01:14 | balrog | left the channel | |
01:18 | balrog | joined the channel | |
11:23 | aombk | left the channel | |
11:35 | aombk | joined the channel | |
16:24 | illwieckz | left the channel | |
16:32 | illwieckz | joined the channel | |
18:54 | BogdanXOR | joined the channel | |
18:56 | BogdanXOR | hello Apertus, it's been a while
| |
18:58 | BogdanXOR | hey Bertl, I would need your help at your earliest convenience
| |
18:59 | BogdanXOR | long story short, trying to use (Atomos) Ninja V+ with the Axiom Beta, but the image I'm getting is in between Green (highlights) and Pink (darks) - rest assured it's not a false color situation
| |
19:00 | BogdanXOR | I have a feeling a config like the one we did (or rather you distilled) for SHOGUN is required
| |
19:00 | Bertl | hey BogdanXOR! :)
| |
19:01 | BogdanXOR | I could crunch at it, but I'm not sure where I could find the documentation for the `scn_reg`isters
| |
19:01 | BogdanXOR | and I'm ofc unsure what the root cause is, to know which registers to tweak (i.e. where to start off)
| |
19:01 | BogdanXOR | I've posted photos and all the tests I did before I came to the conclusion of doing `scn_reg` dance over to the Telegram channel
| |
19:03 | Bertl | well, what firmware and settings are you using?
| |
19:04 | Bertl | btw, green highlights and pink darks sounds a little bit like YCbCr (YUV) vs RGB
| |
19:16 | BogdanXOR | yes, YCbCr expectation on the Ninja side was the hypothesis brought forward by Jaro on Telegram
| |
19:16 | BogdanXOR | he suggested an axiom_mat4_conf invocation to try and fix it - I tried it but to no avail (the behaviour looks the same on the Ninja)
| |
19:17 | se6astian | BogdanXOR also asked about the specific scn_reg HDMI registers background: like what the hdisp_s corresepond to
| |
19:17 | se6astian | what values to cycles through and what parameters are more likely to vary
| |
19:17 | BogdanXOR | Switching to the Shogun I could see the image just fine, albeit more green than before
| |
19:18 | BogdanXOR | as for firmware, v2, as for settings the defaults (normal mode and SHOGUN generator init, as far as I could tell from the scripts)
| |
19:19 | Bertl | well, basically nothing you can adjust on the SCN registers (scan generator) will make the Ninja V+ switch from YCbCr to RGB
| |
19:20 | Bertl | (unless there are weird bugs in the HDMI chip they are using :)
| |
19:20 | BogdanXOR | and I assume getting the Axiom Beta to send YCbCr is more complex than just that matrix transform(?)
| |
19:21 | Bertl | the RGB to YCbCr conversion is actually a 'simple' matrix transform
| |
19:22 | BogdanXOR | yes - I'm just thinking there maybe needs some additional signaling (metadata(?)) to be sent as well
| |
19:22 | Bertl | so with the proper settings, it should be possible to do that, but usually there is another problem here
| |
19:23 | Bertl | YCbCr color spaces usually compress the croma information
| |
19:23 | BogdanXOR | yes, I'm aware
| |
19:23 | BogdanXOR | or I remember it
| |
19:23 | BogdanXOR | that's why 422 subsampling exists, I think
| |
19:23 | Bertl | assuming that the Ninja V+ can handle RGB data
| |
19:24 | Bertl | I would first look for a way to manually force the device to interpret the data as RGB instead of YCbCr
| |
19:24 | BogdanXOR | Ninja V+ can even do RAW, but ofc it's not immediately clear what handshake it expects
| |
19:26 | BogdanXOR | I did some searching (a bit), but there doesn't seem to be anything about it (or it's hidden because of the damn rgb parade or similar rgb functionality that the device has)
| |
19:29 | Bertl | so coming back to the firmware, you are on 2.x you say?
| |
19:34 | Bertl | try to change the following line in axiom_setup.sh
| |
19:34 | se6astian | yes BogdanXOR is on V2 and was instructed to alter the setup axiom_start script to try the different gen_init.sh parameters already
| |
19:34 | se6astian | SHOGUN, 1080p60, etc.
| |
19:34 | Bertl | (should be line 15)
| |
19:34 | BogdanXOR | yes - version the CLI welcome message reports is: 2f3a291c-dirty
| |
19:34 | Bertl | [ "$MODE" == "normal" ] && axiom_gen_init.sh SHOGUN
| |
19:34 | Bertl | change that to
| |
19:34 | Bertl | [ "$MODE" == "normal" ] && axiom_gen_init.sh 1080p60
| |
19:38 | BogdanXOR | so I tried that before by commenting lines 15:16 and uncommenting the lines just before that - and the behavior is the same, with an added vertical image flickering like on the old TVs that are not in tune with the display frequency
| |
19:39 | BogdanXOR | and I tried it now again precisely as you described it and it's the same
| |
19:40 | Bertl | okay, so the Ninja V+ has a similar (weird) chip as the SHOGUN
| |
19:40 | Bertl | (or HDMI IP :)
| |
19:40 | BogdanXOR | video of the behaviour: https://t.me/apertus/5350
| |
19:41 | BogdanXOR | it runs in the family, would doubt they give up their shennanigans :))
| |
19:42 | BogdanXOR | unsure if related, but now the CLI through minicom is not responding :))
| |
19:43 | BogdanXOR | same for screen
| |
19:43 | Bertl | not related unless you messed something up with the script :)
| |
19:44 | BogdanXOR | I've seen before some weird behaviour with vim on the system, so not surprised - unsure if it was due to my direct action, tend to think I didn't fuck up, but who knows
| |
19:45 | BogdanXOR | anywho, will do a nasty reboot and if not will do a reflash
| |
19:46 | se6astian | btw BogdanXORit would be great if we add/document the findings on the wiki
| |
19:47 | se6astian | do you have a wiki account already?
| |
19:47 | se6astian | as we were flooded with spam accounts and mediawiki countermeasures were/are severly lacking we had to disable user registration and create accounts manually
| |
19:51 | BogdanXOR | don't remember if I do, but will check
| |
19:53 | BogdanXOR | in any case, don't worry, will go through Telegram and IRC and put relevant info and conclusions in wiki
| |
19:53 | BogdanXOR | right now it almost feel the only thing we can conclude is that NINJA V+ sucks :))
| |
19:54 | BogdanXOR | *feels
| |
19:54 | se6astian | :P
| |
19:54 | Bertl | well, it is certainly a little weird
| |
19:55 | BogdanXOR | maybe I should re-check my NINJA with another camera, just to make sure
| |
20:01 | Bertl | can't hurt to do that
| |
20:01 | BogdanXOR | Ninja V+ with officially-supported camera (Fuji X-H2s): https://t.me/apertus/5351
| |
20:01 | BogdanXOR | i.e. works fine
| |
20:02 | BogdanXOR | all colors are reproducing well, no pink, no green, not even a cast of it
| |
20:06 | Bertl | well, send one over and I promise I'll take a look :)
| |
20:07 | BogdanXOR | can send you mine. do you think you need the camera as well?
| |
20:07 | Bertl | nah, I have an AXIOM Beta here for testing, so no problem with that
| |
20:10 | BogdanXOR | at the same address as in 2018?
| |
20:10 | Bertl | probably, let's double check in pm
|