Trying to understand get_dofinfo()

eponymous

38
Canon A4000

Trying to understand get_dofinfo()

« on: 16 / September / 2014, 20:24:11 »

Advertisements

I'm trying to understand how dof readings work on my A4000. Using this script to get the readings before and after a shot, I get the results below:

Code: [Select]

--[[
@Title Testing DOF
--]]
d={}
props=require("propcase")

print_screen(1)

press("shoot_half")
repeat sleep(50) until get_shooting() == true
d=get_dofinfo()
shoot()
e=get_dofinfo()

for key,value in pairs(d) do print(key,value) end
for key,value in pairs(e) do print(key,value) end

print_screen(-1)

sleep (3000)

Before the shot (d):

hyp_valid true eff_focal_length 28000 coc 5 aperture 3051 near 925 min_stack_dist 29 dof -1 focal_length 5000 hyp_dist 1644 far -1 focus 2118 focus_valid true

After the shot (e):

hyp_valid true eff_focal_length 28000 coc 5 aperture 3051 near 660 min_stack_dist 29 dof 2681 focal_length 5000 hyp_dist 1644 far 3341 focus 1103 focus_valid false

Why the changes?

Logged

John

waterwingz

12140

Re: Trying to understand get_dofinfo()

« Reply #1 on: 16 / September / 2014, 21:03:09 »

The press("shoot_half" ) makes the camera focus and set the exposure. Typically a script would then issue a press ("shoot_full"). When you chose to use a shoot() instead, you reset the process and the camera set a new focus point and exposure. Hence the new and slightly different focus values.

Logged

Ported : A1200 SD940 G10 Powershot N G16

msl

1280
A720 IS, SX220 HS 1.01a

Re: Trying to understand get_dofinfo()

« Reply #2 on: 17 / September / 2014, 04:15:20 »

The first result shows all values how they are stored as default values after turn on the camera. The second result says, the camera has determined a subject distance, but could not find the focus (focus_valid = false).

For a better understanding you can also use the CHDK DOF calculator.

CHDK Settings => OSD Parameters => DOF Calculator => Show DOF calculator

Use the option [Separat+]. This shows all valid results in green after a half press of the shutter.

msl

Logged

CHDK-DE: CHDK-DE links

waterwingz

12140

Re: Trying to understand get_dofinfo()

« Reply #3 on: 17 / September / 2014, 10:13:27 »

Quote from: msl on 17 / September / 2014, 04:15:20

The first result shows all values how they are stored as default values after turn on the camera.

I think there is some confusion here based on how eponymous asked his question. As I read it, the first value (d) that he is referring to is initialized AFTER the script does a press("shoot_half") & get_shooting() wait. So it should have the various focus values set at that point based on what the camera "sees" when the shoot_half goes to the ready state.

Logged

Ported : A1200 SD940 G10 Powershot N G16

eponymous

38
Canon A4000

Re: Trying to understand get_dofinfo()

« Reply #4 on: 17 / September / 2014, 10:14:18 »

OK, thanks for the help.

I changed the script to:

Code: [Select]

--[[
@Title Testing DOF
--]]
d={}
props=require("propcase")

print_screen(1)

press("shoot_half")
repeat sleep(50) until get_shooting() == true
d=get_dofinfo()
click("shoot_full")
release("shoot_half")
e=get_dofinfo()

for key,value in pairs(d) do print(key,value) end
for key,value in pairs(e) do print(key,value) end

print_screen(-1)

DOF still seems to be flakey though. I set the display to [+Separat] and the camera itself seems to have trouble sometimes, with varying readings of the same scene.

However d should be the settings used for the photo, right?

Logged

John

waterwingz

12140

Re: Trying to understand get_dofinfo()

« Reply #5 on: 17 / September / 2014, 10:27:15 »

Quote from: eponymous on 17 / September / 2014, 10:14:18

DOF still seems to be flakey though. I set the display to [+Separat] and the camera itself seems to have trouble sometimes, with varying readings of the same scene.

With the huge DOF of the little Powershot lenses, the actual focus point for any particular shot (other than extreme closeups) is not very important. I suspect that Canon makes the focus mechanism to be just "good enough" and thus not very precise when you look at the actual numbers.

Quote

However d should be the settings used for the photo, right?

AFAIK, the values for d & e should be the same with your revised script. Is that not the case?

Logged

Ported : A1200 SD940 G10 Powershot N G16

eponymous

38
Canon A4000

Re: Trying to understand get_dofinfo()

« Reply #6 on: 17 / September / 2014, 11:25:34 »

Quote from: waterwingz on 17 / September / 2014, 10:27:15

Quote from: eponymous on 17 / September / 2014, 10:14:18
However d should be the settings used for the photo, right?
AFAIK, the values for d & e should be the same with your revised script. Is that not the case?

Not always. Still testing, but it looks like they're different when the camera doesn't get a good focus. That is, the DOF display is not all green.

Logged

John

waterwingz

12140

Re: Trying to understand get_dofinfo()

« Reply #7 on: 17 / September / 2014, 11:28:16 »

Do you have things like continuous autofocus, servo autofocus, and safety autofocus turned off in the Canon menus?

Logged

Ported : A1200 SD940 G10 Powershot N G16

eponymous

38
Canon A4000

Re: Trying to understand get_dofinfo()

« Reply #8 on: 17 / September / 2014, 15:50:31 »

Quote from: waterwingz on 17 / September / 2014, 11:28:16

Do you have things like continuous autofocus, servo autofocus, and safety autofocus turned off in the Canon menus?

Ah, that might be it. I had servo AF on. When I turn it off, the two readings are the same.

On the idea that the camera is getting close enough, I note that the dof_info.focus value is definitely too short (as is the sd reported by CHDK's dof). For example, on a regular photo of a blank wall in my house, focus = 845, but the wall is ~1850mm away. Since the DOF is reported as 1175, that's enough to get the wall in focus. The CHDK DOF will also report different focus numbers upon repeated manual half-shoots, where I just put my finger on and off the button, so I don't think this is a very precise thing.

(The focus reading, btw, is always one of the stepped values I posted over in this thread.)

Logged

John