EOS M3 porting [M3] [EOS_M3]

pigeonhill

removed

Ant

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

fc43cd96 - AllocateEflensComBuffer
fc2f2c0e - EFLens.Send
fc43cdf4 - ReleaseEflensComBuffer

So these values are obtained directly from the lens and are not stored anywhere.

If someone want to get all distance info values using LUA there need to add new lua script command.

reyalp

Ant
So these values are obtained directly from the lens and are not stored anywhere.

If someone want to get all distance info values using LUA there need to add new lua script command.

Or make the underlying shooting.c functions get_dofinfo / get_focus do something different under the hood for EOS cams.

Ant

reyalp
Or make the underlying shooting.c functions get_dofinfo / get_focus do something different under the hood for EOS cams.

AFAIK these functions return only one value, but consumers want at least two values. How to do that?

pigeonhill

@Ant

One thing we need to be cognisant of is that whatever focus distance we get access to, in mm or cm or upper or lower or something else, this distance is referenced to the sensor plane.

However, most who try and use the focus distance info want the distance from the front principal plane, which we donâ€™t know.

The DoF equations that we tend to use make use of the simple lens equation and can be off by a lot, for DoF calculations in the near field.

One can make use of a split lens model, but itâ€™s most probably not needed for non macro work, ie see my post here http://photography.grayheron.net/2019/05/splitting-things-apart.html

Iâ€™ve convinced myself, for my use case, ie landscape focus bracketing, accessing upper in cm will likely give me everything I need. This even more the case if you sensibly build in focus bracket overlaps.

Bottom line: yes it would be great to get â€˜trueâ€™ focus distance info, but for most sensible use cases I think having access to upper and lower cm distances, via peeking, will likely be sufficient IMHO

c_joerg

Ant

reyalp
AFAIK these functions return only one value, but consumers want at least two values. How to do that?

But at least dof=get_dofinfo() could deliver dof.near and dof.far. So, the peeking in the lua code is not necessary.

c_joerg

Ant
AFAIK these functions return only one value, but consumers want at least two values. How to do that?

But at least dof=get_dofinfo() could deliver dof.near and dof.far. So, the peeking in the lua code is not necessary.

pigeonhill

Although CHDK provides near and far DoF info, I personally prefer to calculate these myself, ie in Lua using my formulation, and, of course, accounting for diffraction.

I can then handle vis or IR cameras.

Thus, â€˜allâ€™ I need is focus position 😉

Ant

pigeonhill
Although CHDK provides near and far DoF info

These values are calculated by CHDK based on _GetFocusLensSubjectDistanceFromLens
But EOS lenses report two distances. It's not a problem to provide only one(near) subject distance.
But there need to modify the core/shooting.c file to return the values from the lens instead of the values calculated by CHDK.

Thus, â€˜allâ€™ I need is focus position 😉

Getting an absolute position of focus is not possible for EF lenses.

pigeonhill

@Ant

Yes I know.

As I said before, a close enough value is good enough most of time, at least for me 😉

Ant

Here you can try simple patch allowing to get DoF values(Subject distance) from EF lens. Not tested on fw 120F

subject_distance.patch

5.72KB

pigeonhill

@Ant

Thanks...but not much help to me 🙁
Can't compile CHDK.
I'm 'just' a Lua scripting guy 😉
Also replacement M3 doesn't arrive until Monday, hence don't know FW version.

pigeonhill

@c_joerg

Donâ€™t know if you can throw some light on this question.

When, for example, I use FocusSearchNear, to move from the current position to the macro end of the lens, is there a way to find out how many steps the lens took?

Knowing this will allow me to move back to the start position, as I use MoveFocus, ie by keeping track of the steps.

Ant

AFAIK after receiving(executing) commands like FocusSearchNear the position counter in the lens is reset to 0.

pigeonhill

🙁

As I feared.

pigeonhill

@c_joerg

Replacement M3 arrived and experimentation restarted 😀

First observation is that call_event_proc("EFLensCom.FocusSearchNear") requires a much longer sleep than MoveFocus, ie to get repeatable upper and lower focus values.

With MoveFocus I find I can 'get away' with sleep(15), ie 15 microseconds. If I use, say, 5, then the upper and lower values vary from run to run.

With FocusSearchNear I need to multiple the sleep by, say, 20, ie get up to 300.

The above observation is from simply repeating my test script, ie move to near then to a stated focus, ie upper, say, greater than the requested distance.

Knowing the above, I can now move on to the next stage and write a demo focus bracketing script.

c_joerg

pigeonhill
First observation is that call_event_proc("EFLensCom.FocusSearchNear") requires a much longer sleep than MoveFocus, ie to get repeatable upper and lower focus values.

That probably also depends on the last position. The worst case is when you go from far to near.

pigeonhill
With MoveFocus I find I can 'get away' with sleep(15), ie 15 microseconds.

Sleep is milliseconds.

pigeonhill
With FocusSearchNear I need to multiple the sleep by, say, 20, ie get up to 300.

I would go at least to 1sâ€¦but only an estimateâ€¦

pigeonhill

@c_joerg

😉 I meant milli seconds 😳

pigeonhill

@c_joerg

In your experiments did you see any difference in response when changing speed in MoveFocus?
I tried to see a difference and can't, ie speed = 0 works OK.

c_joerg

pigeonhill
In your experiments did you see any difference in response when changing speed in MoveFocus?

I have not done very much but could not see any differences.

pigeonhill
I tried to see a difference and can't, ie speed = 0 works OK.

I never checked 0. Only values between 1-255.
May be 0 is slowest possible speed.
Is there any functionality in ML to control the speed?

« Previous Page Next Page »