CHDKPTP: S90 Primary Focal Plane Configuration - hacking out the CCD - page 43 - Creative Uses of CHDK - CHDK Forum

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

  • 704 Replies
  • 140782 Views
*

Offline SticK

  • *****
  • 779
Re: CHDKPTP: S90 Primary Focal Plane Configuration - hacking out the CCD
« Reply #420 on: 19 / April / 2013, 07:58:43 »
Advertisements
Done.
I'm not sure if you had attempted the donut fix here and whether it's in the assembla list, since I can't identify a new filename.  Is so, please post the update and let me know.  If not, I still have planned to test your suggestion with the 100c and stock CHDK over weekend. 

BTW, you can install the ISOFIX into the 100c and make the 101a no longer a TRIAL.
Quote
Hard to believe, as all of them are lower end, with a certainly less sophisticated lens.
Perhaps the reason why they didn't implement counter-donut code in yours *is* because the lens is less sophisticated.

edit:  the magiclantern site you referred me to was down last time, and having looked it up now, yes, that seems the right area concern and supports my initial guess that indeed there is vignetting correction going on, and to a large extent, in the 101a.
« Last Edit: 19 / April / 2013, 08:54:58 by SticK »

*

Offline srsa_4c

  • ******
  • 4450
Re: CHDKPTP: S90 Primary Focal Plane Configuration - hacking out the CCD
« Reply #421 on: 19 / April / 2013, 20:26:15 »
Done.
I'm not sure if you had attempted the donut fix here and whether it's in the assembla list, since I can't identify a new filename.
No, it's only about trying to reproduce the issue.
Quote
If not, I still have planned to test your suggestion with the 100c and stock CHDK over weekend.
I'd be still interested in knowing why your two cameras produce different results.
Quote
BTW, you can install the ISOFIX into the 100c and make the 101a no longer a TRIAL.
Will do, the latter is just a matter of changing the file name though.
Quote
Perhaps the reason why they didn't implement counter-donut code in yours *is* because the lens is less sophisticated.
The current trend is to replace costly hardware (good lens) with software correction. Since Canon already corrects lens distortion, I find it a bit odd that there is no visible sign of correction against vignetting. One thing could explain it: if every camera gets its own correction profile in a manufacturing step (which is only done for higher end models).

Anyway, it means I won't be able to test whatever I may find.

Re: CHDKPTP: S90 Primary Focal Plane Configuration - hacking out the CCD
« Reply #422 on: 24 / April / 2013, 14:17:15 »
I have tested the MECHA control simulator version I on a 100c firmware camera with the latest build of CHDK from srsa_4c and can confirm that power up and mode switch are successful. I am currently changing over to the smt resistors to slim my package down but I will edit this post to confirm functionality of version II with the new PUP sequence as soon as I get my hardware set.

On another note, I am not sure if it is by your doing, or by the forum downtime yesterday, but the wiring diagram in post #414 no longer shows up.

*

Offline SticK

  • *****
  • 779
Re: CHDKPTP: S90 Primary Focal Plane Configuration - hacking out the CCD
« Reply #423 on: 24 / April / 2013, 14:26:44 »
Yes, the first thing I noticed when I logged in is none of the attachments show up at all for the whole thread.  In fact the thread appears to be several pages longer, and yet without attachments.  It's the same with my other thread too.  Hmmm .. looks like other threads are missing attachments.  Perhaps a moderator can shed some light on the subject.
« Last Edit: 24 / April / 2013, 14:31:02 by SticK »


*

Offline reyalp

  • ******
  • 13840
Re: CHDKPTP: S90 Primary Focal Plane Configuration - hacking out the CCD
« Reply #424 on: 24 / April / 2013, 15:56:28 »
Yes, the first thing I noticed when I logged in is none of the attachments show up at all for the whole thread.  In fact the thread appears to be several pages longer, and yet without attachments. 
See http://chdk.setepontos.com/index.php?topic=9739.0
Don't forget what the H stands for.

Re: CHDKPTP: S90 Primary Focal Plane Configuration - hacking out the CCD
« Reply #425 on: 24 / April / 2013, 15:59:59 »
Ah, it seems like they are on top of it. I might suggest that we use an image hosting site like imgur.com to upload pictures, they are consistent and provide short static urls.


EDIT: I haven't been on here too much, but I have been making a little bit more progress. I should have some good black and white pictures from removed bayer layer sensors shortly. I have also almost finished a board layout that will plug in directly on top of the s90 board, I want to get this thing REALLY compact.
« Last Edit: 21 / June / 2013, 14:39:44 by aggsquirrel »

*

Offline SticK

  • *****
  • 779
Re: CHDKPTP: S90 Primary Focal Plane Configuration - hacking out the CCD
« Reply #426 on: 27 / July / 2013, 16:00:56 »
@srsa_4c

RE:  Anti-vignetting hack ?  Can you still help ?

You said in an earlier post:
Quote
The current trend is to replace costly hardware (good lens) with software correction. Since Canon already corrects lens distortion, I find it a bit odd that there is no visible sign of correction against vignetting. One thing could explain it: if every camera gets its own correction profile in a manufacturing step (which is only done for higher end models).

Anyway, it means I won't be able to test whatever I may find.

I am pretty sure you might be right.  The central dark area is not perfectly centered appearing slightly to the left of frame center in Fig 1, and the right two corners are brighter than the left two.  This suggests that the position of CCD dead center is not exactly on the lens optical axis and must vary from camera to camera during assembly.   Hence I agree this is likely a correction profile that is calibrated during production, even though this is a pocket camera.

I have a new idea that might help us test a fix at your end.   

Last time we were using dark noise in an attempt to demonstrate vignetting in your cameras.  I determined noise was of little use, because it would vary and the results between the 100c and 101a were not consistent.

Here I use signal rather than noise.   The figures were acquired with a new optical test setup that emits a uniform-intensity collimated beam to the lens-less CCD (clean, with protective coating removed from its glass window // you can see some projected dust particles).  I can confirm that all three are affected: LIVEVIEW, JPG and DNG.  Figures show DNG and JPG.  The state of the 101a telephoto is fixed to always full wide (cannot be manipulated).

My idea: Figure 1 shows what I get with with the 101a looking into the collimated beam.  Fig 2 is a digital photonegative of Fig1, in effect, the kind of monochromatic shading you should see if your fix works and you can successfully disable vignetting correction on an intact camera with a lens, say by shooting a sheet of paper (or computer screen or wall)  that is uniformly illuminated (back-lit --  like a printer paper on a window) ... you would see results like from the old poor lenses on film cameras.

Sounds reasonable ? The the DISKBOOT currently on the 101a is the most recent on assembla:
s90-101a-1.1.0_2592_disableisdriveerror_nodistcorr_no_startup_lens_check_isofix_trial.zip

If we could get this to work, the 101a would be ready to be installed in the instrument !
« Last Edit: 28 / July / 2013, 09:22:14 by SticK »

*

Offline srsa_4c

  • ******
  • 4450
Re: CHDKPTP: S90 Primary Focal Plane Configuration - hacking out the CCD
« Reply #427 on: 28 / July / 2013, 09:43:26 »
@SticK

Looks like I was wrong. There is an easy method to check for this correction.
Below images are the upper left part of two test shots (chdk raw).

Wide angle


Max. zoom


I have seen this before on DIGIC 4+ cameras, but didn't know what it is. Now I do.

I suspect that your camera applies a better vignetting correction. Can you do a couple of test shots, and post a corner of those images? There is a possibility that the obvious sign of this correction (like on my "wide angle" test shot) will get cropped in CHDK DNG, in that case try it with CHDK RAW. Do not use native RAW or JPG, as both are corrected.

Make a test shot in all zoom steps, this will help to identify which steps produce corrected images.


*

Offline SticK

  • *****
  • 779
Re: CHDKPTP: S90 Primary Focal Plane Configuration - hacking out the CCD
« Reply #428 on: 28 / July / 2013, 10:06:29 »
This is very good news.  Here is the test shot from the previous figures in higher resolution // hope it helps for the moment.  I have to leave for a few hours but will be back to handle requests when I return.  I will attempt the cropping idea then.

IMPORTANT NOTE: on the 101a, I can't move the zoom because the camera will shut down, so if you're referring to getting images on the 101a at different zoom levels, it's not possible.  That's by design of the lens mecha simulator.


*

Offline srsa_4c

  • ******
  • 4450
Re: CHDKPTP: S90 Primary Focal Plane Configuration - hacking out the CCD
« Reply #429 on: 28 / July / 2013, 12:31:47 »
if you're referring to getting images on the 101a at different zoom levels, it's not possible
I know, I meant the other (hopefully fully working) S90.
Make sure that the uncorrected edge (if any ...) is visible in the crop.

Although this finding is
Quote
very good news
, I still don't know where to find the piece of code responsible for setting this correction.


There is something else that could help. You could dump a certain amount of RAM in each zoom step. You'll need CHDK 1.2 for that: http ://mighty-hoernsche.de/trunk/ http://mighty-hoernsche.de/. Note that you don't need to take pictures in the below process.

1) Set up the dumping parameters
 - miscellaneous stuff -> debug parameters -> alt +/- debug function: set it to DmpRAM
 - RAM dump settings -> start address -> press SET, make sure that this address is 0x00000000, press SET to confirm
 - RAM dump settings -> dump size -> press SET, enter 0x150000, press SET to confirm
 - RAM dump settings -> dump delay -> enter 5 (5 seconds), or possibly more
2) Switch to shooting mode, wide angle
3) activate ALT mode, press the button that is the debug shortcut on your camera(*), exit ALT mode, half-press the shutter, wait until the dump completes (if there's an LED on the camera that lights when CHDK RAW is saved, the same LED will show when RAM is dumped)
4) zoom one step in, repeat 3) for all zoom steps
5) there will be 'fake' .jpg files in the DCIM/100CANON folder, with the specified file size (0x150000 = 1 376 256). These are the RAM dumps. Since these will have similar content, a good file compressor utility such as 7-zip can compress all of them into a single archive very efficiently. You can then use a hosting site such as www.sendspace.com to share the dumps.

If you made a mistake, erase all fake .jpg files from the 100CANON folder, before you retry the whole process. I'll have to know which zoom step the camera was in when I evaluate the dumps.

(*) I don't know which button is this 'debug shortcut' button on the S90, but it's most probably the DISP button.

late edit: URL updated, CHDK 1.2 is the current stable release
« Last Edit: 16 / September / 2013, 15:17:32 by srsa_4c »

 

Related Topics