01:13 | Bertl | off to bed now ... have a good one everyone!
| |
01:38 | gcolburn | joined the channel | |
03:38 | gcolburn | left the channel | |
03:44 | aombk | left the channel | |
09:40 | se6astian | joined the channel | |
09:41 | se6astian | good morning
| |
09:58 | Bertl | morning folks!
| |
10:24 | danieel | i would wonder what is the preferred medium for 4k raws :) playing with it internally in a pc is okay, but to actually put the data onto something else is insanely slow
| |
10:28 | se6astian | what are you trying to do exactly?
| |
10:29 | danieel | e.g. take the footage to postproduction site
| |
10:29 | danieel | if it takes 12h to put it on a usb drive, it will take another 12 on that site to offload it
| |
10:29 | se6astian | so why not carry it to them?
| |
10:29 | rexbron | left the channel | |
10:30 | danieel | wasted 24 hours, by using "consumer media"
| |
10:30 | danieel | anything better out there?
| |
10:30 | Bertl | high speed striped SSDs :)
| |
10:34 | danieel | then there is a questions if taking a whole NAS would not be simpler
| |
10:34 | danieel | e.g. on 10GE
| |
10:35 | danieel | i saw recently a mobo with onboard 10GE, for lowpower xeons
| |
10:35 | Bertl | probably depends, if the SSDs are 1.8", 4 or 5 of them might still be easier to transport
| |
10:36 | danieel | but you have to agree on raid type / strip files in round robin fashion.. and also on filesystem
| |
10:37 | Bertl | similar for the NAS
| |
10:37 | danieel | can share both nfs and samba...
| |
10:37 | danieel | or iscsi
| |
10:37 | danieel | better than trying to put ext4 drive into windows :)
| |
10:37 | Bertl | still you have to agree on one :)
| |
10:37 | danieel | at nas, it can coexist, at drive the FS is one time choice
| |
10:38 | Bertl | depends on the filesystems
| |
10:45 | rexbron | joined the channel | |
10:56 | danieel | and one more thing worries me - sending media abroad
| |
10:56 | danieel | how to avoid taxation on both ends upon receipt of the item
| |
10:57 | danieel | as the primary reason is to transfer data, not the actual hardware
| |
12:01 | se6astian | left the channel | |
12:22 | rexbron | left the channel | |
12:22 | rexbron | joined the channel | |
12:22 | rexbron | left the channel | |
12:22 | rexbron | joined the channel | |
12:22 | rexbron_ | joined the channel | |
12:25 | rexbron | left the channel | |
13:11 | se6astian | joined the channel | |
13:24 | rexbron_ | left the channel | |
13:24 | rexbron | joined the channel | |
13:24 | rexbron | left the channel | |
13:24 | rexbron | joined the channel | |
15:22 | intracube | joined the channel | |
16:33 | troy_s | danieel: Disks.
| |
16:34 | troy_s | danieel: Basically SSDs as Bertl said. Mostly proprietary.
| |
16:52 | danieel | proprietary like those from field recorders?
| |
16:59 | troy_s | danieel: Similar. Google Alexa Codex (now onboard) or look through the Sony site for the F65 / F55 disks.
| |
17:01 | intracube | left the channel | |
17:03 | danieel | the sony system does compress the content?
| |
17:04 | danieel | 2.4gbps write speed for AXS-512 does not handle 4k 16bit
| |
17:20 | troy_s | danieel: The Sony system handles native 4k fine (native 6k Bayer)
| |
17:23 | danieel | i see, they use some lossy compression
| |
17:24 | danieel | "Choice of F65RAW-SQ mode with mild 3.6:1 compression or F65RAW-Lite mode with 6:1 compression "
| |
17:25 | danieel | so same as red, wavelet based compression
| |
17:51 | troy_s | danieel: Never shoot the compressed modes.
| |
17:52 | troy_s | danieel: Only time it comes up is for high frame rate.
| |
17:52 | troy_s | danieel: And even then the vfx folks get grumpy.
| |
17:56 | danieel | got some raw frame from f65? the few googlable ones are from blender and are jpegs
| |
17:58 | troy_s | danieel: Won't do you any good. The raw format is a blob.
| |
17:59 | danieel | i meant some tiff export, with 2.2 gamma (no color matrix needed) - just to judge the noise/blur
| |
18:00 | troy_s | You never find 2.2
| |
18:00 | troy_s | Only logs as
| |
18:00 | troy_s | I think the options are linear exrs or Cineons
| |
18:00 | danieel | ah, i thought linear16bit is default
| |
18:00 | troy_s | danieel: You can find Blender exrs
| |
18:01 | troy_s | danieel: But you can't judge noise without the color transform
| |
18:01 | troy_s | (and even then shooter ability has an impact)
| |
18:06 | danieel | downloading the mxf from blender... we'll see
| |
18:08 | danieel | there was some time back a discussion on file per frame, or file per shot - (e.g. dng is frame=file) ... the benefit of single blob is that it is easier to copy continuously
| |
18:09 | danieel | e.g. total commander in windows will copy my dngs in random order if i copy a folder
| |
18:09 | danieel | that makes the data on the destination disc scattered
| |
18:10 | Bertl | maybe 'total commander' isn't the best tool then :)
| |
18:11 | danieel | i wanted to point out that such things happen - i would not expect it
| |
18:11 | danieel | anyway it is random in mc too :)
| |
18:14 | Bertl | maybe 'mc' isn't the best tool either :)
| |
18:53 | danieel | the half-float type in exr is quite ugly... on strong gradients it shows posterization
| |
18:57 | danieel | and the sensor is still quite noisy on the green screen - but hard to tell if that is due to noise amplification by the half-float
| |
18:58 | Bertl | well, what do you expect from 10bit
| |
18:59 | danieel | if the sensor can deliver 16bit linear, i would wonder why they used the 10bit float afterwards
| |
19:03 | Bertl | which sensor?
| |
19:03 | danieel | f65
| |
19:04 | Bertl | how do you know the sensor can do 16bit linear ?
| |
19:04 | danieel | sony states 16bit linear storage...
| |
19:05 | Bertl | storage, yes, but sensor?
| |
19:31 | troy_s | danieel: Uh what?
| |
19:31 | troy_s | danieel: Gradients are fine in half.
| |
19:31 | troy_s | danieel: Remember too, that you have to account for knowledge gaps.
| |
19:32 | troy_s | danieel: And the "noise" can only be evaluated after a color transform.
| |
19:32 | troy_s | danieel: Remember that judging noise is purely a measure of distance between values, and a color transform impacts that quite immensely.
| |
19:33 | troy_s | (And the levels the greens were exposed to etc.)
| |
19:33 | troy_s | IIRC the EXRs may not be raw dumps from the software. I cannot remember entirely.
| |
19:33 | troy_s | But if they are raw dumps from the software, then you _must_ use the ACES RRTs and such for viewing.
| |
19:34 | troy_s | If they are not, all bets are off and the data isn't terribly useful to evaluate, as I remember the primaries were used but the RRTs have some perceptual adjustments that smooth out nuances of the ACES color space.
| |
19:34 | troy_s | TL;DR Don't read too much into the EXRs as it was a learning experience.
| |
19:35 | troy_s | (And I have no clue how you are viewing the EXRs, and that can be at least as important)
| |
21:37 | se6astian | time for bed
| |
21:37 | se6astian | good night
| |
21:37 | se6astian | left the channel |