CHDKPTP: S90 Primary Focal Plane Configuration - hacking out the CCD - page 70 - Creative Uses of CHDK - CHDK Forum  

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

  • 704 Replies
  • 73199 Views
*

Offline SticK

  • *****
  • 779
Re: CHDKPTP: S90 Primary Focal Plane Configuration - hacking out the CCD
« Reply #690 on: 22 / September / 2013, 09:57:21 »
Advertisements
Does the visual corruption go away if you select a menu item?
It does remain if selecting menu items, however, the top part changes morphology (fig, right panel).. instead of upward curves, they go downward.  If I highlight menu top, it's back to the original bad LUT (left).  It goes to normal if I REC and SET SET SET SET a few times (even number), and PLAY MENU, then I can get a good LUT.
« Last Edit: 22 / September / 2013, 10:14:07 by SticK »

*

Offline srsa_4c

  • ******
  • 4038
Re: CHDKPTP: S90 Primary Focal Plane Configuration - hacking out the CCD
« Reply #691 on: 23 / September / 2013, 13:12:09 »
It does remain if selecting menu items
That means, the active bitmap buffer is correct.
About the palette (based on my earlier experience on the ixus110): if I'm not wrong, the corrupted palette buffer is the one that gets fed into the display hardware. In other words, I think the LCD would show the same corruption.
Additionally, this buffer gets rewritten quite often, so the condition causing the issue is persistent. I'll try to look further...

*

Offline SticK

  • *****
  • 779
Re: CHDKPTP: S90 Primary Focal Plane Configuration - hacking out the CCD
« Reply #692 on: 23 / September / 2013, 13:27:07 »
In other words, I think the LCD would show the same corruption.
I may have mentioned I have occasionally seen quick transient bad LUTs on the 100c LCD, I think ..if not mistaken. 

Quote
I'll try to look further...
Thanks ... but in case you run into a brick wall, but you are confident there are no other serious side effects other than display, then keeping it this way I wouldn't consider a major problem.

*

Offline srsa_4c

  • ******
  • 4038
Re: CHDKPTP: S90 Primary Focal Plane Configuration - hacking out the CCD
« Reply #693 on: 25 / September / 2013, 16:23:25 »
This isn't about the palette problem, but I thought it could be useful.
I have finally made a utility for myself that enables me to look for changed/unchanged bytes/words in multiple RAM dumps. Based on your earlier 100c (zoom-related) dumps, I found the following addresses that seem to hold the current optical zoom step:
0x00002bb0
0x00002bba
0x00002d3a
0x00004e7a
0x0000bd2e
0x00039ff4
0x0003ddd8
0x0007fae0
0x000803dc
0x00080818
0x000efed5

Many of the above fw variables are bytes, so if you try to experiment, don't forget to set the "size" argument of the poke command to 1, like this: poke(0x2bba, 9, 1).
As a reminder, the above fw vars should be 0 when the lens is at the wide end, and 9 at the tele end. Since you dumped the RAM in half-shoot, the above list includes addresses that will only update when half-shooting.
The reason why I'm posting this is that one or more of the above fw variables could have an instant effect on live view. Unfortunately it could also be that the live view parameters will only update on real zoom events, which you can't have on your 101a cam.


*

Offline SticK

  • *****
  • 779
Re: CHDKPTP: S90 Primary Focal Plane Configuration - hacking out the CCD
« Reply #694 on: 25 / September / 2013, 16:54:44 »
Apologies srsa, but I don't seem to understand the link here.  I do recall having worked with changing zoom levels when you were looking at vignetting, or maybe ISO levels it was as integers ??

The reason why I'm posting this is that one or more of the above fw variables could have an instant effect on live view. Unfortunately it could also be that the live view parameters will only update on real zoom events, which you can't have on your 101a cam.

Could this possibly help in suppressing vignetting in liveview?

By the way, now that the 101a is mounted in the instrument, the results are absolutely glorious, and technically, the system is working to near-perfection.  I am now doing engineering and validation tests etc.

Something interesting that maybe you may have some insight on .... Now that vignetting is disabled, I note that my optics do introduce some vignetting, in the same pattern as Canon.  The strength difference however is something like 8% (me) vs 30% (Canon).  For future consideration, could there be a strength parameter lurking in there somewhere?  Could you comment?

Also ........

I have had only one USB freeze and that happened when I had UI off, pressed SET, realized UI was OFF and enabled UI while SET was in progress: boom ... the CHDKPTP equivalent of a blue screen!  That's not a big problem at all because those kind of things can happen, but, the result is that the image counter does not get saved.  I spoke to reyalp about that a very long time ago about persisting the counter, and now it is haunting me a bit (especially that we're getting close to Halloween).  He said it was not possible, and I think if it were, I would be writing to SD each count, adding extra wear.  Canon allows resetting to zero.  Can one of you guys find an equivalent solution, so the I can set the counter through a lua call after a crash, for example, and on the next clean shutdown, have Canon update persistence as usual ??

*

Offline srsa_4c

  • ******
  • 4038
Re: CHDKPTP: S90 Primary Focal Plane Configuration - hacking out the CCD
« Reply #695 on: 25 / September / 2013, 17:50:48 »
Could this possibly help in suppressing vignetting in liveview?
That's the "last" remaining issue that depends on zoom, so yes, it could.

Quote
For future consideration, could there be a strength parameter lurking in there somewhere?
Most probably, but we're missing the DIGIC programming manual...
Remember, all I could do is switch certain corrections on/off.

Quote
I have had only one USB freeze and that happened when I had UI off, pressed SET, realized UI was OFF and enabled UI while SET was in progress
You don't happen to have a romlog of this crash, I guess? It could be the same issue as with the hints in Canon menu.

Quote
That's not a big problem at all because those kind of things can happen, but, the result is that the image counter does not get saved.  I spoke to reyalp about that a very long time ago about persisting the counter, and now it is haunting me a bit (especially that we're getting close to Halloween).  He said it was not possible, and I think if it were, I would be writing to SD each count, adding extra wear.  Canon allows resetting to zero.  Can one of you guys find an equivalent solution, so the I can set the counter through a lua call after a crash, for example, and on the next clean shutdown, have Canon update persistence as usual ??
I think we already had this conversation. Setting the image counter param could be possible, but nobody has ever done that, AFAIK, and therefore it can be really risky. Can the camera alter the counter if it finds one or more image on the card? I don't know when it does that.

*

Offline SticK

  • *****
  • 779
Re: CHDKPTP: S90 Primary Focal Plane Configuration - hacking out the CCD
« Reply #696 on: 25 / September / 2013, 20:03:37 »
You don't happen to have a romlog of this crash, I guess? It could be the same issue as with the hints in Canon menu.
There was no crash since, so this should still be good.  By the way, I have to have hints off for all to work well ... that was established many moons ago.

Quote
Setting the image counter param could be possible, but nobody has ever done that, AFAIK, and therefore it can be really risky.
This would be very useful .. could we try it on the 100c first?

*

Offline srsa_4c

  • ******
  • 4038
Re: CHDKPTP: S90 Primary Focal Plane Configuration - hacking out the CCD
« Reply #697 on: 26 / September / 2013, 14:32:08 »
There was no crash since, so this should still be good.
The time stamp in the dump shows Occured Time  2013:09:19, and it's related to some kind of motion tracking activity of the camera - is all face- smile- tracking AF related setting off in the Canon menu?

Quote
This would be very useful .. could we try it on the 100c first?
I'll have to think about this first. Your cam has date based folder naming, I don't know how to handle the non-image counter part of the image counter parameter in this case. Incrementing the image counter (to correct it for the forgotten shots) may also be non-trivial. And, as I surely mentioned before, if there is another image counter related parameter, leaving the unknown one unchanged could brick the camera.

The manual says the following:
Quote
If you use a memory card already containing images, the newly assigned file numbers may continue the existing numbering for both the [Continuous] and [Auto Reset] settings. If you want to start over with file number 0001, format the memory card before use.

Can you try this method (putting back the latest image on the card, into an appropriately named folder)?


*

Offline reyalp

  • ******
  • 12174
Re: CHDKPTP: S90 Primary Focal Plane Configuration - hacking out the CCD
« Reply #698 on: 26 / September / 2013, 15:51:55 »
Can you try this method (putting back the latest image on the card, into an appropriately named folder)?
I suspect you can also use a dummy file with a recognized prefix and possibly extension.
Don't forget what the H stands for.

*

Offline SticK

  • *****
  • 779
Re: CHDKPTP: S90 Primary Focal Plane Configuration - hacking out the CCD
« Reply #699 on: 27 / September / 2013, 10:17:31 »
The time stamp in the dump shows Occured Time  2013:09:19, and it's related to some kind of motion tracking activity of the camera - is all face- smile- tracking AF related setting off in the Canon menu?
Date looks right.

AF Frame: Center (the other option is Face AiAF which I never set)
AF Frame size: Normal
AF Point zoom: OFF
Servo AF: OFF
AF assist beam: OFF
MF Point zoom: OFF
Safety MF: OFF
Review: OFF
Blink detection: OFF
IS Mode: OFF

Quote
if there is another image counter related parameter, leaving the unknown one unchanged could brick the camera.
Hmmm.

Quote
Can you try this method (putting back the latest image on the card, into an appropriately named folder)?
I suspect you can also use a dummy file with a recognized prefix and possibly extension.
That is a *very* interesting possibility ... guys, thanks for the insight.  I am preparing for a demo next week so I may not have an answer until I have a bit of time freed up that I can do this carefully.  It involves a forced PDN to simulate and although very safe so far, perhaps I am thinking I should avoid playing with it until the demo is over ... I'll see what I can do.
« Last Edit: 27 / September / 2013, 10:19:55 by SticK »

 

Related Topics