest
joined the channel
09:47
philippej
nick selection feature works on the irc webpage, as long as one fills the nickname field :-)
09:50
philippetest
left the channel
10:09
se6astian
great
10:56
Morethink
left the channel
11:11
Bertl_zZ
changed nick to: Bertl
11:11
Bertl
morning folks!
11:17
se6astian
hello!
11:22
mars_
hi se6astian
11:50
Morethink
joined the channel
11:50
Morethink
left the channel
11:56
seku_
morning Bertl :)
12:03
Bertl
hey, how's going?
12:04
seku_
quite fine after a good night's sleep
12:05
seku_
been reading up a bit about HDR mode (and how it affects motion blur in highlights) ... all so intriguing
12:49
aombk
joined the channel
12:52
aombk
hi
12:55
ItsMeLenny
hello
12:56
ItsMeLenny
#apertus is attracting many the #magiclantern users
12:56
seku_
it sure is :)
12:57
philippej
... and there is a reason for that : developpers, on axiom, will have complete freedom and complete documentation :-)
12:58
ItsMeLenny
yeah
12:58
ItsMeLenny
i'm confused as to the multiple options for sensors, is that just because it was easy to integrate? do they have similar pinouts. at the same time im not sure what advantage that truesense sensor offers (except for it being cheaper)
12:58
philippej
it depends on what you want to do I guess
12:59
Bertl
well, after we finished the AXIOM Alpha, with the CMV12000, we decided that we would like to investigate a different sensor as well
13:00
ItsMeLenny
was there a blog post or similar on findings?
13:00
Bertl
and the price of the CMV12000 didn't seem very appealing for developers
13:00
ItsMeLenny
ah
13:00
philippej
They don't have similar pinouts, but the board supporting it will have the same pinout to attach it to the processing board. Think of it as a kind of adapter board, with curently two kind being planed, one for cmosis, one for truesense.
13:00
Bertl
so we decided to go with the much cheaper but somewhat compareable TrueSense chip
13:01
Bertl
but, we did an opinion poll, what sensor would be preferred for the upcoming Beta, and the results clearly featured the CMV12000
13:01
ItsMeLenny
that makes sense, i see the advantage there for developers
13:01
ItsMeLenny
yeah
13:01
Bertl
so as they are 'reasonably' similar, we decided to offer both versions
13:02
ItsMeLenny
ah
13:02
ItsMeLenny
does the cmv12000 have a similar pinout to the cmv8000, as that also is on the beta page
13:02
Bertl
i.e. both have SPI/I2C register sets, both have LVDS blocks, etc
13:02
Bertl
no, they all have different pinouts, but that's not a big problem per se
13:03
Bertl
only the CMV2k and CMV4k are pin compatible
13:03
ItsMeLenny
just so long as the same register sets and etc are onboard
13:03
ItsMeLenny
ah
13:03
Bertl
register sets are just software, i.e. you have to adjust them differently
13:03
ItsMeLenny
oh
13:04
Bertl
don't forget, with the AXIOM you have control over all parts
13:04
ItsMeLenny
i had that cmosis page open for a couple of months at one point looking mainly at the CMV2000
13:05
ItsMeLenny
im definitely interested in getting one, might have to wait for the gamma though, however i blew a bit of money on a DSLR (which i got for video, which is also many peoples story)
13:05
derWalter
left the channel
13:06
ItsMeLenny
an axiom that is ^ (not the cmv2000)
13:08
ItsMeLenny
thanks for the insight Bertl clears things up a bit, and philippej
13:09
Bertl
you're welcome!
13:10
philippej
you are welcome! Don't forget that gamma will be more expensive, and not available as soon as beta (and also don't forget the upgrade path from beta to gamma)
13:10
sashacohen
joined the channel
13:10
ItsMeLenny
should i wait for the 200 dollar point and shoot version? :P
13:11
philippej
I would not bet on that :-)
13:11
Bertl
more importantly, if the beta will not succeed, I'm not sure there will be a gamma :)
13:11
sashacohen
changed nick to: Sasha_C
13:11
ItsMeLenny
ah
13:12
ItsMeLenny
if i had the money i'd pay for the whole project
13:12
ItsMeLenny
or we'll say, when i have the money i'll pay for the whole project, but it might take a little longer than duke nukem forever
13:14
Bertl
no problem there, I was just pointing out that if everybody decides to 'wait' for the Gamma, it is unlikely that there will be one :)
13:14
ItsMeLenny
yeah, i know what you mean
13:19
aombk
you ditched the cmv8000 ?
13:20
philippe_
joined the channel
13:21
se6astian
changed nick to: se6astian|away
13:23
Sasha_C
Hi everyone :)
13:23
philippe_
Hi sasha, how goes?
13:25
WAYOFF_
joined the channel
13:25
Sasha_C
I'm well thanks. I feel bad that it's been a long time since I've come online here. Just been very busy with other things in my life. Nice work with the crowd-funding videos!
13:26
Bertl
you should really feel bad about australia not contributing to the crowdfunding :)
13:26
Bertl
welcome WAYOFF_!
13:29
philippe_
http://www.cnet.com/news/crowdfunding-helps-apertus-new-open-way-to-make-a-camera/
13:29
WAYOFF
left the channel
13:29
Sasha_C
@Bertl: I know, we're really backwards here at the bottom of the world...
13:31
se6astian|away
changed nick to: se6astian
13:34
fabio_
joined the channel
13:35
aombk
axiom delta sounds nice
13:35
ItsMeLenny
left the channel
13:35
ItsMeLenny
joined the channel
13:36
Sasha_C
@phillipe_: That's great, we're the No. 1 'must read' article on C/NET!
13:37
aombk
has anyone gathered the sites and blogs that the campaign is mentioned?
13:39
seku_
yeah, awesome öD
13:44
tyrone_
aombk: you can check on news.google.com when you enter "apertus".
13:45
se6astian
ah nice Stephen finished the cnet article!
13:50
se6astian
changed nick to: se6astian|away
13:53
aombk
so, how realistic is the spec of 15 stops with hdr that cmosis gives?
13:53
Bertl
probably as realistic as all the other sensor specifications given by all the other manufacturers
13:54
Bertl
welcome fabio_!
13:54
aombk
have you tested?
13:54
seku_
i guess its complicated; also depends which one of the 3 modes you actually use
13:56
philippe_
dynamic range per se is defined by what you judge acceptable on image quality (noise fonr instance)
13:56
philippe_
(for)
13:59
aombk
yes i know but when you shoot dr charts things are less complicated, you see how many stops you can distinguish from the noisefloor and up
14:00
aombk
for some reason i really like shooting dr charts
14:02
fabio_
Thank you Bert! I've been following axiom for a long time,
14:02
fabio_
and it's getting better and better. congrats guys
14:03
philippe_
left the channel
14:05
sebastian1
joined the channel
14:08
philippej
left the channel
14:10
aombk
are you planning on making updates on the campaign? maybe a more technical video?
14:12
sebastian1
yes
14:12
aombk
like this. this video was absolutely amazing. http://www.youtube.com/watch?v=mvI7rJ_AZys
14:13
alexML
aombk: you can find the HDR specs on the sensor datasheet, http://www.betopcom.com/public/images/products/20131126180222.pdf page 32
14:14
alexML
the first is just like dual iso, but with shutter speed
14:14
ItsMeLenny
when i was looking up the cmosis 4000 about half a year ago i came across this http://www.lumenera.com/usb3/index.php
14:15
alexML
the second seems somewhat like red's HDRx (so highlights will get less exposure time), but I bet there will be motion artifacts (sure, not as bad as ML HDR video)
14:16
alexML
and here https://www.apertus.org/axiom_imagesensor they have a third mode, which is just like ML HDR vido, maybe at higher FPS
14:20
aombk
yes alexML thanks. i was just wondering how many stops dr are actually achievable with the first hdr mode
14:22
alexML
with the one like dual iso? I guess about 15 or 16, maybe more if you accept some noise in midtones (but the aliasing can be noticeable)
14:24
aombk
no i know that from ml already. i mean that with the altered photosite response curve
14:24
alexML
and with the 3-slope method, it's basically the ZeroNoise algorithm applied on 3 images, in hardware
14:24
alexML
imagine it takes 3 images, sums a+b+c, if it's clipped it sums b+c, and if it's still clipped, it keeps the data only from c
14:25
alexML
you can configure the exposure times and the clipping thresholds
14:25
alexML
(datasheet page 34)
14:26
aombk
i am still downloading the pdf. its very slow.
14:26
aombk
and these 3 images are taken at the same time?
14:26
alexML
lol, no
14:26
alexML
you have a single sensor
14:27
aombk
so there may be some motion artifacts even with that mode?
14:27
alexML
well, my description was a bit inaccurate
14:27
alexML
first, it captures a
14:27
alexML
then it resets clipped pixels
14:28
alexML
(unclipped pixels will keep integrating)
14:28
alexML
then it captures b
14:28
cfelton
joined the channel
14:28
alexML
so, where a was not clipped, the accumulator now contains a+b
14:29
aombk
thanks. actually i will look into that pdf once it downloads
14:29
aombk
and ask if i dont get something
14:29
seku_
isnt that similar to mode 2? (shorter exposure time for highlights)
14:29
Jc_
joined the channel
14:30
Jc_
Hi there,
14:30
Jc_
is Philippe Jadin around ?
14:31
ItsMeLenny
he left about 25 minutes ago
14:32
sebastian1
hi
14:32
sebastian1
Jc_: can we help you instead maybe?
14:32
sebastian1
Latest article: http://www.heise.de/newsticker/meldung/Axiom-Beta-Crowdfunding-fuer-die-erste-quelloffene-Filmkamera-2389950.html
14:35
seku_
mh, i think the hdr modes are quite interesting tho ... just that motion blur needs to be added in post
14:35
aombk
is it me or are europeans more interested in this camera?
14:39
seku_
being a european project helps, i suppose
14:40
seku_
left the channel
14:40
seku_
joined the channel
14:46
alexML
the trick is quite powerful though, I wouldn't be surprised if one would squeeze 20 stops by tweaking the curve
14:51
seku_
fun times ahead
14:57
ItsMeLenny
left the channel
15:11
seku_
left the channel
15:11
seku_
joined the channel
15:11
seku_
left the channel
15:12
seku_
joined the channel
15:13
seku_
left the channel
15:16
seku_
joined the channel
15:36
WAYOFF_
left the channel
15:36
sebastian1
left the channel
16:03
seku_
left the channel
16:20
philippej
joined the channel
16:23
fabio_
left the channel
16:43
Sasha_C
left the channel
17:10
Topic
apertus° - open source cinema | www.apertus.org | Currently Hot: AXIOM Beta Crowd Funding is Live: http://igg.me/at/axiom-beta | IRC Logs available at: http://irc.apertus.org
17:10
se6astian
has set the topic
17:49
dmjweb
joined the channel
17:51
jucar
left the channel
17:53
jucar
joined the channel
17:54
se6astian|away
changed nick to: se6astian
17:56
philippej
joined the channel
18:00
dude
joined the channel
18:01
dude
changed nick to: Guest70687
18:05
danieel
left the channel
18:08
Guest70687
left the channel
18:12
willr
joined the channel
18:17
willr
left the channel
18:20
philippej
left the channel
18:55
dmjweb
se6astian: congrats on the speedy start to the campaign
19:03
Achyu
joined the channel
19:03
jucar
left the channel
19:03
Achyu
left the channel
19:04
achyu
joined the channel
19:05
jucar
joined the channel
19:12
achyu
left the channel
19:25
philippej
joined the channel
19:37
se6astian
dmjweb: thanks :)
19:37
se6astian
now we need to keep up the momentum
19:45
dmjweb
se6astian: perhaps an update is in order?
19:57
se6astian
yes, we are working on it as we speak
21:06
yeehi
joined the channel
21:07
aombk
this datasheet is very interesting even though i understand 1/10th of it
21:08
yeehi
The sensor at the moment is cmv12000. Is it somehow possible to back this crowdfunding without getting that sensor, and instead waiting for a cmv20000, if that is in the pipeline?
21:10
aombk
yeeji, maybe you can ask that on the campaign page too
21:11
aombk
i am very disapointed by http://www.eoshd.com/ he didnt even mention axiom beta and the campaign ta all
21:12
yeehi
aombk, why do you like the project?
21:13
yeehi
Is high fps sensor important feature to you?
21:13
yeehi
Which other components / features do you think will be important for the camera?
21:14
aombk
the sensor specs are very good and seems to have great potential
21:14
Bertl
yeehi: well, there is no guarantee that we ever build an AXIOM Beta with the cmv20000, but of course you can back the project and hope for a cmv20000 version lateron
21:15
yeehi
Hi, Bertl.
21:15
yeehi
What would I do with the cmv12000 sensor?
21:15
aombk
global shutter, hdr mode with 15 stops lots of frames for slowmotion raw 12bit etc
21:15
yeehi
Is there a secondary market?
21:15
aombk
but the main reason is openess
21:15
yeehi
Ah, there is hdr mode - that is good.
21:16
yeehi
How good is this sensor in low light, relative to canon sensors?
21:22
dmjweb
yeehi: There's no guarantee anyone will build a given thing you want (unless you pay them enough to do it), but one of the nice things about the Axiom is that a third party could build a CMV20000 module for it.
21:22
dmjweb
That all depends on demand of course.
21:25
alexML
yeehi: an APS-C Canon sensor (7D) has 12 electrons of read noise at ISO 100, and 3 electrons at ISO 1600 (source: http://www.clarkvision.com/photoinfo/evaluation-canon-7d/index.html )
21:25
alexML
CMV12000 has 13 electrons according to the spec
21:25
alexML
so it's not exactly good in low light
21:25
yeehi
thank you, dmjweb
21:26
alexML
(it's roughly as noisy as ISO 100 on APS-C Canons in low light, pushed in post)
21:27
yeehi
thank you, alexML. 1 electron of noise different? That isn't much, is it? Wouldn't that be almost the same as Canon's top sensor?
21:29
yeehi
alexML, I am not so technical, but if the cmv12000 is achieving a level of noise comparable to iso 100 on canon's top sensor, that is brilliant news.
21:30
alexML
Canon at ISO 100 is very noisy, compared to Nikon/Sony
21:31
yeehi
i didn't know that, alexML. It is good to talk to somebody who is familiar with this sort of stuff.
21:31
alexML
if you already have a Canon, try shooting video at ISO 100 in low light
21:32
alexML
and push the shadows in post
21:32
yeehi
alexML, i would like to shoot in available light. I usually need a fast, wide open lens and high iso, say 400 or even 800. Would this cmv12000 (or even cmv20000) be good in these settings?
21:33
yeehi
Also, I am surprised at the relatively small size of the image, around 12 Mp. I thought it would be around 20 Mp.
21:33
yeehi
I think that 12Mp wouldn't be so good for cropping digitally.
21:33
alexML
compared to ISO 800 on 7D, the cmv12000 is 1.77 stops worse (per pixel)
21:34
alexML
(regarding read noise in electrons)
21:35
alexML
the cmv20000 is full frame, and has 8 electrons of read noise
21:36
alexML
the 6D (best Canon sensor yet) has 6 electrons at iso 100 and 1.7 at iso 6400 - http://www.clarkvision.com/photoinfo/evaluation-canon-6d/
21:36
yeehi
Yes, I noticed the cmv12000 is a crop (1.3 or 1.6, i can't remember)
21:37
alexML
so, in low light, cmv20000 would be log2(8/1.7) = 2.23 stops worse than a 6D (assuming you are shooting at iso 6400)
21:38
alexML
I don't know if the noise on the CMV sensors gets better with analog amplification (if it would, the spec would say it, right?)
21:39
alexML
btw, I didn't analyze any images from the sensor yet - I'm just interpreting the spec, but I might be wrong
21:40
alexML
I will ask se6astian to take a few test pictures from which I can do some SNR plots, estimate the DR and compare with other cameras
21:41
yeehi
Is there a forum for this project, alexML?
21:41
alexML
yeah, they have one here: https://www.apertus.org/forums/
21:42
yeehi
thank you for all your help, alexML
21:42
tyrone_
alexML: ask se6astian is has plenty of recorded shots... :-)
21:43
alexML
I need two DNGs of some static scene, covering from deep blacks to clipped highlights
21:44
alexML
I can subtract those to see the noise, plot the SNR, and get the DR from there
21:44
yeehi
alexML, how is the dynamic range relative to a canon sensor? Especially rescuing low light
21:45
alexML
ideally, I need a pair of DNGs at each setting (analog gain, hdr mode, whatever you can control at this stage)
21:47
alexML
DR in basic mode (without HDR tricks) seems comparable to Canons at ISO 100 (roughly, from spec)
21:47
alexML
but with HDR tricks it can get quite high
21:47
alexML
the spec says 15, but I think it can get more
21:48
yeehi
alexML, â¬1350 is quite a lot of money. The investment would start paying off when it wasn't necessary to pay a fortune to upgrade the entire camera for a new sensor.
21:48
alexML
but the increase in DR is on the lower side of the ISO
21:48
alexML
(it will capture more highlights, but it won't improve the shadows - so you need plenty of light)
21:49
yeehi
I am very happy to talk with you, alexML.
21:50
yeehi
You must like sensors a lot!
21:50
philippej
left the channel
21:50
yeehi
it talks about 4k RAW video via experimental... what would this be?
21:51
yeehi
Aren't modern day production cameras capable of shooting 4k video RAW, using non experimental?
21:55
yeehi
Is the camera going to have a LED to preview/view pictures?
21:56
alexML
correction from above: 6D has 29 electrons at ISO 100 (I copied the wrong number)
22:03
dmjweb
yeehi: those 4K cameras have an ssd recorder built in or as a module
22:03
dmjweb
there's nothing preventing this for the axiom, it just hasn't been made yet
22:04
dmjweb
and it's not expected to be done by the time the beta is ready
22:08
anton_
joined the channel
22:08
anton__
joined the channel
22:08
anton_
left the channel
22:11
anton_
joined the channel
22:13
anton__
left the channel
22:14
aombk_
joined the channel
22:16
aombk
left the channel
22:16
aombk_
changed nick to: aombk
22:33
se6astian
good night
22:33
se6astian
early flight tomorrow :)
22:40
dmjweb
left the channel
22:52
se6astian
changed nick to: se6astian|away
22:53
aombk
btw a solar flare is coming
23:06
anton_
Guys, re future plans for SSD. Do you plan to use smth like Plextor M6e/Samsung XP941 ? And then use Zynq 7015's implementation of PCI-e to communicate with these?
23:08
anton_
Looks like Zynq 7015 can only do PCI-e 2.0 where as one would need 3.0 to fully utilise XP941
23:08
anton_
Is there a softcore?
23:16
anton_
My thinking was - if there was a proof of concept that XP941 can be communicated with from a Microzed then perhaps anticipated Axiom Beta specs could be improved with an internal M.2 SSD interface - and bang! people can start looking forward to internal recording. Could that liven up the campaign?
23:18
anton_
...and if that was 3.0 PCI-E perhaps they could even start looking forward to internal RAW recording at high FPS. You know some glitter. Smth other cameras don't have any hope of doing any time soon
23:21
anton_
Not even necessarily speaking of having it 100% working at the time of initial hardware shipment. Rather about having internal space for one (or several?) M.2 SSD-s and the internal connectors for them. If the hardware is there software can be added a little later
23:26
anton_
Alternatively to PCI-E would it make sense to add a USB3 port? Or several? Then external SSD-s could be connected also providing recording w/o a specialized recorder - and also in RAW. Perhaps writing could be done to 2 external SSD-s simultaneously to record RAW at high res and fps?
23:32
anton_
perhaps both internal M.2 for SSD and a couple of external USB3? does 7015 have enough juice for that? :)
23:35
anton_
is it worth to sacrifise 1 of the 3 hdmi outputs for this if needed? after all it will be much easier for people to buy another external encloser with SSD inside and USB3 interface than to get a video recorder which takes video over 2 hdmi inputs simultaneously. These SSD-s are already close to 1$ per Gb
23:42
anton_
Still trying to answer question of that forum member on divxuser who asked "how do you get 300 fps raw out of the camera" - and my answer is - via USB3 or internal SSD. And then the question is if hardware for these needs to be in the very first instances of Axiom Beta already and in the current ads
00:06
Bertl
let me try to answer some of those questions :)
00:07
Bertl
the 7015 (which we do not plan right now, for several reasons, but which might be a potential choice for a Beta version based on the PicoZed
00:08
Bertl
has 4 GT tranceiver with up to 6.6Gigabit per tranceiver
00:08
Bertl
for PCIe 2.0 5.0Gigabit per lane is required
00:10
Bertl
the CMV12k has a resolution of 4096x3072 pixel, at a color depth of 10 bit, it can achieve 150 FPS at this resolution
00:11
Bertl
which is about 19Gigabit per second uncompressed data
00:11
Bertl
so a PCIe v2.0 x4 interface will be able to transport all the data available
00:13
Bertl
USB 3.0, which is a protocol nightmare, allows for 5Gbit/s transfer speeds
00:13
Bertl
so one would require 4 USB 3.0 interfaces (which would also be doable with the 7015) to transport/record the amount of data generated
00:14
Bertl
USB 3.1, which is already defined, could do 10GBit/s per connection, so two connections would suffice, but as far as I know there are no real world devices which can handle that speed
00:15
Bertl
also, the 7015 would not be able to handle those speeds without external USB 3.1 transceiver
00:16
Bertl
now, for the potential solutions on the AXIOM Beta as planned (i.e. without the 7015, as there is no MicroZed with 7015 available)
00:17
Bertl
we planned for 2 high speed connectors, each featuring 10 differential pairs and each of those pairs can easily transfer 800MBit/s without any tricks, and probably twice the amount with careful handling of the data
00:17
Bertl
i.e. double data rate, synchronous transfers and such
00:17
Bertl
(of course that is untested, as we do not have a Beta yet)
00:18
Bertl
this means, that a properly designed shield could easily achieve the same data throughput as the 7015 via those gigabit tranceivers
00:20
Bertl
now, looking at existing SSD solutions, anything which can even remotely handle this amount of data looks like a larger graphics card
00:20
Bertl
so it is significantly larger than the entire Beta
00:21
Bertl
realistic options are to use some kind of RAID to store the data to smaller SSDs, which still would require about 4-6 SSDs to store this amount of uncompressed raw data
00:24
anton_
left the channel
00:24
Bertl
so building a storage interface for the Beta is quite feasable, but it is a rather complicated project on its own, which probably requires a lot of new code to be written (PCIe, SATA, etc) to make it work.
00:26
alesage
left the channel
00:26
Bertl
As we do not want to develop the Beta for 10 years and make it a final, finished product, we decided to focus on the essentials, while giving everyone enough room to implement whatever they want to do