#timvideos

ysionneau: I assume you are asleep, but I'm pretty sure if those i2c-tools above can read your virtual EEPROM, then the fx2 will be able too

mithro: when versions get promoted to testing and stable, how do we know what the source is?

CarlFK: the git ID is encoded in the firmware and in the firmware name

oh cool

CarlFK: see https://github.com/timvideos/HDMI2USB-firmware-...

Title: HDMI2USB-firmware-prebuilt/archive at master · timvideos/HDMI2USB-firmware-prebuilt · GitHub (at github.com)

CarlFK: stable / testing are really just symlinks

mithro: how can I help to fix/progress this issue talking to the hdmi2usb?

you said it was an issue with the driver?

any workarounds that come to mind?

xfxf: The issue is with the exart-uart-driver USB serial adapter

xfxf: so, unless you think you can work on that driver there isn't much you can help with that

i'm no kernel hacker, that's for sure

xfxf: the code I wrote tries to work around the pyseral / exart-uart-driver weirdness

is this something we can report upstream?

xfxf: there is no "upstream"

xfxf: upstream is currently shenki :-P

oh, right, you guys wrote the driver

i assumed it was in the kernel

so i should prod shenki? :P

xfxf: well, no we didn't - we make the old crappy obsolete driver work again

right, but it's still evidently crappy

xfxf: correct

like i'm happy hacking away at application level stuff hence my interest in getting this basic switcher more fleshed out as basic capture software

to use 'for now' until voctomix or gst-switch is in a usable state

and i'm keen to get my hands dirty with python again, been doing too much php/rails lately

oh, while i remember too - i'm keen on solving the issue with gstreamer only capturing at 15fps

you wanted to know what the encoer was running at?

i can test that now

xfxf: have tried commenting the line I sent you the other day for the 15fps?

Title: HDMI2USB-misoc-firmware/encoder.c at master · timvideos/HDMI2USB-misoc-firmware · GitHub (at github.com)

_florent_: ah, sorry, i completely forgot about that. will do! i haven't been home, am now

ok thanks

hmm

encoder says

encoder: 1280x720 @ 30fps (19Mbps) from input0 (q: 85)

captured file when poked with avprobe

Stream #0.0(eng): Video: mjpeg, yuvj422p, 1280x720 [PAR 1:1 DAR 16:9], 15 fps, 15 tbr, 1k tbn (default)

capture command:

gst-launch-1.0 v4l2src device=/dev/video1 ! image/jpeg,width=1280,height=720 ! matroskamux name=mux alsasrc device='hw:0,0' ! audio/x-raw,channels=2,rate=48000 ! audioconvert ! vorbisenc ! queue ! mux. mux. ! queue max-size-bytes=100000000 max-size-time=0 ! filesink location=hdmi2usb_recording_`date +%s`.mkv

xfxf: does the file actually have 15fps or does it have 30fps in it?

totem is claiming 15fps, it looks less juddery than the previous content captured

i'd normally rely on what the file reports though and not my perception, not really scientific

i assume you're implying the headers might say 15fps but the content is 30?

xfxf: break the file apart and see how many frames per second are in the file?

what tools would i use to splice apart the files and inspect the frames?

well, mplayer with -fps 15 displays slower than -fps 30

-fps 30 looks the same as if played normally

so perhaps it is 30 fps now

aah

okay

so

the content i captured the other day

with the gstreamer line above

actually is 30fps i think

just because the headers say 15fps then my ffmpeg conversion just converted it to a 15fps file

as it looked at the fps of the original file

i'm playing back the mjpeg recording now, looks fine

although the player is claiming it's 15fps

why is that happening?

so _florent_ commenting out that line doesn't make any difference

ok, so the issue is probably in the UVC header inserted by the FX2 (fixed to 15fps)

also wierd, so each mjpeg frame can contain data about the playback rate? i had always thought that was the responsibility of the container

right

kind of

the source is likely to be the UVC header, but it goes USB->V4L2->gstreamer

Title: HDMI2USB-misoc-firmware/descriptors_hdmi2usb_alt.c at fx2-refactor · mithro/HDMI2USB-misoc-firmware · GitHub (at github.com)

Probably either line 316 or 331

ah, i assume they're defaults? why not default to 30?

xfxf: because that is what the old code did

This is the previous version -> https://github.com/mithro/HDMI2USB-misoc-firmwa...

Title: HDMI2USB-misoc-firmware/descriptors_hdmi2usb.a51 at fx2-refactor · mithro/HDMI2USB-misoc-firmware · GitHub (at github.com)

I spent a whole bunch of time converting that a51 into that .c file so I could understand what was going on

right

Title: HDMI2USB-misoc-firmware/uvclocal.h at fx2-refactor · mithro/HDMI2USB-misoc-firmware · GitHub (at github.com)

so i see FRAME_INTERVAL_30FPS is defined there

should i just change 15 to 30 in that c file?

xfxf: Yes, I just added that

xfxf: I have no idea if that .c file works yet

oh, right, you're still using the a51

xfxf: this is my fx2-refactor branch

right

which is trying to get the FX2 firmware into a better state were we can improve it

and get it working reliably under Linux, Windows and Mac

that explains why trying to egrep for FRAME_INTERVAL locally found nothing

neat, sounds good

so i should hang tight until you've done that?

xfxf: well, it seems like you might have a work-around for now?

xfxf: btw did you try my changes to your code?

sort of, i know what the issue is now, i'll just need to figure out how to get to create a 30fps file (it's only the mjpeg dumped off the device that is playing back properly)

i did a basic straight capture + manual conversion the other day just to prove this works to myself

my intention was to mangle up the gstreamer pipeline more so it's transcoding to disk as part of the capture

not sure if the 15fps wierdness might complicate that

I'm sure thaytan and help you figure out a gstreamer pipeline which overrides the information that v4l2 is telling it

cool, can you do that to the above so the mjpeg saved out does have the right header info?

above line i pasted, that is

i assume if that's done early enough in the pipeline then i shouldn't have an issue adding extra stuff after it

pretty much as soon as i have some basic switching working

i'll start recording stuff with it

even if it requires some wierd workarounds

as long as i get usable video out at the other end

xfxf: I don't know how

thaytan: ping :)

hmm?

see above - the hdmi2usb device with the gstreamer line i pasted seems to be writing out 30fps data with a 15fps header (i assume)

need a way to force it to 30fps

capssetter

gst-launch-1.0 v4l2src device=/dev/video1 ! image/jpeg,width=1280,height=720 ! capssetter caps='image/jpeg,width=1280,height=720,framerate=30/1' ! ....

like that

right, ta, was just looking up the syntax

will give it a go

neat, works, thanks again!

mithro: another thing that happens too is my canon HDMI camera when plugged into the HDMI2USB doesn't actually feed it any data

you have to turn it off and turn it back on again then HDMI2USB will see it

given you force the output resolution on the camera my assumption is it's possibly not handshaking (as much, at all? i don't know)

may be common with other cameras or scalers, not sure

(this is my XA20 camera, not the LCA ones)

morning

xfxf: the camera or the HDMI2USB?

the camera has to be power cycled

this throws one out in a normal setup routine because you plug everything in, turn it on, and the camera doesn't work

xfxf: have you tried resetting the video_mode with the camera plugged in?

i believe i did, but i'm happy to try that now

Please do

I guess that to use the tools you linked mithro I need to connect somehow my i2c pins to my computer motherboard?

since those tools run on host

ysionneau: I was going to ask if you had a beaglebone or buspirate or something?

xfxf: I was wondering if it was a form of https://github.com/timvideos/HDMI2USB-misoc-fir... or not?

Title: Support resetting an video input without resetting the output · Issue #80 · timvideos/HDMI2USB-misoc-firmware · GitHub (at github.com)

hmmm I don't, but I think I might have some compatible board

I think there are alternative firmware for other boards

oh, I have a minnowboard Max

let's hope it has i2c

ysionneau: if you had a beaglebone or similar board, Rohit might actually be able to help

he's done a whole bunch of i2c work for talking to the VGA chipset

yay i2c

mithro: nope, doesn't work