Current Server Time: 01:13 (Central Europe)

#apertus IRC Channel Logs

2015/01/07

Timezone: UTC


00:35
intracube
left the channel
00:42
ItsMeLenny
joined the channel
00:51
fsteinel_
joined the channel
00:54
fsteinel
left the channel
01:05
lab-bot
BAndiT1983 created T249: Add histograms for image analysis. http://lab.apertus.org/T249
02:27
ItsMeLenny
left the channel
03:57
Bertl
off to bed now ... have a good one everyone!
03:57
Bertl
changed nick to: Bertl_zZ
06:37
davidak
joined the channel
07:08
davidak
left the channel
10:13
HAL-1999a
joined the channel
10:34
HAL-1999a
Guys a quick question:
10:34
HAL-1999a
What are the components which will be common to both Beta and Gamma?
10:40
daFred
joined the channel
11:23
lab-bot
sebastian created T250: Research Packaging foam molding solutions. http://lab.apertus.org/T250
11:24
daFred
left the channel
12:14
Bertl_zZ
hmm
12:15
Bertl_zZ
changed nick to: Bertl_oO
12:17
Bertl_oO
changed nick to: Bertl
12:17
Bertl
morning folks!
12:17
HAL-1999a
morning!
12:17
Bertl
HAL-1999a: most likely the sensor :)
12:18
HAL-1999a
No PCBs? No body parts?
12:24
Bertl
we don't know yet, we'll see
12:30
daFred
joined the channel
12:55
HAL-1999a
Ok...Too early for that I guess.
13:10
designbybeck
joined the channel
13:56
MariusRumpf
joined the channel
14:32
HAL-1999a
left the channel
14:39
MariusRumpf
left the channel
15:42
daFred
left the channel
15:55
intracube
joined the channel
15:57
intracube
does Phabricator still end verification emails with "Get well soon"?!
15:57
intracube
just noticed on his email from a few months back :)
16:00
Bertl
maybe it does, no idea :)
16:14
Topic
apertus° - open source cinema | www.apertus.org | join the apertus° Lab: http://lab.apertus.org/ | IRC Logs available at: http://irc.apertus.org
16:14
se6astian
has set the topic
16:32
HAL-1999
joined the channel
16:34
aombk
joined the channel
16:49
davidak
joined the channel
16:53
daFred
joined the channel
17:08
troy_s
Alright... where's the damn Gamma?
17:35
Bertl
in our heads :)
17:40
Bertl
so, the missing connectors finally arrived (more than a week later than expected), which means that I can now finally build the first Beta Test Boards \o/
17:44
troy_s
Bertl: Serious question, how many revisions do you estimate? I'm guessing three or four?
17:44
Bertl
yes, but hopefully not on all boards
17:45
Bertl
the sensor board (cmv12k) seems good to me, for example, so I expect only one more revision
18:09
Bertl
(unless we find a major problem with that :)
19:01
Blackpoint
joined the channel
19:01
Blackpoint
left the channel
19:11
aombk2
joined the channel
19:12
jucar
left the channel
19:15
aombk
left the channel
19:15
jucar
joined the channel
19:22
jucar
left the channel
19:25
jucar
joined the channel
20:04
aombk2
left the channel
20:20
troy_s
slikdigit: You probably didn't want to read that tweet. LOL
20:21
slikdigit
not really :) this is an early wip, there's not a final render in the whole thing
20:22
slikdigit
main purpose of this render is for me and bishop to evaluate the character animation throughout and see if there is anything that should be fixed
20:22
slikdigit
s/if there is anything that/what
20:24
philippe_
joined the channel
20:27
troy_s
Yep.
20:27
troy_s
You are close though!
20:27
troy_s
Congratulations!
20:27
troy_s
slikdigit: Can't wait to see it.
20:28
troy_s
slikdigit: PS... the only way I can see to conform is to render out a non-dissolved / crossed / overlayed version. That is, as you already have, reduce the thing down to the fewest reels you can. (oldschool basically)
20:28
troy_s
So assuming shot A, B, and C, where B overlays part of A and C, you'd need two reels. One for A and C, and the separate for B.
20:29
troy_s
Render our the reels in full, with black slugs in between where the shots go (so in our example, hide the top reel, render A's reel which is A <Black slug space> C, then hide it and unhide B and render the same <black slug> B <black slug>
20:29
slikdigit
hmmm
20:29
troy_s
Once you have all your reels, you can do an _actual_ per shot grade at 32 bit float in the compositor.
20:30
troy_s
AND you get correct film transitions.
20:30
troy_s
;)
20:30
slikdigit
sounds complicated :D
20:30
slikdigit
copy and pastes log
20:30
troy_s
It's not that complex. It actually mirrors the exact same thing I used to do with film.
20:30
troy_s
Your workprint had a fade say, which was merely a grease marked marker on the single workprint with a squiggly line.
20:31
troy_s
When you had to craft a negative blueprint for your negative cutter, you used to have to do a paper map. We'd call that an interchange or EDL these days.
20:31
troy_s
But that map showed the negative cutter where you wanted your dissolves / fades / overs / blended bits to go.
20:31
troy_s
They'd deliver to you generally two reels exactly as I just described, with slug in them.
20:32
troy_s
Sometimes, on extremely complex cuts you'd need a third reel.
20:32
troy_s
Then the lab would print those precisely as you choose, with per-shot tri-light grades.
20:32
troy_s
So again, it's not terribly complex. If you have dissolves, you end up with two reels likely. If you have a triple dissolve, then you have a third.
20:33
troy_s
(or say a double cross where you have A dissolve to B, and before A is finished, C dissolves across the other two.
20:33
troy_s
The point is, you want a perfectly clean shot, end to end, with no opacity etc.
20:33
troy_s
Then you take those two reels into the compositor and grade them for a week (I suspect on your project of that length)
20:33
troy_s
Manual conform in that you have to mix manually between the reels for the dissolves, but... and this is a large but...
20:34
troy_s
All of the processing is 32 bit float.
20:34
troy_s
All of the mixing is display linear.
20:34
troy_s
And all of the grading options are much more massive, including tracking regions or masking etc.
20:34
troy_s
It just 'makes sense' in truth.
20:34
troy_s
Per shot grades _and_ proper blending.
20:34
troy_s
:)
20:34
slikdigit
actually there are going to be very few dissolves for tube
20:35
slikdigit
some of the ones in the current edit are a crutch standing in for lighting changes
20:35
troy_s
Still applies on the grading front.
20:35
slikdigit
yup
20:35
troy_s
And easier for trim passes etc.
20:36
troy_s
The upside is that if you tried it, you might be able to shake a tree and get a proper conform button.
20:36
troy_s
Which would _rock_
20:36
troy_s
(as well as likely explain to many folks why it would be excellent.)
20:36
troy_s
I remember explaining to sebastian ages ago why codec control is key, and sure enough, many ages later, I think he has come around on it.
20:44
slikdigit
codec control?
20:44
slikdigit
is this sebastian koenig?
20:46
troy_s
Yes.
20:46
troy_s
Had a chat earlier... he was having issues with codecs going to broadcast.
20:47
troy_s
Tricky things. You can't exactly take FooBar application, encode it, and then decode it in the same FooBar application to evaluate... because chances are the mistakes are created and then undone and the result looks perfect.
20:47
troy_s
(Remember the PNG transparency issue in Blender? Same thing.)
20:49
philippe__
joined the channel
20:49
philippe_
left the channel
20:51
slikdigit
yup
20:51
troy_s
Do you have any broadcast screenings locked in yet?
20:58
philippe__
left the channel
21:06
slikdigit
troy_s: not yet :)
21:52
davidak
left the channel
22:25
daFred
left the channel