EOS M3 porting [M3] [EOS_M3]

pigeonhill

@c_joerg

BTW just confirmed the DoF info upper and lower is updated, but seem to need a sleep for info to update.
Looking good 😀

c_joerg

srsa_4c
You may want to check again (watch the splash screen).

That is what i did. Check M3, SX50 and G1x. All CHDK DE.
May be the M3 is different because I build it by myself last time.

pigeonhill
BTW just confirmed the DoF info upper and lower is updated, but seem to need a sleep for info to update.
Looking good 😀

Sure?
Did you compare the values with the result of my script?

pigeonhill

@c_joerg

Haven't done that but focused at about a 1m away (guess) and ran the script to drive the lens to near.
DoF values recorded as
108/93 at the start and 15/15 at the end.

I guess the middle of 108/93 is a reasonable estimate and of course 15cm in the focus min as stated by Canon.
Need to work out a way to know how long to sleep to ensure dofs updated.

c_joerg

pigeonhill
Haven't done that but focused at about a 1m away (guess) and ran the script to drive the lens to near.
DoF values recorded as
108/93 at the start and 15/15 at the end.

Reply 2: These functions (DOF) were not implemented
https://chdk.setepontos.com/d/13756

@Ant
Is DOF implemented now?

pigeonhill

Here's my test script, just to get me going 😉

Thanks to @c_joerg

--[[
@title M3 Focus Test
@chdk_version 1.4

--]]

--  calls to use
--  call_event_proc("EFLensCom.FocusSearchFar")
--  call_event_proc("EFLensCom.FocusSearchNear")
--  call_event_proc("EFLensCom.MoveFocus",steps,speed)

bi=get_buildinfo()

function get_focus_distace_upper()
    local  x=-1
  
    if (bi.platform=="m3") and (bi.platsub=="101a") then
      x=peek(0x00244916, 2)
    elseif (bi.platform=="m3") and (bi.platsub=="120f") then
      x=peek(0x0024495A, 2)
    else
      print('Wrong platform')
    end
  
    return x
  
  end
  
  function get_focus_distace_lower()
    local  x=-1
  
    if (bi.platform=="m3") and (bi.platsub=="101a") then
      x=peek(0x00244918, 2)
    elseif (bi.platform=="m3") and (bi.platsub=="120f") then
      x=peek(0x0024495C, 2)
    else
      print('Wrong platform')
    end
  
    return x
  
  end

print(get_focus_distace_upper(),get_focus_distace_lower())
call_event_proc("EFLensCom.FocusSearchNear")
sleep(2000)
print(get_focus_distace_upper(),get_focus_distace_lower())

BTW if I comment out the sleep, the script fails, ie upper lower not updated 🙁

pigeonhill

@c_joerg

Not sure if you have experienced this, but I have confirmed you need a sleep after calling the move to register the upper and lower dofs.

Here is a snippet of my test code:

upper_pos = get_focus_distace_upper()
lower_pos = get_focus_distace_lower()
print(upper_pos,lower_pos)

call_event_proc("EFLensCom.FocusSearchNear")
sleep(sleep_time*100)
upper_pos = get_focus_distace_upper()
lower_pos = get_focus_distace_lower()
print(upper_pos,lower_pos)

call_event_proc("EFLensCom.MoveFocus",steps,speed)
sleep(sleep_time*100)
upper_pos = get_focus_distace_upper()
lower_pos = get_focus_distace_lower()
print(upper_pos,lower_pos)

If sleep_time is 0 then upper and lower don't change.

If sleep_time is 1, ie 100ms, then they do change but not as expected, ie distance change hasn't completed.

if sleep_time is 2, ie 200ms, or more then all looks good.

I guess the issue is that call_event_proc("EFLensCom.*") is not waiting for completion.

A sleep does work, but not very intelligent. I'll think about other ways, ie a if test.

c_joerg

pigeonhill
Not sure if you have experienced this, but I have confirmed you need a sleep after calling the move to register the upper and lower dofs.

I don't have experience about that....
But I would expect if you go from upper to lower it need some time.
Would be interesting to know what this speed is when you use SearchFar instead of MoveFocus.

Du you have a Sigma lens?
It looks like you didn't get upper and lower limits with Sigma lenses.

Ant

c_joerg
@Ant
Is DOF implemented now?

No.

It looks like you didn't get upper and lower limits with Sigma lenses.

How did you get it?

P.S. I have Sigma 18-250.

pigeonhill

Iâ€™m only using an 11-22mm EFM lens 😀

c_joerg

Ant
How did you get it?

I have a SIGMA 24mm f1.4 ART. I just checked again. My faultâ€¦
The SIGMA lens only moves in AF mode, not in MF. All my Canon lenses moves in AF and MF

M3_focus_SIGMA_24ART.csv

54.73KB

pigeonhill

I was getting somewhere testing what explicit sleep I needed, then trouble.

I got a memory card error message, no code number.

Tried three different cards, ie camera is the problem 🙁

So further testing on hold until I get the camera replaced, ie just purchased it!

reyalp

pigeonhill
I got a memory card error message, no code number.

Tried three different cards, ie camera is the problem 🙁

Did you try cards without CHDK? (Guessing yes, but just in case)

pigeonhill

Yep 🙁

pigeonhill

@c_joerg

Can only think about things until I get my camera replaced, but clearly we need a way to know that upper and lower have been updated after a move call, even when they don't change in actual value, i.e. new value = old value.

Until these are updated and can be accessed to make decisions, the script needs to pause.

I don't like putting a dumb or explicit sleep.

I wonder if there is a way to check this in the backend?

What do you think?

c_joerg

pigeonhill
I don't like putting a dumb or explicit sleep.

I wonder if there is a way to check this in the backend?

What do you think?

It definitely makes sense. But I have no idea if that works. I would look for EFLensCom in this list. Maybe you will find something useful.

https://app.assembla.com/spaces/chdk/subversion/source/HEAD/trunk/platform/m3/sub/120f/funcs_by_name.csv

pigeonhill

@c_joerg

Thanks for the pointer to that page.

As soon as I get a replacement cam I'll start exploring.

There are a few interesting looking hooks, eg:

0xfc43e2cf,EFLensCom.DistanceInfo

I assume I just use as:

call_event_proc("EEFLensCom.DistanceInfo")

reyalp

pigeonhill
There are a few interesting looking hooks, eg:
[code]0xfc43e2cf,EFLensCom.DistanceInfo[/code]

I assume I just use as:

[code]call_event_proc("EEFLensCom.DistanceInfo")[/code]

From the disassembly, it looks like that prints the info (almost certainly to the camera stdout, which could be captured with UART redirection https://chdk.fandom.com/wiki/Debugging#UART_redirection_.28DryOS.29) so probably not directly useful, but the underlying functions and data structures might be.
It references the strings

"\nSubject Distance Near = %ld cm"
"Subject Distance Far  = %ld cm\n"
"Defocus Near = %ld mm\n"
"Defocus Far  = %ld mm\n"

Preceding calls (to fc43cd96 and fc2f2c0e in firmware 101a) appear to get the data those values are derived from.

pigeonhill

@reyalp

Once again thanks for the insight.
Without a cam at the moment 🙁
Also I'm a little lost between looking at the list at https://app.assembla.com/spaces/chdk/subversion/source/HEAD/trunk/platform/m3/sub/120f/funcs_by_name.csv
and not seeing references to fc43cd96 and fc2f2c0.
Just my ignorance.

reyalp

pigeonhill
Also I'm a little lost between looking at the list at https://app.assembla.com/spaces/chdk/subversion/source/HEAD/trunk/platform/m3/sub/120f/funcs_by_name.csv
and not seeing references to fc43cd96 and fc2f2c0.
Just my ignorance.

The funcs_by_name.csv only contains functions recognized by the sig finder.

The functions I mentioned are called by EFLensCom.DistanceInfo, identified by looking at the disassembly of that function (which you could get using capdis or ghidra). Also, as mentioned the addresses I posted are for firmware 101A because that's the one I happened to have a fully disassembly of handy for.

To use them, you'd need to analyze the disassembly of EFLensCom.DistanceInfo to understand how the values you are interested in are returned. It looks like the first one returns a pointer to a structure, and the following call and code does some stuff with it...

pigeonhill

@reyalp

Understand...I think 😉

Getting into disassembly land may be a little beyond me at the moment, ie I feel comfortable with Lua scripting.

I guess, if I knew the addresses and structures, I could access the values in Lua by peeking, eg

x=peek(0xfc43cd96, 2)

Maybe in time I'll be able to do this.

« Previous Page Next Page »