S95 Porting Thread

philmoz

Microfunguy
So, on the SX30, what is HDR shooting ?

Is it continuous mode shooting ?

David

The SX30 doesn't have HDR mode.

The S95 and G12 have this mode, where the camera takes three bracketed images and combines them in camera to create a HDR (High Dynamic Range) image.

Regards,
Phil.

philmoz

waldo

Also, I noticed that you did not mention calling capt_seq_hook_set_nr(). Is that not needed?

I haven't figured out what this does, or what the correct location for the nrflag variable is yet so I haven't included it.

Regards,
Phil.

philmoz

waldo

Maybe I'm mistaken about this, but I had thought that a function with the name "wait_until_remote_button_is_released" has something to do with the USB remote feature. I'm not using that feature. I run a script containing two "shoot" statements and nothing else. Maybe wait_until_remote_button_is_released is also an issue, but I'm just trying to get the basic stuff working at the moment.

...Unless you are saying that wait_until_remote_button_is_released affects whether capt_seq_hook_raw_here gets executed ?

That's my understanding; but I could be wrong.

Can you send me the script you are trying to use and I will test it on the SX30.

I know scripted shooting works because the badpixel.lua script works to create the badpixel.bin file (it takes two images using the 'shoot' function).

Regards,
Phil.

waldo

philmoz
Can you send me the script you are trying to use and I will test it on the SX30.

@title Test
shoot
shoot
end

sikahr

Hi,

sorry for offtopic,

please someone test how much memory is free after chdk loading on S95 in play mode with empty memory card ( black screen - no pictures),

thanks in advance, this is needed for SX130 chdk development

PS
To find free memory:
Press ALT and then press MENU, navigate to the 'Miscellaneous stuff' menu and select 'Show memory info".

philmoz

waldo

philmoz
Can you send me the script you are trying to use and I will test it on the SX30.

[code]@title Test
shoot
shoot
end[/code]

OK, couldn't get much simpler - I'll give it a try tonight to make sure it works on the SX30 😀

Had a quick look at the code - are you sure you have the correct value for PROPCASE_SHOOTING?

Update:
Had another look at your 'capt_seq.c' code I downloaded yesterday.

In the version I have (v6) you have the patched call to 'sub_FF98AEC0_my' (jumptable entry 0) commented out and are calling the firmware version of this function. You need to call the patched version because it in turn calls the patched version of sub_FFAF9990_my (which calls 'capt_seq_hook_raw_here').

My understanding is that the code path through jumptable entry 0 gets triggered if you go from half press to full press very quickly - perhaps the script code is triggering this path.

Hope this helps,
Phil.

emutier

sikahr
Hi,
sorry for offtopic,
please someone test how much memory is free after chdk loading on S95 in play mode with empty memory card ( black screen - no pictures),

I loaded CHDK with play-startup on a 16GB SDHC after deleting all pictures

Free Memory 545088 bytes
CHDK size 269552 bytes
loaded at 0x16EE30

hope that is helpfull#
greetz

Microfunguy

philmoz

I haven't figured out what this does

Quite simply it sets the Canon 'nrflag' variable to 0,1 or 2 as determined by the noise-reduction option (Auto/On/Off) that the user has chosen in one of the CHDK menus.

It is also the point where ' shutter_open_time' is determined for the DNG EXIF header.

AFTER the image has been captured, the Canon firmware will noise-process the raw sensor data .. or not .. depending on the value of that variable.

The jump to wait_until_button_is released() is normally a few lines above that, BEFORE the image is captured.

David

Microfunguy

waldoI'm not using that feature.

That is unfortunate 😀

...Unless you are saying that wait_until_remote_button_is_released affects whether capt_seq_hook_raw_here gets executed ?

No, that first function will pause the capture task.

Without it, the photo will be taken and when the raw data has been saved capt_seq_hook_raw_here() will report 'raw saved' to spytask.

sikahr

Hi emutier,

Thank You for Your work, very helpful to me.

sika

Microfunguy

@Waldo

I am attempting to compile a very preliminary port to SDM.

rand and srand are not in stubs, you have defined them in wrappers.c.

The compiler is complaining that they are unresolved addresses in stubs_entry.s.

Any ideas ?

philmoz

Microfunguy
@Waldo

I am attempting to compile a very preliminary port to SDM.

rand and srand are not in stubs, you have defined them in wrappers.c.

The compiler is complaining that they are unresolved addresses in stubs_entry.s.

Any ideas ?

Comment out OPT_GEN_STUBS=1 in buildconf.inc

You may also need to rename stubs_entry.S.err back to stubs_entry.S

Microfunguy

philmoz
Comment out OPT_GEN_STUBS=1 in buildconf.inc

SDM does not have that option.

I will check the other Make files.

EDIT:

Cannot see anything that disables stubs.

In platform\makefile_sub.inc, CHDK has :-

ifdef OPT_GEN_STUBS
rm -f stubs_entry.S.err

As SDM does not have that, that is the same as commenting it out if it did.

EDIT2:

Well, if I delete the error file and rename the stub file and continue compiling without cleaning, the compiling succeeds.

After cleaning, it will again fail at the same position.

David

philmoz

Microfunguy

philmoz
Comment out OPT_GEN_STUBS=1 in buildconf.inc

SDM does not have that option.

I will check the other Make files.

EDIT:

Cannot see anything that disables stubs.

In platform\makefile_sub.inc, CHDK has :-

ifdef OPT_GEN_STUBS
rm -f stubs_entry.S.err

As SDM does not have that, that is the same as commenting it out if it did.

David

OK, the other way is to add rand and srand to stubs_entry_2.s using the NSTUB macro instead of the NHSTUB macro. The address you use doesn't matter as these will get overridden by the versions in wrappers.c

Phil.

Microfunguy

Well, using NSTUB gets past the stubs problem and many other modules then compile.

generic\wrappers.c then fails to find most firmware functions because stubs_entry.S is always empty !

I have now replaced it with Waldo's original and it compiles fine.

Whether the camera will boot is another matter 😀

philmoz

Microfunguy
Well, using NSTUB gets past the stubs problem and many other modules then compile.

generic\wrappers.c then fails to find most firmware functions because stubs_entry.S is always empty !

Did you undo any changes to makefile_sub.inc?
With the original makefile_sub.inc, and the NSTUB entries in stubs_entry_2.s a full clean and build should work.
(You should not have an empty stubs_entry.s file).

waldo

philmoz
Had another look at your 'capt_seq.c' code I downloaded yesterday.

In the version I have (v6) you have the patched call to 'sub_FF98AEC0_my' (jumptable entry 0) commented out and are calling the firmware version of this function. You need to call the patched version because it in turn calls the patched version of sub_FFAF9990_my (which calls 'capt_seq_hook_raw_here').

The code you have is from November 21 - before I attempted to do anything in capt_seq other than the exposure overrides. That code was commented out to prevent the camera from crashing.

waldo

I sent some more time trying to see why capt_seq_hook_raw_here is not being called. Using Phil's suggested locations, I found that the spot for wait_until_remote_button_is_released is probably pretty close since the line before that is executed just before the picture is taken and the function that it is in exits just after the picture is taken.

The spot for capt_seq_hook_raw_here is probably not correct, since that part of the code is never called (or maybe is called in some shooting mode I have not tried yet). If I move capt_seq_hook_raw_here to just after the function that calls wait_until_remote_button_is_released, then my ubasic shoot command works as it should and the script does not hang waiting for the flag that is set by capt_seq_hook_raw_here.

I don't know if the other functions performed by capt_seq_hook_raw_here are working, but at least capt_seq_hook_raw_here is now being called after the picture is taken.

waldo

Microfunguy
@Waldo

I am attempting to compile a very preliminary port to SDM.

rand and srand are not in stubs, you have defined them in wrappers.c.

The compiler is complaining that they are unresolved addresses in stubs_entry.s.

Any ideas ?

I would guess that you are compiling games. I don't include the games (funny how no one has complained 😀). I think games may be the only reason to need rand(). I would think you could add the prototype for it to lolevel.h

Microfunguy

waldo
I would guess that you are compiling games.

Certainly not, no games are allowed on SDM 😀

I am just curious why your version compiles even though 'rand' and 'srand' are commented-out in stubs_entry_2.s.

« Previous Page Next Page »