Night timelapse (was Re: considerations for SX260 external power supply for time lapse situations) - page 3 - Script Writing - CHDK Forum

Night timelapse (was Re: considerations for SX260 external power supply for time lapse situations)

  • 52 Replies
  • 23493 Views
*

Offline reyalp

  • ******
  • 14080
Advertisements
So we set the Overexp protection to 10. Or perhaps higher?
Yeah, that seems right. Over exposure protection does have some effect before the threshold is reached, so you might want to go a little higher if you want to completely ignore the moon. The actual formula is a bit complicated but it goes by the square of the fraction of the threshold value reached.

For a clear night with just the moon, setting it much higher should have no effect, but will affect what happens for sunset, sunrise and possible clouds.

Again, I'd recommend doing some tests before you do a full night, just a few minutes with the moon in the scene should be enough to see if it's working as you expect.
Don't forget what the H stands for.

*

Offline udo

  • ***
  • 123
I highly recommend using Bv Ev shift to get a more natural effect. This will lower the target exposure at night, but this doesn't really matter because the script will be on the limits once it gets really dark.
As I am slowly preparing for a small holiday I updated the S110 to a recent 1.4.1 chdk and copied rawopint 0.20 on the card.
I can now calculate the ui_exp_over_thresh_frac value I need (bigger than circa 10) but, as I would like to implement your recommendation, the bv options ui_bv_ev_shift_pct and ui_bv_ev_shift_base_e puzzle me a bit, even after reading the readme.
The runs I did thus far in daylight did not use this feature and looked OK; for darker scenes, sunset, sunrise, etc I would like to preset some sensible values before I run some experiments.
Does the bv feature interfere with daylight scenes? Or can I tweak it so that it only works in darker scenes?

*

Offline reyalp

  • ******
  • 14080
I would like to implement your recommendation, the bv options ui_bv_ev_shift_pct and ui_bv_ev_shift_base_e puzzle me a bit, even after reading the readme.
Without bv/ev shift enabled, the script tries to expose every shot the same: So a shot in full daylight will be the same brightness as one after sunset (Tv and ISO limits permitting)

With bv/ev shift, the script adjusts the exposure based on the actual brightness of the scene. So a shot in full daylight will be brighter than one after sunset.

The "Bv" refers (roughly) to the APEX "brightness value". Without getting into the technical details (see http://dougkerr.net/Pumpkin/articles/APEX.pdf for that) Bv is a way of representing the real amount of light in terms of the same "stops" that you use with exposure settings like shutter speed. So if you use a shutter speed of 1/64th for a Bv of X, then a 1/128th would give the same exposure at a Bv of X+1.

The shift % option controls how much the image exposure changes per actual change in brightness. If shift is 50%, then a change of 1 in Bv will cause a 1/2 stop change in exposure. If it's 100%, then the script would essentially have a fixed exposure. 0% means there is no change, effectively turning the option off. I would suggest starting with something like 10%-20%

The "Base" value sets the Bv value where there will be zero shift, meaning the exposure at this light level will be normal, something like what Canon auto exposure would use. Generally, you want this to be somewhere near the brightest the scene will be. For daylight, something around 12 is probably reasonable. For sunsets, or anything you start in daytime, you can use "initial". For sunrise, or if you are starting the script under clouds that might clear, you should pick a specific value.

Quote
Does the bv feature interfere with daylight scenes?
It shouldn't, if the base value is set to a reasonable value. Even if it's wrong, over-exposure limiting should keep things from getting too bad.
You may notice some difference, but IMO the bv/ev shift behavior generally looks more realistic. Without it, shadowed areas often get brighter when a cloud passes in front of the sun. Because the overall scene has gotten darker, exposure is increased, but contrast is reduced by the cloud. With bv/ev shift, this effect is reduced.
Quote
Or can I tweak it so that it only works in darker scenes?
There is no way to have it enabled just part of the time.
Don't forget what the H stands for.

*

Offline udo

  • ***
  • 123
Thanks for the explanation.
I adjusted the settings accordingly: 10% for shift % and 12 for teh base value.
I will try to run a test this weekend.


*

Offline reyalp

  • ******
  • 14080
To get an idea of the range of Bv, you can look at the first plot in this post https://chdk.setepontos.com/index.php?topic=12766.40

The values are multiplied by 96, so you can see in MarkB's sunrise Bv went from ~ -4.5 to 9.5

Bv / ev shift reduced the initial exposure by ~2 stops.

Note that the meter limits (meter low thresh, meter low limit etc) also set the limits on how far the exposure will be changed by bv/ev shift.
Don't forget what the H stands for.

*

Offline udo

  • ***
  • 123
Overexp protection at 10 or so appears to cause dark exposures when doing a sunset timelampse, so we increased to 3000.

Also during night shots the exposure does not go much over 1.37 seconds; what causes this?
Still the over exposure? (this is in an urbanish area)
I attached the csv.
« Last Edit: 27 / March / 2016, 05:56:01 by udo »

*

Offline reyalp

  • ******
  • 14080
Overexp protection at 10 or so appears to cause dark exposures when doing a sunset timelampse, so we increased to 3000.
The numbers are in parts per 100,000, so 10 means 0.01%. That's very low.

You can control how dark over exposure protection can make the screen using the meter low and under limits (explained in https://chdk.setepontos.com/index.php?topic=12766.50)

Quote
Also during night shots the exposure does not go much over 1.37 seconds; what causes this?
Still the over exposure? (this is in an urbanish area)
I attached the csv.
If you look in the CSV, you can see the over_weight column is 200 for a lot of the time, because over_frac is 0.04-0.05. So yes, your over-exposure limit is keeping the exposure short.
Don't forget what the H stands for.

*

Offline udo

  • ***
  • 123
Another try with:
#ui_exp_over_thresh_frac=4000 "Overexp thresh x/100k (0=Off)" [0 100000]

Exposure still doesn't go up very much.
Also over_tresh_flac shows up as 40000 in the csv? (zero too many)

Any idea why these happen?
Max exposure time is lower than before; how can I make the ISO and exposure time go up more?
(in a less urban area would this be already enough to make ISO and exposure go up to the max?)
« Last Edit: 31 / March / 2016, 11:11:19 by udo »


*

Offline reyalp

  • ******
  • 14080
Also over_tresh_flac shows up as 40000 in the csv? (zero too many)

Any idea why these happen?
This is normal. The UI shows parts per 100,000. The script works internally with parts per 1,000,000, which is what is recorded in the log.

I will try to look at the log when I have time later, but I encourage you to look at it yourself and try to understand what limits were driving the exposure, as I explained in the post linked previously.
Don't forget what the H stands for.

*

Offline udo

  • ***
  • 123
I will try to look at the log when I have time later, but I encourage you to look at it yourself and try to understand what limits were driving the exposure, as I explained in the post linked previously.

Well, w.r.t. trying to understand:

The readme says:
==meter, meter96==
Average value of the metered area: meter is the raw sensor value, *96 gives this as an APEX96 value where the target exposure is 0. Negative values indicate an underexposure, positive is overexposure.

But I see logging that has a positive meter value with a negative meter96 value; what does that mean?
Why does exposure time get shorter when I increase the allowable overexposed fraction?

Your advice is very welcome here!
« Last Edit: 01 / April / 2016, 00:46:19 by udo »

 

Related Topics