22:50 | danieel | america :)
| |
22:59 | intracube | $400 for a FLAC audio player?
| |
22:59 | intracube | mmm
| |
23:12 | aombk | no its not a flac player
| |
23:13 | aombk | it plays 192 kHz/24 bit files
| |
23:13 | aombk | that you but from a special online store
| |
23:15 | aombk | probably converted from lower khz
| |
23:19 | seku | well, if the DAC and headamp were world-class ... it'd interest me more than higher-than CD bitrates. even then, carrying damn expensive IEMs or headphoens around is scary
| |
23:20 | seku | and you dont want too good sounding headphones. at least not at work. stopped bringing my shure IEMs because they distracted me too much with good-sounding music.
| |
23:20 | intracube | aombk: buried on that kickstarter page "The PonoMusic.com Store uses FLAC (Free Lossless Audio Codec) audio format as its standard, for compatibility"
| |
23:21 | seku | well, FLAC is the most sensible choice
| |
23:21 | Q_ | 192 kHz / 24 bit does not make sense for playback.
| |
23:21 | danieel | 96/24 would be enough...
| |
23:21 | seku | audiophily is rarely making sense.
| |
23:22 | intracube | 48kHz would be enough for most
| |
23:22 | Q_ | 40 kHz / 16 bit is enough.
| |
23:22 | aombk | intracube, that makes sense. i meant it is not just flac. it supports 192/24 too
| |
23:22 | seku | flac can be what you want. got some 5.1 192khz/24bit FLAC tracks here
| |
23:22 | seku | and other strange stuiff
| |
23:23 | danieel | well i can hear the difference between that 44/16 and 96/24 :) so i prefer the better resolution
| |
23:23 | Q_ | No you can't. What you probably hear is a problem in your equipment.
| |
23:24 | aombk | https://www.indiegogo.com/projects/the-core-wireless-speaker-system
| |
23:24 | danieel | gear for 192k is not so common (its more like Alexas)
| |
23:25 | seku | danieel, i feel i can hear it sometimes, but i wont ever do a blind test on that :D
| |
23:25 | danieel | i have a demanding content... my wife does classical/opera :)
| |
23:25 | seku | mostly, 24/96 are mastered for a more discerning public. so they do sound nicer.
| |
23:25 | seku | but might sound just as well on 44.1 ... if mastered well
| |
23:26 | Q_ | I won't be claiming that for recording it doesn't mater, but it doesn't for playback.
| |
23:26 | seku | about speakers, ill stick with my linkwitz orions :
| |
23:26 | aombk | apertus should have had this approach. having people trying to communicate what a great experience they had with alpha
| |
23:26 | seku | http://www.linkwitzlab.com/orion_us.htm
| |
23:27 | danieel | playing back a compressed dynamic range is very noticable, with such high dynamic range as opera... yes one can do that to fit into a CD, but it is far from reality
| |
23:27 | aombk | "axiom alpha felt like nothing before!"
| |
23:27 | seku | mh, i dunno, the dynamic part of the cd sounds fine... mostly the speaker limits... imho
| |
23:27 | danieel | so with 96k/24 i am perfectly fine. It is as reality. but if you do not have any match how the content should sound like, you can be fine with a mp3 :)
| |
23:28 | Q_ | The DR of 16 bit is higher than your ears.
| |
23:28 | seku | good ole nyquist
| |
23:29 | Q_ | And you need the music very loud to be able to hear much above 16 kHz.
| |
23:29 | seku | the frequency range barely is, at 22khz... but i dont hear up that high anyway. and nothing interesting there either
| |
23:30 | danieel | seku: nice orions... i have a pair of vxt8 by krk
| |
23:30 | seku | aah monitors, nice \ufffdD
| |
23:30 | seku | :D
| |
23:30 | seku | got some B%W as my computer speakers.... love them too :)
| |
23:31 | seku | one day, when i save up enough money, ill go for nautilus probably. them snails. love them snails.
| |
23:31 | troy_s | Q_: Agree. The old argument is just misplaced.
| |
23:31 | troy_s | Q_: the very same ridiculousness abounds in imaging. The bit depth is required for manipulation, not output.
| |
23:31 | Q_ | troy_s: Exactly.
| |
23:32 | seku | you can upgrade a crazy amount in speakers, amps, DACs, i still use CDs for most of my stuff
| |
23:32 | troy_s | Q_: Same sort of statements. Almost identical. Same misinformation.
| |
23:32 | seku | well, FLACed cds via EAC
| |
23:32 | troy_s | I had a fellow try to insist that there was a difference in the signal (digital) going across silver.
| |
23:33 | troy_s | And that was where I realized the monstrosity I was talking with.
| |
23:33 | Q_ | Not to mention that most recent music doesn't even have a decent dynamic range, it's all been compressed.
| |
23:33 | aombk | audiophile magazines still do recordable cd reviews
| |
23:33 | troy_s | Q_: Why stop there? Aesthetics are emergent.
| |
23:33 | seku | hehe, silver crypgenic cable?
| |
23:34 | seku | i also had an argument with a guy about speaker cables and interconnects ( my point : if cables sound differently, then the cable is badly designed)
| |
23:34 | aombk | cd audio quality reviews
| |
23:34 | seku | was a bit ugly
| |
23:34 | troy_s | There is a professor that has done unscientific questions on MP3 vs lossless. Over the years (not surprising) his students have come to prefer MP3
| |
23:34 | Q_ | troy_s: The only good argument I've ever heard for quality of digital cables is the bit error rate.
| |
23:35 | troy_s | Q_: Point is: You hear it rather obviously.
| |
23:35 | troy_s | It isn't like a subtle timbre. LOL
| |
23:35 | seku | well, there is a concern for Jitter in digital cables. but that can be adressed nicely by asynch buffers
| |
23:35 | aombk | blind a/b testing is always the answer
| |
23:36 | seku | i admit tho that DAC upgrades can make a difference.
| |
23:36 | troy_s | It isn't magic though. You can hear digital oopsies as they are glaring. Not some unnoticeable qualia.
| |
23:36 | troy_s | DACs of course!
| |
23:36 | troy_s | Hence the D to A part.
| |
23:36 | troy_s | But even then in the contemporary market it is hard to argue against massive vendors like Burr B or Wolfson...
| |
23:37 | seku | i'm with Sabre :)
| |
23:37 | Q_ | seku: Yes, once you get from digital to analog, all the analog parts get important again.
| |
23:37 | seku | ESS32
| |
23:37 | troy_s | It is equivalent to someone trying to manufacture thresholds that Apple does in a mom and pop shop. Virtually impossible.
| |
23:37 | seku | Q_, thats why my DAC also does preamp duties, and then feeds active speakers
| |
23:37 | troy_s | Their tolerances are a byproduct of scale.
| |
23:38 | seku | (anedio DAC2) ... cute little thing
| |
23:38 | troy_s | Anyways... I am ghetto in audio.
| |
23:38 | seku | doesnt sound that way, troy_s :)
| |
23:38 | troy_s | But I totally comprehend the foolishness over bit depth.
| |
23:39 | troy_s | seku: I am. I have some older school decade old stuff. Holds up fine. Newer consumer receiver.
| |
23:39 | seku | thats all that matters :)
| |
23:39 | troy_s | I have toyed with bumping up to Paradigms but alas... Hard to justify.
| |
23:39 | comradekingu | There is a difference in silver, it will have slightly better rise and fall times for the same quadrant because it concucts slightly better
| |
23:40 | danieel | at audible frequencies?
| |
23:40 | troy_s | In Analog sure.
| |
23:40 | comradekingu | I dont see the point to be argued though
| |
23:40 | troy_s | Digital? I call bullshit.
| |
23:40 | troy_s | Digital transmits just fine down a fricking coat hanger.
| |
23:40 | seku | common audiophily "wisdom" states that silver sounds faster, and more brittle.
| |
23:40 | comradekingu | You can measure it with an oscilloscope if you want
| |
23:40 | comradekingu | Im not arguing it doesnt
| |
23:40 | seku | me? im using well-shielded copper
| |
23:40 | troy_s | Hell, I could scream beeps and boops out at you.
| |
23:41 | troy_s | LOL.
| |
23:41 | troy_s | Granted it would be a slow transmission...
| |
23:41 | seku | wait, let me get you a link
| |
23:42 | troy_s | I am sure the various metals may have measurable differences in analog though. No clue nor interest how buried that ends up though.
| |
23:42 | Q_ | Doesn't using silver and gold actually cause more noise because of their difference in electrode potentional then the copper?
| |
23:42 | seku | this is one of those crazyheads : http://www.hoererlebnis.de/archive/pdf/059-mk-phonosophie.pdf
| |
23:42 | comradekingu | They do, and gold is worse than copper
| |
23:43 | seku | selling interconnects and power cables as upgrades
| |
23:43 | seku | for thousands
| |
23:43 | comradekingu | So if you ever see someone with expencive cables tell them they will do better by soldering the cables in without gold terminals
| |
23:43 | troy_s | Interesting.
| |
23:43 | Q_ | I don't think I ever saw XLR cables with gold contacts.
| |
23:44 | troy_s | But obviously it is sub Nyquist or whatever in digital land?
| |
23:44 | seku | just use good Neutriks and youre fine
| |
23:44 | seku | (case in point : mine are soldered)
| |
23:44 | troy_s | Q_: How come? Does the difference generate voltage or?
| |
23:44 | Q_ | troy_s: Yes.
| |
23:44 | Q_ | troy_s: Battery principle.
| |
23:45 | troy_s | Genius.
| |
23:45 | Bertl | actually thermocouple
| |
23:45 | troy_s | Oh oh... Nerd alerts.
| |
23:45 | troy_s | That is quite fascinating.
| |
23:45 | comradekingu | the cable is an inductor, with resistive properties going along it, then a capacitive effect in between the wires
| |
23:46 | troy_s | Am I to guess that consistent mediums are important then?
| |
23:46 | comradekingu | Q_: all the neutrik XLRs have gold contacts afaik
| |
23:47 | troy_s | comradekingu: So what is the net sum of difference between capacitive / inductive is a differing voltage and noise?
| |
23:47 | Q_ | comradekingu: Looking at their website, none of their XLR connectors have it.
| |
23:48 | comradekingu | troy_s: if your cable isnt broken/underdimensioned then nothing
| |
23:48 | Q_ | Oh, some do.
| |
23:48 | troy_s | comradekingu: But if you have a context where there is a differential in the voltages?
| |
23:48 | seku | well, if one day i win the lottery, i will get me these : http://www.htguide.com/forum/attachment.php?attachmentid=12520&stc=1&d=1211369490
| |
23:48 | comradekingu | External noise can make a difference, where balanced cables help
| |
23:48 | aombk | acoustic room treatment is much more important on the sound you hear than those technical matters
| |
23:49 | seku | these will make CDs sound really nice... if nicely recorded
| |
23:49 | aombk | and it is almost never being done properly
| |
23:49 | comradekingu | troy_s: there is always a difference, but is academical
| |
23:49 | troy_s | Nyquist levels yes?
| |
23:49 | troy_s | Below them?
| |
23:50 | comradekingu | Nyquist levels determine a frequency, so then you are using a cable dimensioned which has to be non-broken
| |
23:50 | troy_s | All quite interesting stuffs.
| |
23:50 | troy_s | Dark alchemy.
| |
23:51 | troy_s | Some of the most interesting imaging code is frequency domain stuffs.
| |
23:51 | comradekingu | Its not alchemy at all, its plain math
| |
23:51 | comradekingu | Anyone trying to tell you different is also a cable-salesman
| |
23:51 | troy_s | Alchemy. It is lower than the floor of my understanding. ;)
| |
23:52 | aombk | the important thing to remember is. someone got 6mil. funding because he persuaded neil young to persuade his friends to say a few nice words about the "feeling" this portable flac player gives you
| |
23:52 | aombk | like vinyl!
| |
23:52 | seku | too many salesmen and not enough listeners in the industry tho :(
| |
23:52 | troy_s | seku: Again, aesthetics are emergent. :)
| |
23:52 | aombk | it probably has a stereo image expander built in
| |
23:53 | troy_s | To be fair to vinyl, the vinylistas have a valid point on one vantage; most vinyl is mastered very differently to generic MP3s
| |
23:54 | troy_s | And that mastering can outweigh the image disparity.
| |
23:54 | troy_s | Perhaps more so these days, where vinyls are limited / exclusive releases
| |
23:55 | comradekingu | *otherwise
| |
23:55 | troy_s | comradekingu: ?
| |
23:55 | aombk | i knew i shouldnt post these links :P
| |
23:55 | comradekingu | i was in err, and returned to have my offset corrected
| |
23:56 | troy_s | I have entirely failed to comprehend even one word of a sentence online. Again.
| |
23:57 | comradekingu | The joke is you need to align the offset of your cd-rom otherwise you cant rip CDs with matching filesums
| |
23:57 | wescotte | joined the channel | |
23:58 | comradekingu | brb sometime later
| |
23:59 | troy_s | Bertl: how goes the plan?
| |
23:59 | Bertl | forward
| |
00:00 | troy_s | Bertl: Did the EOS mount hit funding?
| |
00:01 | Bertl | what do you mean with 'hit'?
| |
00:04 | seku | the active mount? barely hit funding.
| |
00:05 | seku | was at 120k :)
| |
00:05 | seku | well gnight, bedtime
| |
00:07 | Bertl | night!
| |
00:08 | troy_s | Bertl: Reached goal. That is a helluva lot of work.
| |
00:09 | Bertl | ah, yeah, well, we'll see :)
| |
00:09 | troy_s | Bertl: Seeing as how the first milestone appears to support two sensors, any chance of a file moniker when you get to raw? As in a teeny file indicator somewhere in the blob?
| |
00:10 | troy_s | (I know... Discussed previously.)
| |
00:10 | Bertl | we will most likely use the MLV format, which supports a lot of metadata
| |
00:10 | troy_s | Ugh
| |
00:10 | Bertl | it is well structured, and there is already good support
| |
00:12 | danieel | which sw supports mlv? i do not recall any grading suites to open these...
| |
00:12 | troy_s | Ugh
| |
00:12 | troy_s | Where is the spec?
| |
00:12 | troy_s | Correct
| |
00:12 | troy_s | Going to piss off an entire audience base overnight
| |
00:12 | danieel | as for metadata, even in dng you can have proprietary data islands, if that is your target :)
| |
00:13 | seku | afaik mlv was born out of the misery of Canons firmware : the format needed to index a lot of out-of-order data and save it correctly to the limited speed of a 100mbyte/sec CF card
| |
00:13 | g3gg0 | hi
| |
00:13 | seku | hi g3gg0 , you can explain that way better than i
| |
00:14 | troy_s | (Not to knock a decent format of course... But plays well with others is a pitfall.)
| |
00:14 | troy_s | Where is the spec though? And a sample?
| |
00:14 | troy_s | (Bayered ideally)
| |
00:15 | seku | afaik grading suites are mostly fine with CinemaDNG. compressed cinemadng would be even nicer :)
| |
00:15 | g3gg0 | mlv was born because there is no format that would support raw bayer data without having to tweak on several points
| |
00:15 | troy_s | g3gg0: Spec and sample?
| |
00:16 | g3gg0 | please dont rush
| |
00:16 | troy_s | LOL
| |
00:16 | troy_s | I wasn't aware I was. Apologies.
| |
00:16 | g3gg0 | its already late and it takes a few seconds
| |
00:16 | troy_s | I just would like to try and see if I can get an OIIO stub working
| |
00:17 | g3gg0 | https://bitbucket.org/hudson/magic-lantern/src/4d5cf113a68128c866eb788cd7ea3245e81fcc34/modules/mlv_rec/mlv.h?at=mlv_rec_lj92
| |
00:17 | g3gg0 | thats the format as header
| |
00:18 | g3gg0 | https://docs.google.com/spreadsheet/ccc?key=0AgQ2MOkAZTFHdHJraTVTOEpmNEIwTVlKd0dHVi1ULUE&usp=drive_web#gid=1
| |
00:18 | troy_s | Thanks.
| |
00:18 | g3gg0 | thats some more detailed explanation
| |
00:18 | g3gg0 | this was the base before the header was created
| |
00:19 | troy_s | So RTC is a time code chunk?
| |
00:19 | g3gg0 | we used to edit it that way before creating some headers that could be used to build the files
| |
00:19 | g3gg0 | RTCI is the real time clock information
| |
00:19 | troy_s | Okie.
| |
00:20 | g3gg0 | every (meta)data block has a µsec timestamp
| |
00:20 | troy_s | I wonder if we can generate a simulated MLV using CMV12000 data.
| |
00:20 | g3gg0 | sure
| |
00:20 | g3gg0 | very simple
| |
00:21 | g3gg0 | [MLVI] [RAWI] [VIDF]
| |
00:21 | troy_s | First is header?
| |
00:21 | g3gg0 | yep
| |
00:21 | troy_s | Second data blob?
| |
00:21 | troy_s | Third...
| |
00:21 | g3gg0 | RAWI = RAW information
| |
00:21 | troy_s | Yep
| |
00:21 | g3gg0 | dimensions, depth etc
| |
00:21 | g3gg0 | VIDF= video frame
| |
00:21 | troy_s | Oh OK.
| |
00:21 | troy_s | So a metadata.
| |
00:22 | g3gg0 | yep
| |
00:22 | troy_s | Is the meta variable length?
| |
00:22 | g3gg0 | MLVI appears on every file at offset zero
| |
00:22 | troy_s | (I have an OIIO stub already for the raw CMV12000 shots that loads fine in my lab thingy.)
| |
00:23 | troy_s | Rawi is fixed too?
| |
00:23 | g3gg0 | all blocks basically support variable length. for most blocks it wont change though. only INFO blocks that contain ascii text are variable iirc
| |
00:23 | g3gg0 | all blocks share the same format
| |
00:23 | troy_s | g3gg0: So is this a V1 or pre-v1 implementation?
| |
00:24 | g3gg0 | ?
| |
00:24 | g3gg0 | v1?
| |
00:24 | troy_s | Version one. How stable?
| |
00:24 | troy_s | I expect there is a 1.1 or something likely for Axiom
| |
00:24 | g3gg0 | this is v2.0
| |
00:25 | troy_s | g3gg0: In your estimation then, how much would potentially (guess) shift to support Axiom?
| |
00:25 | g3gg0 | for axiom - as bertl explained - there would be an addition to support raw data that is not going from TL to BR, but that is written as stripes
| |
00:25 | danieel | g3gg0: the structures are LE or BE ?
| |
00:25 | g3gg0 | LE
| |
00:25 | troy_s | Temporal stripes?
| |
00:25 | g3gg0 | spatial
| |
00:26 | troy_s | So unordered?
| |
00:26 | troy_s | But same frame?
| |
00:26 | g3gg0 | iirc bertl explained that the sensor data isnt read out from left to right, but its sectioned and interlaced
| |
00:26 | troy_s | Right. So same temporal frame, but shuffled batches.
| |
00:27 | Bertl | the idea is to allow for recording sensor data without reordering it
| |
00:27 | troy_s | And dumping accordingly frees up processing.
| |
00:27 | g3gg0 | this format isnt explicitely supported by the format's metadata yet. i.e. the writer cant tell the reader how the data is ordered.
| |
00:27 | danieel | come on, you can do that little shuffling in the fpga
| |
00:28 | Bertl | yes, we do it in the FPGA already on the Alpha
| |
00:28 | danieel | so why not on beta?
| |
00:28 | troy_s | It is like having the old guys shouting from the muppets.
| |
00:28 | Bertl | but if you want to maximize battery for example, you don't want to do that
| |
00:28 | troy_s | Ah. Good point.
| |
00:29 | g3gg0 | MLV sample files: http://magiclantern.fm/forum/index.php?topic=11899.msg130484#msg130484
| |
00:29 | troy_s | Not that batteries are a tremendous issue given the contexts. But still valid.
| |
00:29 | troy_s | g3gg0: What sensor there though? Those raws?
| |
00:29 | Bertl | or reduce heat :)
| |
00:29 | g3gg0 | 14bpp canon eos
| |
00:30 | troy_s | g3gg0: is there any lab processing software out there right now for the MLV format?
| |
00:30 | g3gg0 | to use tools like mlrawviewer from chmee, i recommend to stick to 14bpp formatting until he added variable width
| |
00:30 | g3gg0 | you mean for debugging the format?
| |
00:30 | troy_s | I do like the leaner idea of MLV. OpenDNG has a pig of crap in it.
| |
00:31 | g3gg0 | yeah thats one reason for mlv.
| |
00:31 | danieel | g3gg0: future additions shall include multi-stream (frame based hdr), multi roi support (to tag the parts) and possibly specifying dense bit packing.... the nonstandard application specific striping makes little sense
| |
00:31 | seku | mlv also converts fine to cinemadng.
| |
00:32 | troy_s | Hrm. I wonder what he(?) Is using for DNG reading.
| |
00:32 | troy_s | Probably a tiff hack.
| |
00:32 | danieel | what dng does better is to tag the optically black areas - you might want to do that too
| |
00:32 | g3gg0 | danieel: can you note that somewhere (wiki, forum) for discussion?
| |
00:32 | danieel | copy it yourself. i will not repeat myself :)
| |
00:33 | seku | me? using chmee's raw2cdng
| |
00:33 | troy_s | danieel: Optically black?
| |
00:34 | troy_s | Who is the author of MLRawViewer?
| |
00:34 | g3gg0 | danieel: uhm its half past two and i am typing from my phone in bed. maybe i cannot remember every detail you wrote.
| |
00:34 | danieel | g3gg0: is there any indexing for quick seeking?
| |
00:34 | danieel | pm my email, will copy it there and you can sort it out tomorrow
| |
00:34 | g3gg0 | troy_s: baldand, andrew baldwin
| |
00:34 | danieel | *me
| |
00:35 | g3gg0 | danieel: .idx files containing a header and only mlv_xref entries
| |
00:36 | g3gg0 | danieel: any reader application may store those .idx for later processing or recreate on its own
| |
00:38 | troy_s | Interesting. He wrote the tag parser.
| |
00:40 | g3gg0 | troy_s: baldand?
| |
00:41 | troy_s | Yes
| |
00:45 | g3gg0 | troy_s: main goal for MLV was to get the raw data out of the camera as quick as possible, with no preprocessing except cropping out of the full frame and as much metadata as possible. converting to interchange formats like CDNG is stuff for a workstation
| |
00:45 | troy_s | g3gg0: It's a decent enough goal, but the overlap in the design is questionable to skip.
| |
00:46 | troy_s | g3gg0: Debatable of course.
| |
00:46 | g3gg0 | many tools are very picky about CDNG.
| |
00:46 | g3gg0 | yeah
| |
00:46 | troy_s | g3gg0: I've sort of been plucking away on a lab software because I'm interested in getting footage into an offline state, so dealing with the demosaic etc. is useful.
| |
00:47 | troy_s | g3gg0: But he has a LUT support in there, so you are almost into the one-stop-shop for editorial... might as well take it that last 2% IMO.
| |
00:47 | troy_s | g3gg0: A few things that can be imrpoved almost immediately is tacking on OCIO and OIIO support. Gains you offline dumping to EXR / DPX immediately and full LUT / color support.
| |
00:47 | danieel | g3gg0: send few extras to think about
| |
00:47 | danieel | *sent
| |
00:47 | troy_s | g3gg0: But I am just looking at his code.
| |
00:48 | troy_s | I haven't looked at amaze much. I probably should. I need to start doing RMSEs on generated bayers and see which ones hold up ideally.
| |
00:49 | g3gg0 | OIIO sounds interesting, never tried though
| |
00:49 | troy_s | g3gg0: OCIO and OIIO are reference standards really.
| |
00:50 | troy_s | g3gg0: And about as dead easy to implement. With python bindings.
| |
00:50 | troy_s | I'd be skeptical of the shader code
| |
00:50 | troy_s | OCIO allows for GPU acel.
| |
00:50 | g3gg0 | as they are already 4 byte names, consider them being perfect MLV tags already ;)
| |
00:50 | troy_s | g3gg0: Well adding OIIO gets you TIFF/DPX/EXR immediately, with the latter two being most important.
| |
00:51 | troy_s | and OCIO opens up CDL for free.
| |
00:51 | g3gg0 | i will read into it and check if there is a simple way to add it. thanks for pointing there.
| |
00:51 | g3gg0 | but its already late, 4h left for sleep
| |
00:51 | troy_s | Among a plethora of other useful things. Ideally someone adds in CinemaDNG or OpenDNG to the OIIO base. Mr. Gritz is a god and would surely add it.
| |
00:51 | troy_s | g3gg0: Ping me on the morrow.
| |
00:52 | g3gg0 | will do. good night everyone
| |
00:52 | troy_s | Wish my GL were better.
| |
00:52 | danieel | gn
| |
00:54 | troy_s | Wholly god does the shader code suck in ellipses.
| |
00:54 | troy_s | Drives me to the point of being unable to read it. At all.
| |
00:54 | troy_s | There has got to be a better way.
| |
00:56 | g3gg0 | left the channel | |
01:01 | danieel | the shader code is usually ugly, depends what version of the compiler you can use
| |
01:01 | wescotte | left the channel | |
01:01 | danieel | opencl is better in that respect, but does not run everywhere
| |
01:06 | troy_s | danieel: I've found a decent method to load files. I like the idea of the increased performance, but little things bug me like how to get bit depth etc of the shader.
| |
01:08 | danieel | you mean the computational format?
| |
01:22 | troy_s | danieel: I guess so. I'd like to know that a shader is operating at 32 bit float for example.
| |
01:27 | danieel | should be 32bit (SP)... the DP rating for gpus is 64bit float
| |
01:27 | danieel | it depends on the texture format you use
| |
01:28 | troy_s | danieel: Just a standard fragment shader. I'm just curious how to enum / query that.
| |
01:28 | danieel | make a test to see when it looses precision
| |
01:28 | troy_s | No. I won't. There has got to be a query.
| |
01:28 | troy_s | That's junk.
| |
01:28 | troy_s | Egads.
| |
01:28 | troy_s | If that is how you have to test depth, shaders are fricking hopeless. I cannot imagine that is the method.
| |
01:29 | danieel | well... in the gpu terminology we can start complaining about dimensions being specified in floats
| |
01:29 | danieel | how do you get a pixel perfect mapping from that?:)
| |
01:29 | troy_s | I'm cool with that... I just want a query.
| |
01:29 | troy_s | Partials I'm cool with.
| |
01:29 | troy_s | But I'd like to know what depth I'm at.
| |
01:30 | danieel | ... texture format
| |
01:31 | troy_s | danieel: Any links?
| |
01:36 | danieel | the GLSL specs state 3 precisions - highp/mediump/lowp, with no meaning :)
| |
01:37 | danieel | section 4.7.2 at https://www.opengl.org/registry/doc/GLSLangSpec.4.50.pdf
| |
01:45 | Bertl | unsing floats for dimensions/coordinates isn't a bad idea per se, normalizing them OTOH ...
| |
01:48 | danieel | it causes rounding problems...
| |
01:49 | danieel | so while a nearest neighbour works, a better interpolation will produce slightly different result
| |
02:25 | intracube | left the channel | |
02:33 | 1JTAANITI | joined the channel | |
02:33 | 7JTAA6UAQ | joined the channel | |
02:34 | Bertl | off to bed now ... have a good one everyone!
| |
03:02 | taschenguger | left the channel | |
03:14 | 1JTAANITI | left the channel | |
03:14 | 7JTAA6UAQ | left the channel | |
03:17 | intracube | joined the channel | |
03:32 | wescotte | joined the channel | |
03:43 | intracube | changed nick to: intracube_afl
| |
03:43 | intracube_afl | changed nick to: intracube_afk
| |
03:57 | wescotte | left the channel | |
03:57 | wescotte | left the channel | |
04:35 | wescotte | joined the channel | |
04:52 | aombk | is anyone here?
| |
05:01 | intracube_afk | left the channel | |
06:01 | Gegsite | joined the channel | |
06:32 | wescotte | left the channel | |
06:57 | g3gg0 | joined the channel | |
06:59 | Gegsite | left the channel | |
07:03 | KurtAugust | joined the channel | |
07:12 | KurtAugust | Goodmorning! Rise and shine.... Question to @ApertusOSCinema Beta backers: would you like it to be designed so you won't need an extra camera cage? Sturdy housing with cable protection and lots of threaded holes. For instance for a monitor. Also strong enough to use it a follow focus and not having any flex (is often difficult with the rubber bases of a dslr). This will of course increase the cost of manufacturing, so I can imagine some
| |
07:12 | KurtAugust | developers wouldn't mind something less. I would just think that the added expense is still much less than any aftermarket solution and prolongs the camera's life. And since it's designed as a box anyhow...
| |
07:13 | se6astian|away | changed nick to: se6astian
| |
07:15 | KurtAugust | Hey, are the irc logs broken?
| |
07:16 | se6astian | good morning
| |
07:16 | se6astian | indeed the sunday one is empty
| |
07:16 | se6astian | I will check it later
| |
07:17 | KurtAugust | And I was thinking you finally took some rest.
| |
07:19 | KurtAugust | left the channel | |
07:27 | KurtAugust | joined the channel | |
07:29 | KurtAugust | Burglary at Apertus: millions worth of logs stolen!
| |
07:30 | se6astian | "a thief copied large amounts of highly scandalous internal communication from the high security server"
| |
07:30 | se6astian | ok gotta go :)
| |
07:30 | se6astian | need to sell the logs on the black market!
| |
07:30 | se6astian | changed nick to: se6astian|away
| |
07:35 | KurtAugust | left the channel | |
08:04 | KurtAugust | joined the channel | |
08:10 | ItsMeLenny | joined the channel | |
08:11 | KurtAugust | left the channel | |
08:38 | __anton__ | joined the channel | |
08:47 | philippej_ | joined the channel | |
08:47 | philippej | joined the channel | |
08:56 | KurtAugust | joined the channel | |
08:59 | KurtAugust | left the channel | |
09:18 | intracube | joined the channel | |
09:23 | ThatCantBe | left the channel | |
09:44 | ItsMeLenny | left the channel | |
10:43 | danieel | its funny how one spends $100M's on a movie, and at the end it sucks because all the VFX is global shutter like while real scenes got heavy rolling shutter artefacts
| |
10:49 | intracube | danieel: movie in question?
| |
10:52 | danieel | newest transformers
| |
10:53 | danieel | bay's look is no way possbile with rolling shutter cameras
| |
11:04 | intracube | danieel: I couldn't see any obvious signs in the transformer 4 trailer
| |
11:24 | Bertl | morning folks!
| |
11:35 | __anton__ | hi, looks like irc archive isn't showing today's log
| |
11:35 | __anton__ | http://irc.apertus.org/index.php?day=12&month=10&year=2014 is empty for me
| |
11:37 | Bertl | indeed, let me check
| |
11:44 | Bertl | test
| |
12:15 | tyrone_ | joined the channel | |
12:16 | tyrone_ | uhhh this sunday is quiet :-)
| |
12:17 | Bertl | nah, the logger ran out of space, is already fixed
| |
12:17 | Gegsite | joined the channel | |
12:17 | Gegsite | left the channel | |
12:17 | Legendin | joined the channel | |
12:18 | tyrone_ | ahh okey thaks bertl..... the log was looking strange :-)
| |
12:20 | tyrone_ | bertl do you know if somebody from the apertus team will go to the vision exhibition in stuttgart?
| |
12:37 | 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
| |
12:37 | se6astian | has set the topic | |
12:38 | Bertl | test3
| |
12:38 | tyrone_ | you can talk directly to the image sensor manufactures and much other companys from europe...
| |
12:38 | Bertl | okay, logs should be back and working
| |
12:40 | tyrone_ | last time i was there in i could talk to the person from frauenhofer who has created of the image seonsor for the first digital camera from arri... it was intressting
| |
12:41 | danieel | d20 or that 1990's military one?
| |
12:41 | tyrone_ | d20
| |
12:42 | danieel | most of the people working on arri came from frauenhofer research..
| |
12:43 | tyrone_ | that i didn't know....
| |
12:45 | tyrone_ | do you know if there is a change with the access from the documents from truesense because on semi has bought them?
| |
12:46 | tyrone_ | okey it was in april the takeover...
| |
12:47 | danieel | probably not, they finished the transfer like a week or two ago
| |
12:52 | danieel | do you plan to go stuttgart?
| |
12:53 | tyrone_ | i hope i can go it's for me just 2h drive so not far away....
| |
12:59 | tyrone_ | danieel does ON Semi has for the other sensors(example VITA) a NDA for the specification manual or is it public?
| |
13:03 | danieel | public, but you could figure that out if you've been actually looking :)
| |
13:09 | tyrone_ | okey i think i'm stupid :-) i couldn't finde the spec on the site..... google was better....
| |
13:35 | danieel | the DS is on the product page :)
| |
13:45 | tyrone_ | perfect just the vita12k and 16k have a reduced datasheet...... every other one is with full specs even the vita 25k.....
| |
14:11 | Legendin | left the channel | |
15:11 | tyrone_ | left the channel | |
16:20 | sebix | joined the channel | |
16:33 | se6astian|away | changed nick to: se6astian
| |
16:33 | se6astian | good evening
| |
16:54 | sebix | left the channel | |
16:59 | troy_s | danieel: I can check, but pretty sure it was shot on a global shutter camera.
| |
17:11 | aombk | well, rolling shutter was not transformers biggest problem
| |
17:11 | aombk | *if it was shot whit rolling shutter equipment
| |
17:15 | troy_s | Let me be clear... There are no problems
| |
17:15 | troy_s | Domestic:Â $245,435,927Â Â Â 22.7%+Â Foreign:Â $835,500,000Â Â Â 77.3%=Â Worldwide:Â $1,080,935,927
| |
17:16 | troy_s | When your metric of problem / not problem is money, there are no problems. At all.
| |
17:17 | troy_s | That puts it at slot 11 of all time worldwide grosses.
| |
17:17 | danieel | troy_s: imdb/technical says red dragon, but the artefact is worse/comparable to what we measured on epic
| |
17:18 | intracube | danieel: are there any clips online that show the issue?
| |
17:18 | danieel | probably not.. i can take 2 snapshosts if you need
| |
17:18 | troy_s | danieel: May have been. You also have to remember that Epics are used when cameras are going to get crunched.
| |
17:19 | troy_s | danieel: So any high-action sequence, where high risk to camera, you can almost be certain it is a POS Epic; the Eyemo of the new era.
| |
17:19 | danieel | it is most visible in nonvfx action shots (people figting, camera panning quickly), let me find it
| |
17:19 | troy_s | And doubly certain on a film where the primary camera is a Dragon.
| |
17:20 | danieel | wasnt it supposed to be shot on imax/3d ?
| |
17:20 | troy_s | Dragon has a global no?
| |
17:20 | aombk | the seccond transformer had compositing problems too
| |
17:20 | danieel | no
| |
17:20 | danieel | global can be simulated for <180deg with an liquid crystal shutter
| |
17:20 | troy_s | aombk: Cannot be avoided. There are tracks that come back from Mumbai with slippage during tracking for example, and if the schedule has slipped, it gets rolled with.
| |
17:21 | troy_s | danieel: Rare to shoot under the 180. Is that what they were doing?
| |
17:22 | danieel | nope, give me a sec to dump the frames
| |
17:23 | aombk | the new red sensor does that thing with liquid crystal shutter
| |
17:23 | aombk | actually the new mount
| |
17:27 | gcolburn | joined the channel | |
17:28 | aombk | http://s578.photobucket.com/user/rajeshseeni/media/1.jpg.html
| |
17:30 | troy_s | Tapered pillars.
| |
17:42 | danieel | http://home.rozsnyo.com/hollywood-loves-rolling-shutter.jpg
| |
17:44 | intracube | danieel: that is from what film?
| |
17:44 | danieel | transformers, age of extinction
| |
17:45 | danieel | once you start making a camera, you cant watch movies... RS besides the poor dynamic range is most noticable :)
| |
17:45 | Bertl | wb gcolburn!
| |
17:48 | intracube | speaking of dynamic range and highlights: http://images.static-bluray.com/reviews/6812_20_1080p.jpg
| |
17:49 | troy_s | intracube: Your point?
| |
17:49 | troy_s | White and black are creative decisions.
| |
17:50 | troy_s | I would hate to see a world where everyone avoids white. See Robert Richardson's earlier work where he blew out with rear nets etc.
| |
17:50 | danieel | prometheus was quite good.. was not distracting much
| |
17:51 | intracube | troy_s: it's that they didn't avoid it (when shooting) then tried to bring it down in post
| |
17:51 | intracube | at least that's what it looks like
| |
17:52 | danieel | intracube: if you see the whole movie it had a point.. was set in averagely dark gray, leaving brighter levels for when needed
| |
17:52 | danieel | you can adapt your range level to the darks... and then they surprise you with a sun or light...
| |
17:52 | danieel | if you keep the mood wide range, you cant do exceptional things
| |
17:52 | intracube | danieel: I think you're missing my point
| |
17:53 | intracube | it's that the highlights on the upper screens look heavily clipped
| |
17:53 | intracube | then they've been brought back to non peak levels where you'd expect to see detail
| |
17:53 | intracube | it looks jarring IMO
| |
17:53 | danieel | its intentional bring down as i wrote.. but yes, there is lack of detail :)
| |
18:01 | intracube | in practice you'd need some frame of renference to peak white
| |
18:01 | intracube | so it really depends on the shots leading up to it
| |
18:01 | troy_s | Pretty sure the peeps helming know how to shoot. ;)
| |
18:01 | troy_s | Just saying'.
| 18:02 | intracube | loves to be an arm chair critic
|
18:02 | troy_s | If at any point someone felt that was a mistake, I can state beyond a shadow of a doubt that particular shot would have been reshot. Nothing to it.
| |
18:03 | Bertl | intracube: do you have a nice arm chair?
| |
18:03 | troy_s | I will say it again for folks that perhaps didn't grasp / embrace / comprehend:
| |
18:03 | troy_s | Aesthetics are emergent.
| |
18:04 | troy_s | Could Bertl craft a nice PLE that gathers up a much larger DR? Absolutely. Should a DP throw out all but one stop of it? Absolutely.
| |
18:04 | intracube | Bertl: :P
| |
18:05 | troy_s | Aesthetics go through the typical cycles of “out of place with purpose†“fresh new bleeding edge†“mainstream†“cliched vernacularâ€
| |
18:05 | troy_s | See Jump Cut.
| |
18:05 | troy_s | See Cross Cut.
| |
18:05 | troy_s | See cinema as coffee house spectacle, to narrative / non-narrativr
| |
18:06 | Bertl | very interesting what I'm saying here?
| |
18:06 | troy_s | ?
| |
18:06 | Bertl | ah, nevermind, window update problem
| |
18:07 | aombk | aesthetics are emergent but still, sometimes people, even professionals, dont know what they are doing
| |
18:11 | sebix | joined the channel | |
18:11 | aombk | and this is very evident and clear in transitional periods, for example analog to digital
| |
18:12 | aombk | which in film is, among other things, chemistry to math (the color correction aspect of it)
| |
18:26 | troy_s | aombk: Perhaps surprisingly, many high profile projects go out of their way to research before shooting due to the experience front.
| |
18:27 | troy_s | So I'll call into question the trope that many professionals “don't know what they are doingâ€
| |
18:27 | troy_s | Mainstream film is the most conservative industry you would ever see. Mistakes are made, heck yes, but in terms of attempting to insulate against them from a shooting vantage? There are many layers.
| |
18:28 | troy_s | Everyone from the DP, to assistants, to other HoDs have varying experiences that insulate.
| |
18:29 | troy_s | There are many legendary stories from the film era of mistakes too. They happen, but by and large, studios don't generally let Dondald Duck and Mickey helm hundred million dollar projects.
| |
18:32 | aombk | research doesnt always reward us with results
| |
18:32 | aombk | so you disagree that sometimes people, even professionals, dont know what they are doing
| |
18:34 | troy_s | aombk: I am offering my opinion purely as anecdotal from experience. In my personal experience, one can have a gripe with aesthetic choices etc, but I have met very few that “don't know what they are doingâ€.
| |
18:34 | troy_s | And the system is designed against that sort of failure, as there are many layers of insulation.
| |
18:35 | troy_s | (EG a DP may be thinner on color knowledge, but has a generally experienced DIT on their side.)
| |
18:35 | aombk | maybe we should define "dont know what theyre doing"
| |
18:35 | intracube | troy_s: who has said that many professionals don't know what they're doing?
| |
18:35 | troy_s | I just think it is an overly simplistic trope peddled by the masses to try and explain Bay-esque choices.
| |
18:36 | troy_s | aombk: Exactly. Privileged rubbish terms.
| |
18:37 | troy_s | intracube: Scroll back.
| |
18:37 | aombk | intracube, i said something like that
| |
18:38 | intracube | oh ok
| 18:38 | intracube | wasn't paying full attention
|
18:39 | troy_s | It is just too simplistic to attribute that concept to other complex systems.
| |
18:40 | aombk | troy_s, well in the end, i dont agree with you on that
| |
18:41 | troy_s | aombk: Sure. Feel free to disagree. There is a smidge of experience behind my anecdotal evidence though.
| |
18:43 | troy_s | There is also the very real fact that seasoned veterans are often doing things in mainstream movie making that well... Haven't been done before. So of course they don't know what they are doing.
| |
18:44 | aombk | the masses never judge the complex process and the amount of work and research professionals put in to produce something
| |
18:44 | troy_s | Sure. But it is also ideological I believe.
| |
18:44 | aombk | what is?
| |
18:45 | troy_s | If we say “That movie sucked†is is basically a statement of constancy (a myth) and a checksum of extremely complex things.
| |
18:45 | troy_s | I cannot count for example, the number of times my opinion on various bits of cinema have shifted and changed. The myth of constancy is a foolish one.
| |
18:46 | troy_s | And over time, that constancy is more assaultable. See DW Griffith's earlier cinema for example, which at once is celebrated as monumental feats and also simultaneously grotesquely racist.
| |
18:46 | troy_s | Cinematography falling into the precise same zone.
| |
18:47 | troy_s | aombk: Sense?
| |
18:48 | aombk | yes but i dont know where to start. i disagree to the core of it.
| |
18:48 | aombk | are you a dp?
| |
18:49 | g3gg0 | hi
| |
18:50 | troy_s | What core?
| |
18:50 | troy_s | Hello g3gg0
| |
18:50 | g3gg0 | hi troy_s
| |
18:50 | troy_s | aombk: What premise lay at the core that you disagree with.
| |
18:50 | troy_s | That professionals do or do not know?
| |
18:51 | troy_s | There is a huge breadth there, as professional is a rubbish term for anyone that has been paid to ply their body to something.
| |
18:51 | aombk | at the core of what you are saying i think lies relativism
| |
18:51 | troy_s | Yes. Of course it does.
| |
18:51 | aombk | postmodernist relativism
| |
18:51 | troy_s | Plato sort of screwed that pooch for us
| |
18:51 | troy_s | Indeed.
| |
18:52 | aombk | hah well yes, i am not a big fan of this approach
| |
18:52 | troy_s | Or in actuality, a vastly more workable model for sociological manifestations? ;)
| |
18:52 | troy_s | Sure. But you would have a hard time taking that to the mat with this peep.
| |
18:52 | g3gg0 | troy_s: if i may ask, whats your background in film making?
| |
18:52 | troy_s | Spent too long getting my own belief system dismantled to disagree with it any longer.
| |
18:53 | troy_s | g3gg0: Studied film /photography in art school, shot / produced blah blah crap, and also exists as a minion to the machine.
| |
18:54 | g3gg0 | troy_s: are you behind the cam or mostly post processing?
| |
18:54 | g3gg0 | just curious
| |
18:54 | troy_s | Depends on what I am doing.
| |
18:54 | g3gg0 | so both
| |
18:54 | troy_s | On independent / personal / freelance, sometimes all.
| |
18:55 | troy_s | On some projects, such as album covers, sometimes none.
| |
18:56 | troy_s | Sadly know an unhealthy amount about colorimetry thanks to epic failures on my end when crafting work. :)
| |
18:56 | troy_s | Still trying to learn.
| |
18:57 | g3gg0 | hehe :)
| |
18:57 | g3gg0 | so colorspace conversion stuff is one of your competences?
| |
18:58 | g3gg0 | fought long to understand how to really convert raw bayer data into RGB
| |
18:59 | g3gg0 | the things behind the scenes. really good to know. then things make a lot more sense
| |
18:59 | PRN | joined the channel | |
18:59 | Bertl | welcome PRN!
| |
18:59 | g3gg0 | btw, hi Bertl
| |
19:00 | troy_s | g3gg0: Yes sadly.
| |
19:00 | PRN | Thanks Bertl
| |
19:00 | troy_s | g3gg0: Take chart. Develop 3D LUT, transform to XYZ. Transform to output +/- a Bradford or CAT02 in there. ;)
| |
19:01 | troy_s | g3gg0: But of course there are times when you want to only assert a linearized model for manipulation, and operate in the native RGB captures. Keying for example.
| |
19:02 | g3gg0 | CAT02? ok looking up. the other things i did myself too
| |
19:02 | troy_s | Linearized to assert your resultant associated alpha values represent occlusion, and native RGB to avoid crosstalk.
| |
19:02 | troy_s | g3gg0: CAT02 is just one of the newest breeds of CA.
| |
19:02 | g3gg0 | LMS color space
| |
19:02 | PRN | anyone from apertus team going to Broadcast India, Mumbai?
| |
19:03 | troy_s | g3gg0: All of the best CAs are LMS domain.
| |
19:03 | g3gg0 | ok is it the one related to XYZ, just with correct chromatic "distance" between colors?
| |
19:03 | troy_s | They have to be.
| |
19:03 | troy_s | g3gg0: Just a different model. Like a Bradford++.
| |
19:04 | g3gg0 | reminds me of http://xkcd.com/927/
| |
19:04 | troy_s | http://web.stanford.edu/~sujason/ColorBalancing/adaptation.html
| |
19:04 | troy_s | Not the same.
| |
19:04 | troy_s | This isn't about standards. It is more about attempting to simulate something. So there is a little more directionality.
| |
19:05 | g3gg0 | thanks for the link. my knowledge mostly is built upon the guides from http://www.brucelindbloom.com/
| |
19:05 | troy_s | Bradford superseded Von Kries, as VK superceded XYZ scaling.
| |
19:05 | troy_s | And CAT02 / Sharp are further refinements.
| |
19:05 | g3gg0 | it is also mostly used for white point adaption?
| |
19:05 | troy_s | Lindbloom is a canonized beast, with a slight ICC twist to most.
| |
19:06 | troy_s | Chromatic adaptation, yes.
| |
19:06 | troy_s | Solely.
| |
19:06 | troy_s | Although CAT02 is part of a larger thing.
| |
19:06 | troy_s | Which is a sort of newer breed of color, being Mr. Fairchild's CIECAM
| |
19:06 | troy_s | Color appearance models.
| |
19:07 | troy_s | Which attempt to solve some of the more nuanced bits of the psychophysical aspectd
| |
19:07 | troy_s | (As in many of the color phenomenon illusiony things)
| |
19:07 | g3gg0 | ah good to know. thanks
| |
19:07 | troy_s | Problem is, some / many may be cultural.
| |
19:08 | g3gg0 | still the most odd thing is the M_RGB<-XYZ which has to be determined
| |
19:08 | troy_s | (See Herskovits et al regarding Influence of Culture on Visual Perception)
| |
19:08 | troy_s | (Some postmodern relativism evidence there too for aombk. ;))
| |
19:08 | troy_s | g3gg0: Explain?
| |
19:09 | troy_s | Matrices will always fall short, and likely have some clusters of colors that suck.
| |
19:09 | troy_s | Same goes for display profiling.
| |
19:09 | g3gg0 | the the camera matrix, for converting XYZ to RGB. the most important thing to get the best images, but still it is being determined by shooting a color patch in sunlight.
| |
19:09 | troy_s | Not exactly
| |
19:10 | troy_s | That is ICCs lovely legacy in a domain it shouldn't be
| |
19:10 | g3gg0 | it is effective, yes. but there are colors you cannot reproduce with passive lighting
| |
19:10 | troy_s | The bottom line is that a camera is post spectral
| |
19:10 | g3gg0 | yep
| |
19:10 | troy_s | Meaning that RGB is not reality.
| |
19:10 | g3gg0 | yeah and every channel gets light that is not filtered ideal
| |
19:10 | troy_s | So in addition to our eyes being metameric, an RGB sensor is.
| |
19:10 | troy_s | Yes. Non-narrow band
| |
19:11 | Bertl | g3gg0: and that's good :)
| |
19:11 | troy_s | So a camera can never really reflect what it sees
| |
19:11 | troy_s | And will behave quite differently at different illuminants
| |
19:11 | troy_s | Blah blah
| |
19:11 | g3gg0 | and you never really can corellate back of which color the thing was that is being imaged
| |
19:12 | troy_s | Bottom line is that A) 3D LUTs are vastly more accurate B) you need one for each varying of contexts (ISO, illuminant, etc.)
| |
19:12 | troy_s | (Hence look at the Sony or Arri LUT sets in OCIO)
| |
19:12 | troy_s | g3gg0: Nope. Hence post-spectral
| |
19:12 | troy_s | But we can get close in many instances
| |
19:12 | g3gg0 | i tried to find the matrices various raw developing tools use.
| |
19:12 | troy_s | Close enough for post work
| |
19:12 | troy_s | Matrices suck
| |
19:13 | troy_s | Sensors just don't behave well enough.
| |
19:13 | g3gg0 | i tried to reconstruct a LUT from that. i somewhen stopped because i couldnt explain what ive seen there :)
| |
19:13 | troy_s | Bertl hammered that into my head a while back
| |
19:13 | g3gg0 | here it is https://www.youtube.com/watch?v=9Fz6V5aJSbk
| |
19:13 | g3gg0 | as said, it was just an experiment to better understand what happens behind the scenes
| |
19:13 | troy_s | The problem is that some shapes may cover a decent volume of the gamut, but when you get to the edges, your DE goes through the roof
| |
19:14 | troy_s | Hence why a 3D LUT will almost always be vastly more accurate.
| |
19:14 | g3gg0 | yeah absolutely
| |
19:14 | troy_s | (Stability of sensor notwithstanding)
| |
19:14 | troy_s | g3gg0: quit watching that crap
| |
19:14 | troy_s | Anytime anyone brings up Light room or ProPhotoRGB... Just worry.
| |
19:15 | g3gg0 | i made that ;)
| |
19:15 | troy_s | You have found yourself too stuck in ICC land.
| |
19:15 | troy_s | The ICC tools can be leveraged to your whim, but you sadly have to know their graphic arts hacks
| |
19:15 | g3gg0 | i dont have any other tools for developing CR2
| |
19:16 | g3gg0 | (aside of RT, dcraw, DPP)
| |
19:16 | troy_s | Namely the damn D50 PCS absoluteness.
| |
19:16 | troy_s | g3gg0: We need to chat more. ;)
| |
19:16 | troy_s | g3gg0: You are obviously power nerd and can compile OCIO and OIIO?
| |
19:17 | troy_s | OCIO was developed to supplant the limitations of ICCs
| |
19:17 | g3gg0 | troy_s: sure, just a lib with simple API
| |
19:17 | troy_s | In particular:
| |
19:17 | troy_s | A) Unlocking the forced chain of input-PCS-output
| |
19:17 | troy_s | B) Discarding the D50 rubbish
| |
19:17 | troy_s | C) Fine grained artistic control of transforms
| |
19:17 | g3gg0 | unfortunately no direct DNG support. didnt check if the TIFF module is capable of writing DNG-alike TIFFS
| |
19:17 | PRN | left the channel | |
19:18 | troy_s | D) Discarding the display referred-centric nature of ICCs
| |
19:18 | troy_s | g3gg0: I think you can manage.
| |
19:18 | troy_s | DPXs are pretty universal. And as I said, if someone added the tags for the TIFF side, quite sure Mr. Gritz would merge them for CinemaDNG.
| |
19:19 | troy_s | DPX and EXR still rule the roost (for good reason) in post land.
| |
19:19 | troy_s | g3gg0: But I think your transforms would greatly bump up a notch. You could end up with archival stored EXRs without breaking a sweat.
| |
19:20 | troy_s | (Scene linear too, which is a huge step toward post work)
| |
19:21 | troy_s | g3gg0: I forced OIIO to load a CinemaDNG, after all, it is a TIFF.
| |
19:21 | g3gg0 | so one of your concerns is to dump the raw sensor data as early as possible into some industry standard image format that uses an universal colorspace?
| |
19:22 | g3gg0 | (if i understood it right)
| |
19:22 | troy_s | There are no universal colorspaces. On the contrary.
| |
19:22 | g3gg0 | well, lets then call it standardized ones :)
| |
19:22 | troy_s | But for imaging, it does help having the image in a useful entry format.
| |
19:23 | troy_s | g3gg0: So for example, one might want to process raw to DPX and archive the DPXs.
| |
19:24 | troy_s | g3gg0: DPX due to the inherent log “compressionâ€, and EXRs due to their outright rejection of transfer curves and float depth that makes them the most ideal entry point for post work.
| |
19:24 | PRN | joined the channel | |
19:25 | troy_s | g3gg0: Sense?
| |
19:26 | g3gg0 | troy_s: what i dont get is - why would we ever care about any color space / LUT when writing data into an universal format like EXR. to my understanding all the data should match the scene as-shot. applying some LUT that looks the best on cinema screen or RGB monitors or CCTV is one of the last steps?
| |
19:26 | g3gg0 | or do you talk about "recovering" the real tristimulus from the bayer pixels using LUTs
| |
19:26 | troy_s | Ultimately most will want to start native.
| |
19:26 | troy_s | To avoid crosstalk for keying or such.
| |
19:27 | troy_s | “match the scene†is really the end of a 3D LUT, but that isn't ideal either. Ideal for an entry to grading, but not for post always.
| |
19:27 | g3gg0 | yes, matches my understanding
| |
19:27 | troy_s | Much of post isn't color critical, as long as there are assertions to not mangle the data.
| |
19:28 | troy_s | (So you still have to keep your eyes on it. Nightmare.)
| |
19:28 | Bertl | off for now ... bbl
| |
19:28 | Bertl | changed nick to: Bertl_oO
| |
19:28 | g3gg0 | bye Bertl!
| |
19:29 | g3gg0 | i just am frightened of baking color conversion matrices/LUTs into the data for archiving.
| |
19:29 | g3gg0 | assume the LUT wasnt good enough
| |
19:29 | troy_s | g3gg0: You wouldn't. Always view transforms.
| |
19:29 | g3gg0 | all your archived data is crap
| |
19:29 | troy_s | It is entirely plausible that an artist will need to see the work as it will appear to the grader's entry point.
| |
19:30 | troy_s | Hence you only tack that onto the view. (Most imaging apps have a distinct breakdown between the input / working slush / and view)
| |
19:30 | troy_s | Where working slush is always scene linear.
| |
19:30 | g3gg0 | yeah thats a viewer story then, right?
| |
19:30 | troy_s | No, it's part of the imaging tool. Nuke uses OCIO.
| |
19:31 | troy_s | You set up your views and can toggle between them. So you chain transforms to have a reference to output (say reference to 709) then tack on your 3D LUT display correction
| |
19:31 | g3gg0 | (ok i am not comfortable with all the words that professional PP guys are using :) )
| |
19:31 | troy_s | All the while your data is pure.
| |
19:31 | troy_s | Make sense?
| |
19:31 | g3gg0 | yes this absolutely does
| |
19:31 | troy_s | It isn't professional; it is all needs.
| |
19:32 | g3gg0 | hey, i shoot MP4 videos. this says it all? ;)
| |
19:32 | troy_s | If you fight through a short music video or indie, you learn pretty quick _why_ it makes sense.
| |
19:32 | troy_s | Even then, you can greatly improve your work by following that path.
| |
19:32 | troy_s | (I am a big believer in independents. Huge.)
| |
19:33 | g3gg0 | i am neither a professional videographer nor a photographer. i am reverse engineer and a programmer with hobby photographing :)
| |
19:33 | troy_s | The only cost is disk storage, which is quite a lot cheaper than it ever has been.
| |
19:33 | troy_s | Sure. You are a perfect candidate to muck with it then.
| |
19:33 | troy_s | And probably enjoy doing it.
| |
19:34 | troy_s | Even breaking your MP4s down into stills will increase the flexibility you have.
| |
19:34 | g3gg0 | ok back to one thing that bugs me...
| |
19:34 | troy_s | Everything from controlling the subsampling (a huge impact on your still image) to being able to composite say, fire, to getting to a correct film-like dissolve.
| |
19:35 | g3gg0 | shooting color patches in D65 sunlight to determine the color conversion matrix liek e.g. dcraw does
| |
19:35 | g3gg0 | to me this is only 50% of the story
| |
19:35 | troy_s | D50 to be more accurate.
| |
19:35 | troy_s | This is because they all rely on ICC tools.
| |
19:35 | troy_s | So matching the PCS is most ideal.
| |
19:36 | g3gg0 | you shoot patches and build a matrix that should match all colors that are physically possible.
| |
19:36 | troy_s | The further you are from the PCS, the more inaccuracies are introduced via the CA to get it there.
| |
19:36 | troy_s | Hard to do with reflective charts. You need light boxes etc. But that said, a HCT is a good chart. IT8s are solid too.
| |
19:37 | g3gg0 | yeah but the RGGB(raw) <- XYZ matrix you get from that is by far not correct
| |
19:37 | g3gg0 | yeah
| |
19:37 | troy_s | Depends. It is actually pretty accurate.
| |
19:37 | troy_s | The metal basically takes a few things:
| |
19:37 | troy_s | A) The chart. Known XYZ positions in 2D space
| |
19:38 | troy_s | B) A data file that contains spectrophotometer readings of the actual chart print unique to each chart.
| |
19:38 | troy_s | C) The native tri-color data (RGB file)
| |
19:39 | g3gg0 | especially with pink highlights a1ex found that a bunch of the common imagery tools produce RGB results that are far off the reality.
| |
19:39 | troy_s | You get two paths: matrix or 3D LUT. (Or tack on a shaper in there too)
| |
19:39 | troy_s | Matrix!
| |
19:39 | troy_s | You get that crap with matrices
| |
19:39 | troy_s | Less so with 3D LUTs
| |
19:39 | troy_s | But you also need to be acutely aware of your sensor and what it is doing
| |
19:39 | g3gg0 | but then you really need lightboxes that cover a wide spectral range, right?
| |
19:39 | troy_s | For example, if you feed it raw unclipped RGB.
| |
19:40 | troy_s | No... Think of this mistake
| |
19:40 | troy_s | You feed the chart (most often underexposed)
| |
19:40 | troy_s | (You need to look at the text file outputs to see ideal exposure. It is often 1-2 stops hotter than most folks shoot charts.)
| |
19:41 | troy_s | But least mean squares to matrix is like drawing a line through points
| |
19:41 | troy_s | Even if 90% are damn close to the libe
| |
19:41 | troy_s | Line, the ends are probably way off.
| |
19:41 | troy_s | Hence pink or magenta or yellow
| |
19:41 | troy_s | Or someone bogs up the CA
| |
19:41 | troy_s | There are _many_ places to screw profiling up.
| |
19:42 | g3gg0 | hehe absolutely
| |
19:42 | troy_s | If you would like, I can step through with Argyll and we can cut much of the crap out of the way and you can see the results
| |
19:42 | g3gg0 | especially when having channels that are overexposed
| |
19:42 | danieel | still the biggest issue remains the proper lighting of the chart
| |
19:42 | troy_s | But they can be damn good, MUCH more representative than the default matrices.
| |
19:43 | troy_s | danieel: That is large. But not largest.
| |
19:43 | troy_s | Many many many places where a minor bumble in knowledge hoops the result.
| |
19:43 | troy_s | Overexposed is bad too, but so is under.
| |
19:44 | troy_s | Also, seeing as how a sensor will never see pure black or white, you have to add those points manually.
| |
19:44 | g3gg0 | this is the reason why everyone hopes that making it community-driven will make it successful
| |
19:44 | danieel | that is easy to check, dont see it hard with exposure assist tools we have :)
| |
19:44 | troy_s | That will correct the glaring badness at the upper and low ends, preventing skewing off into grotesque color.
| |
19:44 | g3gg0 | axiom needs at least one guy who takes some time and produces a perfect* color profile :)
| |
19:44 | troy_s | There are a lot of damn sharp minds out there. Collective efforts are tough though.
| |
19:45 | PRN | left the channel | |
19:45 | g3gg0 | cant we measure the spectral response for every single bayer filter R, G1, G2, B and built a LUT from that?
| |
19:46 | danieel | they are measured
| |
19:46 | danieel | look at the QE charts from the sensor makers
| |
19:46 | troy_s | Not narrow band!
| |
19:46 | troy_s | Can not work.
| |
19:46 | troy_s | RGB != spectral.
| |
19:46 | troy_s | So two problems.
| |
19:46 | g3gg0 | nah i really think about following the spectral curve
| |
19:47 | troy_s | And, the sensors misbehave differently with different factors.
| |
19:47 | g3gg0 | they do?
| |
19:47 | troy_s | So ISO? Different LUT. Color temp? Different LUT.
| |
19:47 | troy_s | Etc.
| |
19:47 | troy_s | But the non-narrow band is the big killer there.
| |
19:48 | g3gg0 | ISO: right. color temp? no, that would suprise me. we talk about raw video :)
| |
19:48 | troy_s | Needless to say, decent profiles can be had if the sensor can be tamed.
| |
19:48 | g3gg0 | but sensor temperature will cause trouble
| |
19:48 | troy_s | Color temp of illuminant. ;)
| |
19:48 | troy_s | And temperature yes.
| |
19:48 | troy_s | And a bunch of other crap.
| |
19:48 | danieel | g3gg0: color temp is another curve, which defines the input light's spectrum...
| |
19:49 | g3gg0 | nah maybe i expressed myself wrong.
| |
19:49 | troy_s | The taming of the sensor is hardest.
| |
19:49 | danieel | might not be just one bump on the chart
| |
19:49 | troy_s | I cannot do anything about that.
| |
19:49 | g3gg0 | danieel: did all of that already by hand, so basics are known
| |
19:49 | troy_s | That is hardware nerdery.
| |
19:49 | troy_s | And mixed illuminants
| |
19:50 | troy_s | I was shocked at how awful the sensor behaved
| |
19:50 | troy_s | Really shocked
| |
19:50 | troy_s | I was always used to the "raw" data I saw off of a DSLR
| |
19:50 | troy_s | Little did I know how massaged that was
| |
19:50 | danieel | i think it is just the issue of the cmosis sensor
| |
19:50 | g3gg0 | troy_s: i really thought about measuring every single CFA (wideband) filter for their spectral response. e.g. like those http://www.astrosurf.com/buil/350d/350d.htm (but he made the mistake and didnt use RAW but PP-ed RGB)
| |
19:50 | troy_s | Bertl and danieel had plenty of insight on that front.
| |
19:51 | g3gg0 | or this one: http://www.astrosurf.com/buil/eos40d2/filter.htm
| |
19:51 | troy_s | g3gg0: Seems like a fool's errand to me, but I wouldn't try to insist someone not try.
| |
19:52 | troy_s | The bottom line is that all glass / plastic adds crosstalk
| |
19:52 | troy_s | To already non-spectral gathering
| |
19:52 | troy_s | So the battle is already fruitless
| |
19:52 | danieel | if the vienna guys can setup a proper lighting, i can bring my head and the hct, to make proper data to start with
| |
19:52 | troy_s | Varying levels of light will mix differently at different zones
| |
19:52 | danieel | around 28-29 oct, will be at pioneers.io
| |
19:53 | troy_s | Hell, people loved shooting film _because_ of the unique crosstalk
| |
19:53 | troy_s | Now we evolve away from that, toward gathering of scene data, and attempt to leave that to the creative ending point to control.
| |
19:54 | troy_s | danieel: Do you have the text file with the HCT?
| |
19:54 | danieel | yes, matched with the batch of the print
| |
19:54 | troy_s | If you get even lighting, you should get a pretty decent set of profiles. I would do ISO and other variables too so you can get a few.
| |
19:54 | troy_s | Yep.
| |
19:54 | troy_s | Solid charts.
| |
19:55 | troy_s | I wonder if there are high end light boxes that tag extremities of the spectral locus.
| |
19:56 | g3gg0 | what i (naively) thought is, getting the spectral response for every CFA filter R, G1, G2 and B to make up a matrix. this matrix is the coordinate system for the colors being measured. e.g. if R/G/G/B is 0/0/0/1 then the real input color can be in range of the spectral response of the blue filter. (e.g. this plot: http://www.astrosurf.com/buil/eos40d2/graph5.png ). i know that recovering a color in XYZ space from the CFA output is as complex a
| |
19:57 | g3gg0 | this would be the most (mathematically) correct method to get the "real" color of the thing being shot
| |
19:58 | g3gg0 | processing an image would take 1k times as long, depending on the accuracy of the spectral response curve complexity
| |
19:58 | g3gg0 | but i bet astroguys would kiss us for that ;)
| |
19:59 | troy_s | Still won't work
| |
19:59 | troy_s | Because you also ignore the objecy
| |
19:59 | g3gg0 | at the moment matrix-tools work as if the CFA would be narrowband
| |
19:59 | troy_s | Object, which will also behave differently
| |
19:59 | g3gg0 | so all colorspaces do
| |
19:59 | troy_s | Read the Argyll list on watercolors recently
| |
19:59 | troy_s | There is some deadly complex stuffs that happen as anyone that does imaging knows too.
| |
20:00 | troy_s | For example, were you aware that RGB rendering models will result in different results for indirect lighting?
| |
20:00 | troy_s | As in your choice of primaries matter.
| |
20:00 | g3gg0 | what my proposed method returns is not the exact color, but the color probability. there are no real primaries anymore
| |
20:01 | g3gg0 | primaries = assuming narrowband
| |
20:01 | troy_s | And on your point of matrices, hence why I am a 3D LUT believer. They model crosstalk vastly more accurately.
| |
20:01 | troy_s | But they aren't narrow band
| |
20:01 | g3gg0 | exactly!
| |
20:01 | troy_s | The humps kill you
| |
20:01 | g3gg0 | thats why so many odd things happen with colorspace conversion
| |
20:01 | troy_s | And kill you on about sixteen fronts
| |
20:02 | troy_s | Hell yes. But again, a good chart (HCT / IT8) captured
| |
20:02 | troy_s | Can give you pretty damn good results if processed with care.
| |
20:02 | g3gg0 | e.g. check the images at http://magiclantern.fm/forum/index.php?topic=5197
| |
20:02 | g3gg0 | especially the blue LED
| |
20:03 | g3gg0 | using primaries and matrices only led the image developing tool to treating the blue LED as pink
| |
20:03 | g3gg0 | when color reconstruction jumps in, everything is a mess
| |
20:05 | g3gg0 | (iirc during the tests a1ex revealed that the matrices used in dcraw have some conceptual errors that lead to odd colors even when channels are not blown out)
| |
20:07 | g3gg0 | anyway. was a nice chat with you. thanks for all the information
| |
20:07 | g3gg0 | its time to get some rest :)
| |
20:10 | troy_s | g3gg0: I think we can do better
| |
20:10 | troy_s | g3gg0: Maybe you, me, and Alex can do a chat soon
| |
20:10 | troy_s | I am reasonably certain we can do better.
| |
20:11 | troy_s | g3gg0: great chatting with you too. It is lovely that there is a little channel where color nerds can hang
| |
20:11 | troy_s | (Was excellent at SIGGRAPH too, as I got to meet and chat with all of the gods like Mr. Selan, Mr. Gritz, and a crap load of other bright heads.)
| |
20:18 | sebix | left the channel | |
20:20 | intracube | alexML: I just looked at the UFRAW thread g3gg0 posted
| |
20:20 | intracube | awesome work!
| |
20:31 | troy_s | I have to wonder what ACR is doing.
| |
20:31 | troy_s | Could be a cheat in there.
| |
22:48 | se6astian | good night
| |
22:48 | se6astian | changed nick to: se6astian|away
| |
22:48 | g3gg0 | left the channel | |
23:45 | Bertl_oO | changed nick to: Bertl
| |
23:45 | Bertl | back now ...
| |
23:56 | wescotte | joined the channel |