Re: [ML] 5Dmk2 AJ 4.7

[Antony Newman]

Hi Moira,

1) Pressing the Auto_Focus button seems to 'interrupt' whatever the 'AJ_TASK' was doing (eg half way through doing a screen update).   There are a number of what look like 'random key presses' that were not well guessed by AJ4.7 that the AF-ON button had just been pressed.    -> When I release AJ5.0 there are many changes in this area (that may solve some of these issues)

2) I've completely rewritten the Menu interface.   There is no longer any ML menu code in there.
Screen shots attached.

3) This area of the code of the code should behave differently in AJ 5.0.

I should have an alpha version of this ready in the next two weeks.

Regards,
Antony

[JaKob]

wow

- Jakob

--
http://magiclantern.wikia.com/
 
To post to this group, send email to ml-d...@googlegroups.com
To unsubscribe from this group, send email to ml-devel+u...@googlegroups.com
For more options, visit this group at http://groups.google.com/group/ml-devel?hl=en

[Ba...@aol.com]

Hey Antony:

 

Looking forwared to 5.0...

[Antony Newman]

bbss.

I am interested in using the ML software for video projects.

I'm happy to share my software stack with anyone that wishes to extend it (or store it in repository!)

AJ

[Alin Simionoiu]

I personally really like to look at your menu code at the least if you have it public somewhere (or you _will_ have it public)

alins

[Antony Newman]

Alin,

Happy to share the code. (done)

It would help you to see how the interface works too .. that can wait until after I've released AJ5.0 beta.

AJ

[Alin Simionoiu]

Thank you very much AJ, really appreciate this

[bbss]

Hi

Sorry, I sent private, instead of on here.

Yes. I am a photographer, so mostly interested in the photographic
tools.

Still, many of the 550d gadgets are nice for video too I think. Like
time lapse hadr time-lapse.

I will look into if I can learn the programming environment. But I am
not so smart, so even if the motivation is there, I am not sure it is
enough.

In any case. I am very interested in this development too. But it's
functions on video is still at a too high level for me to take
advantage of these more advanced functions.

Amazing though what is being done. Looking forward to the next
release. I'm still hoping intervallometer, audiotrigger and the new
hdr function will be possible to implement.

Regards
bbss

[Antony Newman]

Very very quick intro (so the code makes sense):

Top left icon: 
+) INCREASE or DECREASE
+) Use Joy-Centre to toggle between these

Horizontal menu Options:
+) Highlighted (just like the Canon menus)
+) When text is BLUE - this cannot be changed ('X' on left)
+) '-' on left : Value will be decreased when Shutter half pressed
+) '+' on left: Value will be Increased when Shutter half pressed 
+) Numbering on right (when more than 7 items on screen -> this is now scrollable)

Extended Help
+) Displayed at bottom of screen

AJ

[Moira O'Brien LIPF]

Really looking forward to the new menus and glad that the old ML menus have gone.  This is certainly the way forward.

Moira

AJ

--

[cacasodo]

AJ,
Thanks a lot for this great update. One question: About two months
ago, I had compiled and tested Piers 550D code to disable the wind
filter on the 5D. At that time, I noticed that the raw port of the
550D filter disable code did not work. I'm not a C programmer, so I
couldn't debug and fix that issue. But I saw the "Disable Filters"
menu selection on your screen capture, I was wondering..is that
feature ported to the 5D now?

thanks again,
'sodo

[Antony Newman]

Cacasodo,

I believe what is required is a deep understanding of how the Audio chip is configured (so that we know exactly what to send the chip).  Others have had an in-depth look at this.  I have only had a cursory check.

 http://www.asahi-kasei.co.jp/akm/en/product/ak4646/ak4646_f03e.pdf

Changes I have made in the audio area (audio.c):

+) Filters:  Trammell's code explicitly change all the filters one by one.
    The 550D patch did them all in one go -> I've switched to the 550D version.
    There is a huge chunk of the manual on this (and I have not read it)
 
+) Fix of audio in MoviePlay (Patch copied from 550d guys few months ago-> Seems to work)

+) Fix of 'audio margin'.  I copied the 550d patch on 'reordering the index' to get the margin correct (2 days ago)
    I don't know if this works.

+) [Programming] Changed the variable communication to ensure that the latest versions are read (not sure if this has anything to do with the audio issue for the first few secs on the 550d)

+) [Programming] Noted that the calls to audio_configure() (to update the IC with new values) is done
both synchronously (ie do it now), and asynchronously (go do in when you are ready). 
This is inconsistent but I don't (currently) think it causes software issues

+) [Programming] Ripped out all the code that does Screen Updates, Menus and signed arithmetic.

+) Added the 02gain info into the new menu (this was my cursory glance at what the Audio IC manual)

+) [Programming] Renamed all of the ML variables (I didn't understand what the previous ones were).

+) Rewrriten the Audio register debug log to (when activated) to create a new file of output every time audio_configure is called.
    If you wish to test this on the 5D ... You will (I hope) now be able to check that after any change has been correctly applied to the
    AUDIO IC registers (ie you don't need any programming skills, just an understanding of how to configure the Audio chip)

    Might be useful for some to run (something like this) on the 550d to see if the Audio issue (louder for first few secs) is due to
    the Audio chip being reconfigured just before / or just after recording is started.

+) I've attached my own version of audio.c here.     (you may find this is a bit easier to read than the ML version).
  
    The routine:   audio_configure()    is a bit easier to to read. 

    If anyone understands 100% what needs to done to configure the audio_ic differently (and wants to know what the 50 ish lines of
    code do to make the necessary changes),  you can either ask me to turn what it does into English,    or send me what you would
    like it to do.

Hope this helps.

AJ

[cacasodo]

Thanks for the perspective AJ,
I can certainly test the "current state of things" re: your recent
ports. Will I need audio.c only or your entire branch? If your
branch, let me know where I can pick it up. Love to get my hands on
those lovely, proportional font menus!
;)
sodo

[Pelican]

It seems fixed width font for me...

--

[cacasodo]

Sorry about that..I meant to say "non-proportional font menus"
:)

[Antony Newman]

Cacasodo,

If you are not familiar with C, you can either test the AJ4.7 (earlier email), or wait until AJ5.0 comes out.

I haven't stored it at bit-bucket - but anyone is welcome to the code base.

AJ

[cacasodo]

Thanks AJ,
Might as well wait until 5.0. When do you estimate that will be out?
CS

[Antony Newman]

CS,

1 week 5 days.

AJ

[cacasodo]

AJ,
That's a bit too much accuracy for the likes of me.
;)
thanks! waiting with bated breath. cheers,
sodo

[Piers]

On Friday, March 11, 2011 1:52:42 AM UTC+11, aj_newman wrote:

+) Filters:  Trammell's code explicitly change all the filters one by one.
    The 550D patch did them all in one go -> I've switched to the 550D version.
    There is a huge chunk of the manual on this (and I have not read it)

When I put the filter disable code into the 550D I did say "I'm just turning everything off in the filter enable byte, can someone confirm this is having any effect?" (I don't have remotely good enough monitoring conditions at home most of the time to be sure). I don't recall anyone responding, so there's some chance filtration is broken on 550D and 5D. From my reading of the 4646 manual, I'm doing the right thing.

Confirmation would be nice ...

PG

[Morgan Look]

Hi PG.

I remember scouring the 4646 manual at the time you introduced the code, and as I recall it was a very straightforward, but effective way of disabling all filters.

There is scope to disable only parts of the filtering.  The info on the filters themselves is rather abstract, so I don't think I looked into that in too much detail.  

I vaguely remember I nearly piped up with a suggestion to offer the option of dis/en-abling individual filters.  However I didn't see any use for it, since they all work in the digital domain at the same resolution as the final recording.  Meaning they are fairly destructive, and effectively reduce dynamic range.

If there is anything else about the audio IC I can be of help with, I'm happy to take another read of the spec and pass on my interpretations.

Regards.

Morgan.

[Piers]

Yeah, the consensus seemed to be that, like gain, you can do all the digital fiddling you want in post, but you want the least adulterated signal on the card to begin with.

In the very early days (back when the released version had the couldn't-read-MAGICLANTERN.CFG file bug) I recall a couple of guys saying they could hear a difference when they turned it off in their own builds, just wish they'd hung around for long enough to confirm I did what they did.

Re the manual, feel free to scan it for any extra "hidden filtration" or signs that confirm/deny that setting the whole byte to 0 disables all filtration.

PG

[Morgan Look]

Just had another look, the only digital stage I can find which is not controlled with the "Digital Filter Select 1" (11H) register is the ALC.  This is disabled with the ALC<1,2> bits in the "ALC Mode Ctl 1" (07H) Register.

So yeah, setting all bits of the Digital Filter Select 1 register to "0", will disable all digital blocks, except the ALC which is disabled separately.

I guess maybe it's worth checking that these settings remain persistent during recording and playback.

On a side note, ALC on the monitoring output might actually be desirable in some cases.

Morgan.

[Piers]

ALC should be controlled by the Auto Gain parameter (and I think it's still Tram's code - infallible)

PG

[cacasodo]

Morgan/Piers,
From a 5D user's perspective, I was unsuccessful in hearing any
difference with the Disable Filters on or off. This was about a month
and a half ago:
http://groups.google.com/group/ml-devel/browse_thread/thread/f232bcb55243519d/a5a215dbff6c6041?lnk=gst&q=cacasodo+audio+filter#a5a215dbff6c6041

If you can give me the steps to validate the settings remain
persistent during record/playback, I can certainly test with AJ's
upcoming 5.0 release.

'sodo

[Piers]

'sodo,

I think short of play a full frequency sweep into the camera and compare the output with and without filters, there's nothing really specific I could say in terms of "naked ear" tests (if there was, I could do it myself!).

Hmmm, there are some sweeps here if you want to have a go:

http://www.burninwave.com/#freqsweep

Theoretically, the waveform of the recorded data should much more closely resemble the original than one with a lot of filtration. Of course it's possible that Canon aren't doing any filtration (or much anyway). But I do recall folk saying the stereo pre-emphasis filters could be heard.

(if it does look like Canon don't filter, we could probably drop the feature...)

PG

[Antony Newman]

PG,

Maybe the audio registers could be dumped before we audio_configure() them? 

AJ

[cacasodo]

Piers,
Very cool link. I'd been using a king crimson tune with lots of heavy
bass and a spectrograph in reaper, but the frequency sweeps provided
in that link will help, too.
Thanks.
Sodo
http://www.burninwave.com/#freqsweep

[Antony Newman]

Here's the free package on use on a mac (might be slightly OTT for this test)

http://www.hometheatershack.com/roomeq/

AJ