| 23:07 | johnathan | Will do!
|
| 23:09 | RexOrCine | Good to see you engaging with the tough stuff. Well played.
|
| 23:24 | Bertl_oO | johnathan: there are several ways to figure out the various PLL parameters, like for example reading the MCMM/PLL documentation (most of the settings have specific ranges)
|
| 23:25 | Bertl_oO | if you don't want to do that, probably a rather simple way is to use Vivado and the IP configurator to have it figure them out for you, then use those values for your purpose
|
| 23:26 | Bertl_oO | note that you need two clocks for HDMI, a 'pixel' clock and a 'data' clock
|
| 00:17 | johnathan | left the channel |
| 01:24 | johnathan | joined the channel |
| 02:03 | Bertl_oO | off to bed now ... have a good one everyone!
|
| 02:03 | Bertl_oO | changed nick to: Bertl_zZ
|
| 02:19 | johnathan | left the channel |
| 03:02 | RexOrCine | changed nick to: RexOrCine|away
|
| 03:09 | siddhantsahu | joined the channel |
| 03:09 | siddhantsahu | left the channel |
| 03:41 | johnathan | joined the channel |
| 03:43 | johnathan | left the channel |
| 04:50 | BAndiT1983|away | changed nick to: BAndiT1983
|
| 06:25 | intrac | changed nick to: intracube
|
| 06:41 | BAndiT1983 | changed nick to: BAndiT1983|away
|
| 07:20 | BAndiT1983|away | changed nick to: BAndiT1983
|
| 07:28 | siddhantsahu | joined the channel |
| 08:04 | Bertl_zZ | changed nick to: Bertl
|
| 08:04 | Bertl | morning folks!
|
| 08:04 | siddhantsahu | Morning!
|
| 08:07 | Bertl | How's it going?
|
| 08:08 | siddhantsahu | Great. I am new here. Finding it the AXIOM project super interesting!
|
| 08:09 | siddhantsahu | Going through the source code of AXIOM Beta firmware and OpenCine
|
| 08:10 | siddhantsahu | How can this task T987 : Try running TensorFlow DeepLab-v3 inside AXIOM help?
|
| 08:10 | siddhantsahu | I mean how will semantic segmentation help for AXIOM?
|
| 08:11 | siddhantsahu | Is it possible to run such a compute expensive model in the AXIOM processor?
|
| 08:19 | Bertl | maybe, we don't know yet ... you need to talk to se6astian|away about this task
|
| 08:25 | siddhantsahu | left the channel |
| 08:56 | Bertl | off for now ... bbl
|
| 08:56 | Bertl | changed nick to: Bertl_oO
|
| 10:29 | comradekingu | joined the channel |
| 11:30 | siddhantsahu | joined the channel |
| 11:30 | siddhantsahu | left the channel |
| 11:38 | sebix | joined the channel |
| 12:12 | siddhantsahu | joined the channel |
| 12:54 | BAndiT1983 | changed nick to: BAndiT1983|away
|
| 13:15 | comradekingu | left the channel |
| 13:27 | comradekingu | joined the channel |
| 13:44 | se6astian|away | changed nick to: se6astian
|
| 13:47 | se6astian | changed nick to: se6astian|away
|
| 13:47 | se6astian|away | changed nick to: se6astian
|
| 13:58 | BAndiT1983|away | changed nick to: BAndiT1983
|
| 15:06 | se6astian | changed nick to: se6astian|away
|
| 15:32 | sebix | left the channel |
| 16:31 | johnathan | joined the channel |
| 16:39 | johnathan | I got the project setup in Vivado (using 2018.3 version), and it builds all the steps (synthesis and implementation) and I end up with a folder impl_1 with a bunch of files, including a top.bit (which I assume is the FPGA bitsteam), but the file size (4045663) is much different than both the cmv_hdmi3_30.bit (2150479) and cmv_hdmi3_60.bit (2160451). Is this to be expected? I didn't want to risk trying it in case s
|
| 16:39 | johnathan | omething could be wrong. At this point, I haven't made any changes to any of the sources, having only run the vivado.tcl under soc_main.
|
| 16:47 | vup2 | you should be able to find cmv_hdmi3.bit somewhere (in the working directory)
|
| 16:47 | se6astian|away | changed nick to: se6astian
|
| 16:47 | vup2 | that should have a similar size to the old bitstreams
|
| 16:49 | vup2 | xilinx supports some basic compressions for their bitstreams and top.bit is probably not compressed, the cmv_hdmi3.bit bitstream should be compressed
|
| 16:50 | vup2 | in theory both should work, but better ask Bertl_oO for details
|
| 16:54 | vup2 | johnathan: ^
|
| 17:14 | BAndiT1983 | changed nick to: BAndiT1983|away
|
| 17:27 | Dev2608 | joined the channel |
| 17:30 | johnathan | ok yep, I got it to work now, my mistake was creating a new project in the gui and running the script via the tcl console, for some reason it doesn't do the same thing. I started to read the Xilinx 7-series clocking docs, but wondering what is meant by IP Configurator? I see where it has generated the HDMI clocks but not sure how to go about changing it. I will poke around and do more research. As a side topic, I
|
| 17:30 | johnathan | have created a basic video4linux driver for the camera (using cmv_snap as a guide) that is working well and have it working as an input source to OpenCV on the camera.
|
| 17:31 | Adnan | joined the channel |
| 17:31 | Adnan | Hlo
|
| 17:36 | Adnan | left the channel |
| 17:40 | Dev2608 | Hey Bandit1983 | away : I was looking OCcore to understand how MLV files can be supported for openicne for Frameserver task , points i could find .1) The imageprovider.h provides two function (readbinaryfile and load ) , the load funciton is virtual which can be overridden for MLV, TIFF or DNG , The unorder_map will caontian pointer for loader function for each type of file . 2 ) Other files are implementation of demosaici
|
| 17:41 | Dev2608 | Wanted to ask few things ,some files are empty like OCcore.cpp and frameprovider.cpp , what is that for ! ,
|
| 17:43 | Dev2608 | (Demosaicing Algorithms)*
|
| 17:44 | Dev2608 | left the channel |
| 17:46 | BAndiT1983|away | changed nick to: BAndiT1983
|
| 17:48 | BAndiT1983 | Dev2608, some files are empty as the implementation is missing or they are from the prototype time and are obsolete, needs some review to check which ones are still required and which ones can be deleted
|
| 17:53 | vup2 | johnathan: yeah creating a new project using the gui should not be necessary
|
| 17:54 | vup2 | to tweak the hdmi clock you probably need to look in hdmi_pll.vhd
|
| 17:55 | vup2 | here the *_DIVIDE, *_PHASE and *_MULT parameters controll how {hdmi,tmds,data}_clock are generated from ref_clk_in
|
| 17:56 | vup2 | to get these values the easiest is probably to create a new vivado project, create a block design and add a clocking wizard block and configure it how you need it
|
| 17:56 | vup2 | the after generating the block design you can find the MMCME2_ADV config in one of the generated file
|
| 17:58 | vup2 | alternatively you could look at the the documentation regarding the mmcm2_adv at https://www.xilinx.com/products/intellectual-property/clocking_wizard.html
|
| 17:58 | vup2 | the v4l driver sounds interesting, is it published somewhere?
|
| 18:21 | Dev2608 | joined the channel |
| 18:21 | BAndiT1983 | johnathan, how is the opencv used there?
|
| 18:30 | Dev2608 | Ok , Thanks bandit1983 ! , by the way , I sent u email , explaining action plan for frameserver task , wanted to know your opinion
|
| 18:30 | BAndiT1983 | haven't you received it already? i've replied before leaving to work today
|
| 18:31 | BAndiT1983 | it was at 7:14 (german time)
|
| 18:31 | BAndiT1983 | off for a moment, will be back later
|
| 18:31 | BAndiT1983 | changed nick to: BAndiT1983|away
|
| 18:34 | Dev2608 | No , i haven't , :(
|
| 18:36 | Dev2608 | okay got it , thanks
|
| 18:36 | Dev2608 | it was in spam
|
| 18:36 | Dev2608 | folder
|
| 18:40 | siddhantsahu | left the channel |
| 18:41 | siddhantsahu | joined the channel |
| 18:46 | Dev2608 | left the channel |
| 20:02 | RexOrCine|away | changed nick to: RexOrCine
|
| 20:03 | BAndiT1983|away | changed nick to: BAndiT1983
|
| 20:03 | BAndiT1983 | Dev2608, now i'm a bit disappointed by that ;) :)
|
| 20:04 | slikdigit | joined the channel |
| 20:10 | johnathan | vup2: I'll clean up the v4l driver source and send a PR soon
|
| 20:10 | vup2 | cool
|
| 20:12 | johnathan | BAndiT1983: My application heavily relies on OpenCV to do various computer vision tasks, for now I have got working camera calibration (removing lens distortion etc) which is the first step in many CV-related workflows.
|
| 20:15 | vup2 | the opencv stuff certainly sounds interesting, keep us updated
|
| 20:15 | BAndiT1983 | johnathan, sounds very interesting, although it would be overkill for in-camera stuff, but for post-processing definitive necessary
|
| 20:26 | johnathan | yeah the camera SOC is not near fast enough to do real-time rectification but the algorithm is trivially simple and I hope to implement in FPGA once I learn more
|
| 20:38 | BAndiT1983 | Bertl_oO can tell probably more, in my opinion such stuff should be done after shooting, but maybe i'm wrong about it
|
| 20:41 | se6astian | off for now, good night
|
| 20:42 | se6astian | changed nick to: se6astian|away
|
| 20:42 | slikdigit | left the channel |
| 20:47 | Dev2608 | joined the channel |
| 20:51 | seku | joined the channel |
| 20:51 | Dev2608 | Bandit1983 : ohh sorry , :( , in case frameserver task won't be there( fingers crossed) , anything related to image processing stuff .
|
| 20:52 | Dev2608 | left the channel |
| 20:52 | Dev2608 | joined the channel |
| 20:52 | Dev2608 | Which I can look ?
|
| 20:57 | seku | evenings, ole backer here. Just being curious about the sensor interface board v2. are there chances it will be included with the Beta Compact, or will it be available seperately later?
|
| 21:01 | Dev2608 | left the channel |
| 21:03 | RexOrCine | seku: Possibly both (but unlikely), definitely latter. In terms of priorities it's not listed in relation to Compact shipping. Power Board V2 is.
|
| 21:04 | seku | RexOrCine, thanks! fully understandable.
|
| 21:04 | BAndiT1983 | Dev2608, at the moment we are in the planning phase, so no definitive result there yet, let us check on that if we know how many mentors are available, would like to have frame server task this year, but we lack some people to do everything, that's why my suggestion was that firmware should get more attention this time, as we are preparing a lot of fun stuff there
|
| 21:06 | seku | Last nagging : will the Compact first do only the ¨usual¨ resolutions, or will the whole sensor (opengate) be recordable? ... i am somewhat of an anamorphic nut, so usable 4:3 would be great.
|
| 21:06 | johnathan | BAndiT1983, yes I do agree its not useful to most casual shooters but for my application I need to do it real-time and want to avoid any extra hardware
|
| 21:07 | seku | (bandwidth issues, i presume?)
|
| 21:07 | slikdigit | joined the channel |
| 21:07 | Dev2608 | joined the channel |
| 21:09 | Dev2608 | Yes , I understand , np , I guess it will be clear before 6 Feb , I am not leaving oc but also try to understand about firmware
|
| 21:10 | BAndiT1983 | just ask if you need more info on some areas
|
| 21:10 | Dev2608 | Thanks :)
|
| 21:11 | BAndiT1983 | at the moment we are also considering to add axiom remote task, which includes extension of menu which is developed for PIC32 in C
|
| 21:11 | RexOrCine | seku: CMV12000 is max 4096x3072. Capturing full sensor output is a case of running "cmv_snap" command and the resulting image will be 4096x3072.
|
| 21:11 | Dev2608 | Yes , I will surely , let me first go through it by my own , update u soon
|
| 21:11 | slikdigit | left the channel |
| 21:11 | Dev2608 | left the channel |
| 21:14 | seku | RexOrCine, thanks :) once i get my compact, i will test if i can push the 100ish frames also on opengate :D
|
| 22:23 | aombk | joined the channel |
| 22:31 | Bertl_oO | johnathan: sounds interesting ... let me know when you need more information there ...
|