KAP & UAV Exposure Control Intervalometer Script

waterwingz

milobg
So first I try the values you suggest me…. I put camera outside my window to taking photo to bright an dark areas:

Can you enable logging and repeat this test? Use the last parameter in the script menu that says Logging and set it to Both.

Rerun your test and then attach the resulting log file from your SD card ( KAP.log ) to a response here?

milobg

Hi Like you told me yesterday, here is the KAP.log file....

Is it corect???

Milobg

KAP.log

11.46KB

waterwingz

milobg
Hi Like you told me yesterday, here is the KAP.log file....

Thanks. Only the last seven shots in the log are setup as you described in your previous posts. The only shot there not at 1/1250 is IMG_7351.JPG. It's shooting at 1/800 because the ISO is maxed out at 1600. The script seems to be performing exactly as designed.

milobg

ok I understand... script works well,

now if I want to lock exposure with your script...
What do you suggest to me??

Are these the right values??

Av Mode : both
Target Tv : 1/1250
Tv Max : 1/1250
Lowest Av : 2.8
Target Av : 2.8
Highest Av : 2.8
ISO Min : 100
ISO Max1 : 100
ISO Max2 : 100

Let me know, if you can....

thank milobg

waterwingz

milobg
now if I want to lock exposure with your script...
What do you suggest to me??

To get the correct exposure, the script will lower the shutter speed once the ISO is at maximum and the f-stop is fully opened. Otherwise you will have an underexposed picture. You could try setting ISO Max2 to 3200 to help with this.

Otherwise, the script will need to be modified or you will need a different script.

Wreck

Thanks for creating this script Waterwingz.

I've just used it for the first time on an sd960is. I have the target shutter speed set to 1/1600, and most images are being shot correctly at that speed, however I've come across a couple which were shot at a lower shutter speed and iso, and are blurry because of it.

In my kap.log for some of those files, this is the output:
2013Dec06 15:02:52 26) IMG_1022.JPG
2013Dec06 15:02:52 meter : Tv:1/640 Av:2.8 Sv:80 827:827
2013Dec06 15:02:52 actual: Tv:1/1600 Av:- Sv:160
2013Dec06 15:02:59 27) IMG_1023.JPG
2013Dec06 15:02:59 meter : Tv:1/320 Av:5.6 Sv:80 910:910
2013Dec06 15:02:59 actual: Tv:1/1600 Av:- Sv:400
2013Dec06 15:02:59 AvMin:2.8 NDF:none foc:infinity
2013Dec06 15:03:07 2😎 IMG_1024.JPG
2013Dec06 15:03:07 meter : Tv:1/125 Av:5.6 Sv:n/a 838:838
2013Dec06 15:03:07 actual: Tv:1/1600 Av:- Sv:640
2013Dec06 15:03:07 AvMin:2.8 NDF:none foc:infinity
2013Dec06 15:03:13 29) IMG_1025.JPG
2013Dec06 15:03:14 meter : Tv:1/500 Av:2.8 Sv:80 780:780
2013Dec06 15:03:14 actual: Tv:1/1600 Av:- Sv:250
2013Dec06 15:03:14 AvMin:2.8 NDF:none foc:infinity

When I look at the exif for these files, they don't match up to the log:
IMG_1022.JPG is 1/400, f2.8, iso 50.
IMG_1023.JPG is 1/200, f2.8, iso 50
IMG_1024.JPG is 1/1600 iso 640 (as it should be)
IMG_1025.JPG is 1/320, f2.8, iso 50.

Any idea what could be going on?

waterwingz

Wreck
I have the target shutter speed set to 1/1600, and most images are being shot correctly at that speed, however I've come across a couple which were shot at a lower shutter speed and iso, and are blurry because of it.

This is most curious. Would you say the exposure was correct (i.e. not too dark or too bright) even though the shutter speed was not what you required?

Thanks for posting the log entries. I color coded a couple of things that pop out right away and commented below.

In my kap.log for some of those files, this is the output:
2013Dec06 15:02:52 26) IMG_1022.JPG
2013Dec06 15:02:52 meter : Tv:1/640 Av:2.8 Sv:80 827:827
2013Dec06 15:02:52 actual: Tv:1/1600 Av:- Sv:160
2013Dec06 15:02:59 27) IMG_1023.JPG
2013Dec06 15:02:59 meter : Tv:1/320 Av:5.6 Sv:80 910:910
2013Dec06 15:02:59 actual: Tv:1/1600 Av:- Sv:400
2013Dec06 15:02:59 AvMin:2.8 NDF:none foc:infinity
2013Dec06 15:03:07 2😎 IMG_1024.JPG
2013Dec06 15:03:07 meter : Tv:1/125 Av:5.6 Sv:n/a 838:838
2013Dec06 15:03:07 actual: Tv:1/1600 Av:- Sv:640
2013Dec06 15:03:07 AvMin:2.8 NDF:none foc:infinity
2013Dec06 15:03:13 29) IMG_1025.JPG
2013Dec06 15:03:14 meter : Tv:1/500 Av:2.8 Sv:80 780:780
2013Dec06 15:03:14 actual: Tv:1/1600 Av:- Sv:250
2013Dec06 15:03:14 AvMin:2.8 NDF:none foc:infinity

First of all, the script did try to take every shot at 1/1600. You can see that in the log lines that start with the word "actual".

Secondly, the SD960 has a fixed aperture. Yet the camera is trying to use f2.8 for some shots and f5.6 for other shots. One of two things must be happening here. Either the zoom position is changing or the camera's Neutral Density filter is being engaged by the Canon firmware. Are you changing the zoom setting some how between shots? If not, it must be the ND filter, and that will cause unwanted exposure variations. (Notice that the Canon firmware is only trying to use the f5.6 setting in brighter scenes). Set the AV Mode in the script parameters to "NdFlter" and the script will manage that for you. (Its not clear how this would cause the shutter override not to work though .. )

Finally, when taking image IMG_1024.JPG, the camera's light meter system tried to select an ISO setting below the range expected by the script. That's quite curious but should not have caused any problems.

When I look at the exif for these files, they don't match up to the log:
IMG_1022.JPG is 1/400, f2.8, iso 50.
IMG_1023.JPG is 1/200, f2.8, iso 50
IMG_1024.JPG is 1/1600 iso 640 (as it should be)
IMG_1025.JPG is 1/320, f2.8, iso 50.

Do the exif values for other images match the log entries made by the script? ( Other than ISO80 being reported as ISO50 ).

tl;dr : try setting the AV Mode in the script parameters to "NdFlter"

Wreck

Yeah the exposure on those images which had the low shutter speed and iso was more or less correct, so it must be that the ND filter was engaging. It seems that the images where this happened generally had more of a bright feature in the frame than others that were more of a uniform tone.

I've just had a more thorough look through the exif data on the rest of the images, and there are quite a few images that don't have the correct shutter speed/iso, but they were sharp so I didn't look at them initially.

2013Dec06 15:01:28 14) IMG_1010.JPG
2013Dec06 15:01:28 meter : Tv:1/200 Av:5.6 Sv:80 832:832
2013Dec06 15:01:29 actual: Tv:1/1600 Av:- Sv:640

This one ended up shooting 1/125 at f5.6 iso 50. I don't think there's any way that the zoom would have been engaging since the FOV of all of the images is the same.

Wreck

Just did a test with the AV Mode set to NDFilter, and pretty much every shot is way overexposed, generally with a shutter speed of 1-250-1/400 at iso 50 (target shutter speed was set to 1/1250)

2013Dec08 11:17:17 14) IMG_1114.JPG
2013Dec08 11:17:17 meter : Tv:1/320 Av:5.6 Sv:80 923:923
2013Dec08 11:17:17 actual: Tv:1/1250 Av:- Sv:250
2013Dec08 11:17:17 AvMin:2.8 NDF:NDout foc:infinity
2013Dec08 11:17:23 15) IMG_1115.JPG
2013Dec08 11:17:24 meter : Tv:1/400 Av:5.6 Sv:80 927:927
2013Dec08 11:17:24 actual: Tv:1/1250 Av:- Sv:250
2013Dec08 11:17:24 AvMin:2.8 NDF:NDout foc:infinity
2013Dec08 11:17:30 16) IMG_1116.JPG
2013Dec08 11:17:30 meter : Tv:1/400 Av:5.6 Sv:80 958:958
2013Dec08 11:17:31 actual: Tv:1/1250 Av:- Sv:200
2013Dec08 11:17:31 AvMin:2.8 NDF:NDout foc:infinity
2013Dec08 11:17:37 17) IMG_1117.JPG
2013Dec08 11:17:37 meter : Tv:1/500 Av:5.6 Sv:80 973:973
2013Dec08 11:17:37 actual: Tv:1/1250 Av:- Sv:200
2013Dec08 11:17:38 AvMin:2.8 NDF:NDout foc:infinity
2013Dec08 11:17:43 1😎 IMG_1118.JPG
2013Dec08 11:17:44 meter : Tv:1/640 Av:5.6 Sv:80 1039:1039
2013Dec08 11:17:44 actual: Tv:1/1250 Av:- Sv:120
2013Dec08 11:17:44 AvMin:2.8 NDF:NDout foc:infinity
2013Dec08 11:17:50 19) IMG_1119.JPG
2013Dec08 11:17:50 meter : Tv:1/250 Av:5.6 Sv:n/a 947:947
2013Dec08 11:17:50 actual: Tv:1/1250 Av:- Sv:200
2013Dec08 11:17:51 AvMin:2.8 NDF:NDout foc:infinity

From Exif:
IMG_1114.JPG - 1/200, f5.6, iso 50
IMG_1115.JPG - 1/250 f5.6 iso 50
IMG_1116.JPG - 1/250 f5.6 iso 50
IMG_1117.JPG - 1/320 f5.6 iso 50
IMG_1118.JPG - 1/500 f5.6 iso 50
IMG_1119.JPG - 1/1250 f5.6 iso 200

waterwingz

Wreck
Just did a test with the AV Mode set to NDFilter, and pretty much every shot is way overexposed, generally with a shutter speed of 1-250-1/400 at iso 50 (target shutter speed was set to 1/1250)

Looks like you took those shots in bright light? In every one of those shots, the camera appears to be inserting the ND filter and taking an exposure reading that way. This is different behavior from what I've seen on my cameras - I will have to study it a bit.

Meanwhile, I think you will find that the shutter speed used is actually 1/1250. The exif values are not correct.

Wreck

How do you tell if the exif values are correct or not? I was just shooting straight down onto grass (the same area and conditions as yesterday) and it's a reasonably sunny day today. The overexposed images that were produced make me think that it did actually use a slower shutter speed, but I'm not much of an expert with chdk cameras.

waterwingz

Wreck
How do you tell if the exif values are correct or not? I was just shooting straight down onto grass (the same area and conditions as yesterday) and it's a reasonably sunny day today. The overexposed images that were produced make me think that it did actually use a slower shutter speed, but I'm not much of an expert with chdk cameras.

The script log reports that CHDK is trying to override the shutter to 1/1250. I would be very surprised if that does not happen. On my A1200, once you ask for a shutter speed above the normal camera upper range, the exif data does not save the higher shutter speed for some reason even though the image is taken at the faster speed.

Meanwhile, I think that the overexposed images are because the script is assuming the aperture is set to f5.6 and does not realizing its actually at f2.8 with the ND filter engaged. So when it does its exposure calculation it uses f5.6 and then shoots with the ND filter out! This causes it to be about 3 f-stops over exposed.

Clearly a bug - I just need to do some testing to figure out if its an anomaly of your SD960 or normal behavior for most cameras.

Wreck

These overexposed shots all have motion blur too, so I still think the shutter speed is low.

waterwingz

Wreck
These overexposed shots all have motion blur too, so I still think the shutter speed is low.

If that's true, then it would mean the CHDK port for the SD960 is seriously broken. Nothing much I can do about that with a script.

waterwingz

Okay, if you are up for a little more testing, this should handle the ND filter issue properly in bright light (tested taking pictures of my desk lamp's light bulb).

https://app.box.com/s/hslxxa9tiotgwxm35jhw

Let me know if it fixes the overexposure issue when you have the AV Mode in the script parameters to "NdFlter" ?

Wreck

Thanks for that. I'll give it a go and let you know how I get on.

srsa_4c

waterwingz

Wreck
These overexposed shots all have motion blur too, so I still think the shutter speed is low.

If that's true, then it would mean the CHDK port for the SD960 is seriously broken. Nothing much I can do about that with a script.

Bug reports are welcome (preferably with a description on how to reproduce).
I have applied the 'short shutter press' override fix in changesets 3264, 3265, but I'm not sure if that will have any influence on this script.

waterwingz

srsa_4c
Bug reports are welcome (preferably with a description on how to reproduce).

The script does not correctly handle the situation where the camera wants to insert the ND filter and reports using a smaller f-stop setting as a result. It assumes the reported Av value is the actual aperture f-stop setting and does its exposure calculations based on that. When the actual shot occurs, it ends up using calculations based on the ND filter being inserted even though it does not insert that filter.

Once that gets sorted out, we can worry about possible Tv override bugs. The fix for cameras without an adjustable aperture is to use get_prop(props.MIN_AV) to determine the actual Av96 value of the aperture rather than the get_av96() function.

I have applied the 'short shutter press' override fix in changesets 3264, 3265, but I'm not sure if that will have any influence on this script.

Thanks for that. Reading your commit comments, it appear this only happens on the first shot? In this case, according to the script's log, if there is a bug then it is happening through-out the intervalometer run on many shots.

Also, if your second svn commit comment :
The camera may engage the ND filter in bright light, but take the picture without it (not CHDK related).
is about the issue here, that's a script logic problem and not related to the patch?

I posted a patched script file for Wreck to try. With the ND filter issue fixed, we can work on whether there is an issue with the exif value not matching the override or with the override not actually being applied. The shooting part of the script code is pretty much standard stuff :

                -- set up all exposure overrides
                set_tv96_direct(tv96setpoint) 
                set_sv96(sv96setpoint) 
                if( av96setpoint ~= nil) then set_av96_direct(av96setpoint) end    
                nd_string="none"            
                if(Av_mode > 1) then                      -- ND filter available ?
                    if ( insert_ND_filter == true ) then
                        set_nd_filter(1)                         -- activate the ND filter
                        nd_string="NDin"
                    else 
                        set_nd_filter(2)                         -- make sure the ND filter does not activate
                        nd_string="NDout"
                    end
                end

                -- and finally shoot the image
                press("shoot_full_only")
                sleep(100)
                release("shoot_full")
                repeat sleep(50)  until get_shooting() == false

waterwingz

srsa_4c
I have applied the 'short shutter press' override fix in changesets 3264, 3265, but I'm not sure if that will have any influence on this script.

Looking at the code in core/shooting.c, it does not look like the change to the aperture_sizes_table[] in this patch going to change the value returned by get_av96() if the ND filter is inserted ?

srsa_4c

waterwingz
The script does not correctly handle the situation where the camera wants to insert the ND filter and reports using a smaller f-stop setting as a result. It assumes the reported Av value is the actual aperture f-stop setting and does its exposure calculations based on that. When the actual shot occurs, it ends up using calculations based on the ND filter being inserted even though it does not insert that filter.

I thought PROPCASE_AV should show the chosen Av value (it seemed to when I recorded the possible Av values).

The fix for cameras without an adjustable aperture is to use get_prop(props.MIN_AV) to determine the actual Av96 value of the aperture rather than the get_av96() function.

This is probably correct for DOF calculations, but is it really good basis for determining exposure?

Reading your commit comments, it appear this only happens on the first shot?

Testing this is rather boring, but it seemed to only affect the Tv override, when ISO was also overridden, and mostly when doing a short press. I've only seen this right after transitions to rec mode, subsequent shots were correct.

The camera may engage the ND filter in bright light, but take the picture without it (not CHDK related).
is about the issue here, that's a script logic problem and not related to the patch

I was referring to live view, the ND is moved to its final position on half-shoot.

waterwingz

srsa_4c
I have applied the 'short shutter press' override fix in changesets 3264, 3265, but I'm not sure if that will have any influence on this script.

Looking at the code in core/shooting.c, it does not look like the change to the aperture_sizes_table[] in this patch going to change the value returned by get_av96() if the ND filter is inserted ?

Probably, but people without the cam will now at least see what's available.

« Previous Page Next Page »