search

artyom-beilis / OpenLiveStacker

Live Stacker Project - C++ backend and frontend
GNU General Public License v3.0
30 stars 3 forks source link

feature request: indilib support #95

Open lilo-chita opened 2 weeks ago

lilo-chita commented 2 weeks ago

Any chances to add remote camera support via indi interface?

artyom-beilis commented 2 weeks ago

I want to add indi camera support for a while but I actually failed to run ekos locally with camera and have main issues with it.

I probably need to upgrade to later Ubuntu version (I still work with 18.04) to run ekos properly. This is something I need to learn to work with. Indi documentation is far from being good.

Also what camera do you have? Maybe it easier to add support directly :-)

lilo-chita commented 2 weeks ago

Atm it's Canon DSLR. But it's remote setup. RaspPi as indi server with all hw connected to it. And laptop to control, process and view the results.

artyom-beilis commented 2 weeks ago

You can connect OLS directly to canon and access remotely via http... But I understand that unified method is better. Also for DSLR just watch directory should work easily. Framerate is anyway poor.

Astro-Otter-Space commented 2 weeks ago

I think i can be interested too. I'm trying to work in remote: raspberry pi with indi and indi-drivers and desktop with Ekos connected to raspberry pi

lilo-chita commented 2 weeks ago

You can connect OLS directly to canon and access remotely via http

Sure, but I'm afraid that my pi 3b+ is very limited in compute power. Anyway will try and see.

artyom-beilis commented 2 weeks ago

I see so you just run indi server on pi and access it via ekos remotly?

lilo-chita commented 2 weeks ago

indi server on pi and CCDciel as client.

lilo-chita commented 2 weeks ago

Not so bad actually. 0.5sec for each frame on raspberry pi 3b+ raspbian bookworm OS for M44 simulation.

lilo-chita commented 2 weeks ago

Opppss Actually it's a bit longer

2024-06-25 22:34:06; stacker, info: Stacking took 510.945 ms (stacker_processor.cpp:117) 2024-06-25 22:34:06; stacker, info: WB 1.59693,1,1.18148 (post_processor.h:112) 2024-06-25 22:34:06; stacker, info: Scale 2.23362 offset=0.0400782 relative cut =0.0895197 (post_processor.h:535) 2024-06-25 22:34:06; stacker, info: Stretch mean 0.054652-> gamma=2.09679 (post_processor.h:436) 2024-06-25 22:34:06; stacker, info: Real time video frame 10 arrived (video_stream.h:119) 2024-06-25 22:34:06; stacker, info: Preprocessing took 116.694ms (pre_processor.cpp:43) 2024-06-25 22:34:06; stacker, info: apply wb=84.3208ms calc stretch60.247ms stretch=279.019ms (post_processor.h:452) 2024-06-25 22:34:06; stacker, info: post processing took 698.054ms. proccessing 471.335 ms, jpeg took=226.719 (post_processor.cpp:235) 2024-06-25 22:34:06; stacker, info: Stacked video frame 3 arrived (video_stream.h:119) 2024-06-25 22:34:06; stacker, info: Registration at 2:[2, 3]

artyom-beilis commented 2 weeks ago

You are actually right it is ~0.5s if you have multi-core PI. Since pre-processing, stacking and post processing are running in parallel threads. stacking is bottle neck usually. Post processing may take longer but it can skip frames so it wouldn't overload.

Also note simulation frames quite small so for large DSLR frame it can take much more.