CHDKPTP: S90 Primary Focal Plane Configuration - hacking out the CCD - page 49 - Creative Uses of CHDK - CHDK Forum

CHDKPTP: S90 Primary Focal Plane Configuration - hacking out the CCD

  • 704 Replies
  • 142385 Views
*

Offline srsa_4c

  • ******
  • 4451
Re: CHDKPTP: S90 Primary Focal Plane Configuration - hacking out the CCD
« Reply #480 on: 03 / August / 2013, 17:33:30 »
Advertisements
This might be a bit more tedious but if I download and delete DEBUG0.LOG each time, I could tag the filename with shooting conditions and the corresponding JPG # (a more complete version of the above).  That way there is less chance I think of a tagging de-sync with appended data and due to human error.  Would that work for you ?
I think so.
May I ask the image number (the one in the file name) of your 2 test shots? The (internal) image number is in the log, but I'd like to see if I decode it correctly.

*

Offline SticK

  • *****
  • 779
Re: CHDKPTP: S90 Primary Focal Plane Configuration - hacking out the CCD
« Reply #481 on: 03 / August / 2013, 17:46:11 »
6299 and 6300.  I have DNG enabled, so I get the pair on SD per shot (ie 4 files).  I assume if that's the case, I don't have to tag the filename with image number, right?  By the way, if this has any effect on your data, Canon CR2s are more interesting to me than DNGs, but saving is mutually exclusive.  Can I enable CR2s ?  Will that help, or there is no difference?

edit: Oh yes .. another question before I start ...would you like to have a RAMDmp for each ?  Would that be useful ?
« Last Edit: 03 / August / 2013, 17:58:48 by SticK »

*

Offline srsa_4c

  • ******
  • 4451
Re: CHDKPTP: S90 Primary Focal Plane Configuration - hacking out the CCD
« Reply #482 on: 03 / August / 2013, 18:08:29 »
6299 and 6300.
Thanks, the logged number is correct.
Quote
Can I enable CR2s ?  Will that help, or there is no difference?
I don't really need pictures (aside from the jpegs' metadata, as I already mentioned), since there is nothing in the test build that would change them. Switching on cr2 may or may not have influence on the chosen code path.
Some more things you could try (NOT in the matrix, just 1-2 test shots each): change the brightness of the scene (dark <-> bright), focus distance (near - mid - far), or if there's anything that hasn't been mentioned (with possible influence on vignetting) and you're using it on the other cam.
I don't need RAM dumps at this time.

*

Offline SticK

  • *****
  • 779
Re: CHDKPTP: S90 Primary Focal Plane Configuration - hacking out the CCD
« Reply #483 on: 03 / August / 2013, 19:41:45 »
DEBUG0.LOG RESULTS

 > Manual mode.
 > JPG & CR2.

DEBUG0_SET1:  Changed ZOOM LEVEL and f-stop

 > All fixed to manual focus at one arbitrary level.
 > All at 1/8s.
 > Dark subtract is ON for all of these (as in the 101a).
 > Where you see name formats containing brackets ie NNNN [f2(2p9) Z4], eg, that means I had set f2 in the OSD but as I zoomed in, the camera readjusted automatically to f2.9.
 > Z means ZOOM LEVEL, from 0 to 9.  The middle one, ie Z4, could be in error by one level in not more than two files.
 > Suspect files 6306 and 6309 were redone as 6315 and 6316 (I may have not deleted the log file at first).
 > Probably the best subset is the f8 triplet (6312 to 6314).

DEBUG0_SET2: Special cases

> all at f/2
> at 1/8s except where indicated as 2s
> uppercase F is FOCUS

These are easy to do.  All the CR2s and JPG are available if you need EXIFs.  If you need more I can make more.  The 101a will always be at f2 and Z0, so these samples cover the cam, I think.  On the 101a, I have left autofocus ON because going to manual focus shifts the liveview frame slightly de-centering the image, by a liveview pixel or two.  The only parameters that will change on the 101a are Tv and Sv, nothing else.
« Last Edit: 03 / August / 2013, 21:32:49 by SticK »


*

Offline srsa_4c

  • ******
  • 4451
Re: CHDKPTP: S90 Primary Focal Plane Configuration - hacking out the CCD
« Reply #484 on: 04 / August / 2013, 10:44:25 »
@SticK

Thanks for the logs. Some observations:

All your test shots went through the same code path, except for the one with deactivated DFS.
It's confusing though: it appears that the first part of the shot is done with shading correction turned off (?!), and the dark frame is done with correction turned on (?!).
On the other hand, the shot without dark frame was done with shading correction, as expected.

The function that sets the correction parameters takes the following arguments:
- zoom point
- Av
- Tv
- 'flash fire' (this is either true or false, depends on whether flash will fire) ignored

If you could do a few more tests in different shooting modes (Av, Tv, P, scene, low light, auto, ...), that could help me understand what's up with that seemingly weird use of the correction flags. Try to take 2 pictures in each mode (one long expo, one short expo), and set dark frame subtraction control to 'auto'. You don't need to change anything else (zoom / focus / iso / ...), as those don't seem to influence the chosen code path.

edit:
+1 request: continuous drive mode
« Last Edit: 04 / August / 2013, 16:48:41 by srsa_4c »

*

Offline SticK

  • *****
  • 779
Re: CHDKPTP: S90 Primary Focal Plane Configuration - hacking out the CCD
« Reply #485 on: 04 / August / 2013, 11:47:17 »
Some thoughts on your observations ...

Hmmm Tv?  One of the things that might be happening is the correction function you found might also be doing "uniform shading" to handle Tv.   So my expectation there is that the real vignetting correction correction might actually separate from the uniform shading, and both could be deeper down. 

Hence leaving "uniform shading" enabled is OK and preferred because I will be using Tv.  It's the "non-uniform shading" we need to suppress.  One has to keep in mind that there may be a 3rd form, such as pixel gain correction that is only CCD dependent.  Unlikely, but if that exists, it has to remain enabled.

Bypassing shading completely (if this function does not feed amplifier gain), if possible to test out, might be interesting to see what happens.

I have to leave soon for family things but around for an hour or so // when back will address your requests.

*

Offline SticK

  • *****
  • 779
Re: CHDKPTP: S90 Primary Focal Plane Configuration - hacking out the CCD
« Reply #486 on: 04 / August / 2013, 15:21:31 »
It's confusing though: it appears that the first part of the shot is done with shading correction turned off (?!), and the dark frame is done with correction turned on (?!).
On the other hand, the shot without dark frame was done with shading correction, as expected.

In fact what you found does make sense.  My earlier equation is incorrect ... it should be the other way around: in DFS, only one of the shots has to be corrected, otherwise subtracting would nullify the correction (my stupid mistake)!  What you want to subtract out are the common signals generated by the CCD (eg hot pixels) and leave the correction in.  And it really shouldn't matter which one gets the correction, so long as pixel intensity ranges don't get truncated.

These all yield the same results:

a) EXPOSED FRAME -> DARK FRAME -> CORRECT -> SUBTRACT
b) EXPOSED FRAME -> DARK FRAME -> SUBTRACT -> CORRECT
c) EXPOSED FRAME -> CORRECT ->  DARK FRAME -> SUBTRACT

edit:
Based on your findings, I'm going to do a couple of new experiments on the 101a that should help us, when I get back.
« Last Edit: 04 / August / 2013, 15:44:26 by SticK »

*

Offline srsa_4c

  • ******
  • 4451
Re: CHDKPTP: S90 Primary Focal Plane Configuration - hacking out the CCD
« Reply #487 on: 04 / August / 2013, 18:55:31 »
Attached is a new test diskboot.bin. I have implemented a setting to fake maximum zoom for that above mentioned function.
Control over that setting is only available via the following Lua commands (runs on camera):
to enable the fake zoom hack: set_config_value(40,1)
to disable: set_config_value(40,0)
to query the state of that config value: get_config_value(40)

It's still possible that I made a mistake somewhere. All previous debug code is still in.

edit: re-added the attachment, the forum engine ate the previous one
« Last Edit: 04 / August / 2013, 18:57:38 by srsa_4c »


*

Offline SticK

  • *****
  • 779
Re: CHDKPTP: S90 Primary Focal Plane Configuration - hacking out the CCD
« Reply #488 on: 04 / August / 2013, 19:07:40 »
Quote
I have implemented a setting to fake maximum zoom for that above mentioned function.

I'm just back and was just about to start your new set.  This could be a very interesting test ... would you agree that I should see lens vignetting on the 100c, when set to f/2 on the OSD?

set_configs are easy for me to use for ON/OFF stuff etc.  So I'll try this first.

Question: what is "continuous drive mode" ?
« Last Edit: 04 / August / 2013, 19:13:36 by SticK »

*

Offline srsa_4c

  • ******
  • 4451
Re: CHDKPTP: S90 Primary Focal Plane Configuration - hacking out the CCD
« Reply #489 on: 04 / August / 2013, 19:19:25 »
Question: what is "continuous drive mode" ?
The mode that continuously takes pictures while you keep pressing the shutter button.

Quote
would you agree that I should see lens vignetting on the 100c, when set to f/2 on the OSD?
I hope so.

I'm out now, be back 'tomorrow'.

 

Related Topics