Landscape Focus Bracketing Script

pigeonhill

1155

Re: Landscape Focus Bracketing Script

« Reply #120 on: 12 / January / 2020, 12:27:22 »

Advertisements

@c_joerg

On the assumption the problem is in the bookend function, I've changed things.
Could you try this and try and replicate the error: https://gist.github.com/pigeonhill/7056282940415f7d99e0cc8688f48a0d
If this works I'll up issue the code and get rid of the prints.

Cheers
Garry

Logged

c_joerg

1248

Re: Landscape Focus Bracketing Script

« Reply #121 on: 12 / January / 2020, 13:04:58 »

Quote from: pigeonhill on 12 / January / 2020, 12:27:22

On the assumption the problem is in the bookend function, I've changed things.

Three attempts without errors. Lets see how it goes on...

Logged

M100 100a, M3 121a, G9x II (1.00c), 2*G1x (101a,100e), S110 (103a), SX50 (100c), SX230 (101a), S45,
Flickr https://www.flickr.com/photos/136329431@N06/albums
YouTube https://www.youtube.com/channel/UCrTH0tHy9OYTVDzWIvXEMlw/videos?shelf_id=0&view=0&sort=dd

pigeonhill

1155

Re: Landscape Focus Bracketing Script

« Reply #122 on: 12 / January / 2020, 13:06:01 »

Logged

pigeonhill

1155

Re: Landscape Focus Bracketing Script

« Reply #123 on: 12 / January / 2020, 13:26:37 »

@reyalp

If, fingers crossed, we’ve narrowed down the issue @c_joerg was having to this

Code: [Select]

repeat sleep(10) until (get_exp_count() ~= ecnt)

Can you throw some light on this, why the latest version throws up the error, but the previous one didn’t.

Logged

pigeonhill

1155

Re: Landscape Focus Bracketing Script

« Reply #124 on: 12 / January / 2020, 15:15:59 »

@c_joerg

Have tidied up the non-M3 version of the script and the pre-release version is here: https://gist.github.com/pigeonhill/7056282940415f7d99e0cc8688f48a0d

Thanks for all your support in flagging up the issue and helping to sort things out, although I have no idea what was going wrong and still don't

If I don't hear from you in the next couple of days, I'll assume all is OK and I'll update the script on the CHDK site.

Cheers

Garry

Logged

reyalp

14082

Re: Landscape Focus Bracketing Script

« Reply #125 on: 12 / January / 2020, 17:11:26 »

Quote from: pigeonhill on 12 / January / 2020, 13:26:37

If, fingers crossed, we’ve narrowed down the issue @c_joerg was having to this

Can you throw some light on this, why the latest version throws up the error, but the previous one didn’t.

IMO, again, it's extremely unlikely that it actually depended on CHDK build. If I understand correctly, the problem was intermittent for c_joerg, so the simplest explanation is that it depends on other, unknown factors.

The only way I could see

Code: [Select]

    local ecnt = get_exp_count()
    shoot()
...
    repeat sleep(10) until (get_exp_count() ~= ecnt)

failing is if shoot() failed to shoot, or the exposure counter being totally broken on that port. shoot failing is possible, though I think very rare with existing retry logic. If the exposure counter were broken, it would likely have been noticed a long time ago. AFAIK nothing significant in these areas (or really, any other area) has changed for G1 X between the versions mentioned earlier.
But there's a lot of other changes in the script so it's not clear to me that was the problem. Switching myshoot to use shoot() instead of hook loops without timeouts seems more likely.

Logged

Don't forget what the H stands for.

pigeonhill

1155

Re: Landscape Focus Bracketing Script

« Reply #126 on: 12 / January / 2020, 17:55:57 »

@reyalp

Understood and I'm certainly not challenging your insight ;-)

I've been frustrated as I simply can't replicate the issues that @c_joerg has been having.

Thanks for all your support, as always.

Logged

reyalp

14082

Re: Landscape Focus Bracketing Script

« Reply #127 on: 12 / January / 2020, 18:18:15 »

Quote from: pigeonhill on 12 / January / 2020, 17:55:57

Understood and I'm certainly not challenging your insight ;-)

FWIW, if there were changes in the g1x port or core code related to shooting, then those would be the obvious place to start looking. But in this case, pretty much all the recent changes have been specific to other ports or in areas unlikely to be related to the symptom.

There's still the possibility that a minor change in where things end up in memory or timing could expose a bug, but that doesn't really give anything to investigate without clearly identifying what exactly is failing.

Logged

Don't forget what the H stands for.

c_joerg

1248

Re: Landscape Focus Bracketing Script

« Reply #128 on: 13 / January / 2020, 12:42:00 »

Today I made a lot of runs with your latest version. All X2Inf. I did runs with an overlap CoC and CoC/2. The last problem should be solved. But I came to another problem.
On higher focal length (42mm) and CoC/2 the focus did not change anymore but the script still shoots.
Then I add prints again in your script. Log attached.
I find out, the script works for a while, and then it makes 41 shot with same focus and exposure.

Code: [Select]

F= 6849 Near= 5102 Far= 10411Then the script moves, made 2 shots and then it starts again with a lot of shots with same focus.

Code: [Select]

F= 6856 Near= 5106 Far= 10428I think, the script comes to a point, where the focus difference is too small.

I can’t understand why you remove the prints. I would always have prints in my script with logging of script params, focus, far, near, focal length, AV, TV, ISO. That really helps in these conditions.

gfbs2_.lua (7.92 kB - downloaded 12 times.)

LOG_6620.TXT (6.12 kB - downloaded 14 times.)

Logged

M100 100a, M3 121a, G9x II (1.00c), 2*G1x (101a,100e), S110 (103a), SX50 (100c), SX230 (101a), S45,
Flickr https://www.flickr.com/photos/136329431@N06/albums
YouTube https://www.youtube.com/channel/UCrTH0tHy9OYTVDzWIvXEMlw/videos?shelf_id=0&view=0&sort=dd

pigeonhill

1155

Re: Landscape Focus Bracketing Script

« Reply #129 on: 13 / January / 2020, 13:01:36 »

@c_joerg

Personally I would not use the script beyond, say, 30-40mm, as, as you found, the equations may break down and you will end up with loads of focus brackets.

Also you could increase the overlap criterion to coc, ie not so low. Coc/2 is very low.

In an earlier version I did have a feature to handle this situation, I’ll retook at that in the light of your findings.

Bottom line: Focus bracketing works well at the wide end, but don’t go too long.

Plus, looking at your log, I think you have maxed out the info that CHDK can provide. Once again, I’ll look at putting in a check for this situation.

« Last Edit: 13 / January / 2020, 13:11:29 by pigeonhill »

Logged