Open 99991 opened 2 years ago
Sorry for the trouble and thanks for the thorough details here! We're quite behind right now so just a candid heads up that it might be a bit before we can address this.
(And FWIW I personally quite want to see this application happen, will be super cool.)
@Luxonis-Brandon @99991 , I too am interested in the issue of having exposure settings propagated more quickly. Any luck/progress on this lately?
@99991 @rajsinghmn
I think this can now be done by using "direct" set of commands, to manually control those parameters. These should skip over such delays, which become much more prominent with low framerates.
Let me know how this works for you
Thanks for the suggestion, @themarpe! Sorry, I'm fairly new to the luxonis platform. By "direct", do you mean using an alternative to XlinkIn for the exposure changes? Do you have a minimal example of how to directly control exposure?
@rajsinghmn
Its on branch: camera_controls_misc
cam[c].initialControl.setMisc("manual-exposure-handling", "fast")
However, this only works for IMX378/IMX582
for now, so might not apply to your case as is...
There a two issues with the exposure settings of the OAK-FCC-3P in conjunction with the Arducam HQ. Those issues hinder applications in astrophotography where very long exposure (several minutes to hours) are desirable. See for example this StackOverflow answer with a 42 minute long exposure using the RPI HQ camera.
For issue 1, it would be nice if the exposure time limit could be removed entirely.
For issue 2, it would be nice if the exposure settings could be applied more quickly -- perhaps after every frame? 600 ms should be more than enough time to transmit that information.
Here is a plot of exposure and corresponding brightness change where you can see both issues. For the first 5 seconds, the image brightness does not change because the exposure settings are slow to propagate. For the next 30ish seconds, the brightness does not change although the exposure is changed. After setting the exposure from 700 ms to 600 ms, the brightness finally changes, but the change again takes roughly 5 seconds to be applied.
Here is the code I used to measure those values: