Magic Lantern for 5D Mark III - Alpha 3 (1.1.3)

[a1ex]

how is that particular call dangerous? if i understand it correctly, it changes camera settings and can potentially lead to a non-booting camera if invalid values are written for whatever reason?

Exactly, see here for my past mistakes with that function: http://magiclantern.wikia.com/wiki/Unbricking

[wintix]

mhm, i see your point there.

[manojr]

I installed the ML Alpha version into my 5D Mark III today. It worked so perfectly. (So easy to install, if you follow the instruction)
First I updated my camera firmware from the Canon website (Firmware version 1.1.3), then I download the ML Alpha version from the link provided by ML website and put it in my secondary slot (SD Card) of my camera and updated the firmware, and it's done.. so simple...

The auto-boot feature does not work yet.
Waiting for the upcoming version of ML 2.4

Thanx for the Magic Lantern Development team for their hard work and dedication.

[squig]

For uncompressed FullHD 422 at 24p, I need "only" 100MB/s. Maybe with a 32MB write buffer it's exactly what I need.

For 2.35:1, I only need 75MB/s.

That's pure theory, so take it with a grain of salt.

Looks like the 1000x cards sustained write is maxing out at 90Mb/s. 2.35:1 uncompressed would be fantastic. A 2.66:1 option would be even better. There's some good deals on 1000x cards http://www.bhphotovideo.com/c/buy/Memory-Cards/Ns/p_PRICE_2%7C1/ci/1097/N/4270508873+4294955789

Is there any way to read 2048 pixels off the sensor letterboxed?

The other option would be a 4:3 crop for 2-3k anamorphic shooting.

[willb101]

Where is the actual download, going round the house to try and find it?

[nanomad]

first page?

There's an optional updated .fir in post #49 to get a more reliable value in card speed tests

[Phobos]

I've been testing the ML alpha software for a couple of days and it seems pretty stable.

Thanks a lot again for the alpha release! Hope to hear from future progress as well.

[pahyyz]

Tested Alpha 1 this afternoon- very happy my  5D Mk3 was still working after the multiple installs
The firmware was stable , only encountered a few  issues:
-during the firmware update/loading - never got a mssg on LCD to say the install was finished. Only after a few minutes of looking at a black screen and then hitting MENU button, could i see that the firmware had actually been updated. Later figured out, install would be finished once red light beneath LCD stopped flashing
-I could never get any ML features to display in LiveView either when previewing the scene or during actual recording.
-I could not alter preset values of the features (ie - zebras at 99%) , assume this is because it is an alpha version

I finally did find the LIGHT button to toggle features on/off during playback/review modes. Would not work in LiveView on LCD screen

When do you think that Focus features (trap focus, altered  focus patterns,  ...) and Exposure Enhancements will be implemented for 5D Mk3?

I am new to ML, but impressed with its potential. Thanks for all the hard work

[dude]

I tested yesterday during a job,
peaking was perfekt for low light.
Only one time my cam froze, but i can t recreate it... had to put out battery.
i am so happy i don t need en external monitor with peaking!!!

[JasonATL]

I'll continue testing and fill out the form when I feel like I have tried enough.

Here are two things I found today:
Setup on this test: ML on empty SD card. 4 GB CF card used for recording images/movies.

Minor display bug. The GB remaining (top right corner) does not refresh properly. It doesn't appear to erase the previous value before writing the new value.  When you go to the ML menu and come back, the screen refresh cleans this up so that the GB is not messed up (i.e., it displays properly). See screen capture here: https://picasaweb.google.com/lh/photo/gtRb7lAR1hphpAmeVO8Eus6bkbuOKDnQM31wDLF3lbE?feat=directlink

Possible error-inducing feature. This happened to the recording that I was doing while trying to capture the image referenced above. It was tricky, as I had to time it so that the GB remaining would change after I started the screen capture timer, but before the timer elapsed. It took several tries. I believe this is relevant, as you'll see. I let the CF card fill up by continuing to record. When the card filled up on this round, I got Canon error "Err 04 / Cannot save images because Card1 is full. / Replace Card1." To be clear, this did not require a battery pull. The .DAT file remained on the card, but the associated .mov file was blank. The ML debugging log (reported on screen to be writing) was also empty - it was there, just 0 KB. I tried to replicate with both no ML and ML and could not... until I tried the screen capture during recording.

My hypothesis is that the Canon firmware notes space remaining at the beginning of recording and subtracts off the recording size from this, stopping the recording in time to convert the .DAT file to .mov. The screen captures (I did three or four) caused the card to be unexpectedly full before the Canon FW could stop the recording and convert as normal. My own view is that I'm not sure that it is worth trying to fix this with anything other than a warning not to do a screen capture if you plan on filling up the card (perhaps this warning already exists and I just haven't seen it!).

[a1ex]

Actually I've noticed the free space bug two days ago (see changeset).

The other one makes sense, now I know what to try next.

[mojo]

hi,

i intend to use the SD slot for storing ML Alpha 1 and CF slot for storing video files.
Is there a way to 'lock' video recording to the CF slot even after a reboot?

regards

[a1ex]

Proposed feature set for Alpha 2:

- Magic Zoom
- Focus peaking with grayscale image
- HDR video
- Gradual ISO transitions (this is new feature, and 5D3 is the only camera where it works flawless... yet)
- Display adjustments (brightness/contrast/saturation)
- Clean HDMI out with pillarboxes
- UniWB display correction (bleeding edge)
- Zoom on old button or M-Fn (not quite reliable though, I may disable this)
- Display presets

None of these things require changing any permanent camera settings, so they should be pretty safe. Backend things being tested: DIGIC controls and state object hijacking.

[robertdouglas]

I'm OK with the next proposed feature set!
Been working with the Alpha1 since one week, haven't found serious issues.
So ready to test the next one

[crazypig]

Dear Alex,

Only one request....

Would you add autoexec.bin inside the ML please...

More easy to test

Sean

[wintix]

sounds great. hopefully you will get comfortable with using prop_request_change in an alpha soon

[odlan]

All working fine here. waiting for next release

[squig]

- Clean HDMI out with pillarboxes

Are we talking uncompressed 1080p 4:3?

[optik]

Proposed feature set for Alpha 2:

- Magic Zoom
- Focus peaking with grayscale image
- HDR video
- Gradual ISO transitions (this is new feature, and 5D3 is the only camera where it works flawless... yet)
- Display adjustments (brightness/contrast/saturation)
- Clean HDMI out with pillarboxes
- UniWB display correction (bleeding edge)
- Zoom on old button or M-Fn (not quite reliable though, I may disable this)
- Display presets

None of these things require changing any permanent camera settings, so they should be pretty safe. Backend things being tested: DIGIC controls and state object hijacking.

I think that would be fantastic

[smp]

For those loading ML on SD cards...

Did anyone ever validate if this was accurate??

Well…after some testing I have determined that, if you care at all about high speed shooting or clearing you buffer quickly, YOU DO NOT want to put a card in the SD slot. Why? Because, for some reason unbeknownst to me, Canon decided to build the 5D Mark III with one very fast CF slot which supports the newer UDMA7 protocol and a standard SD card slot which does NOT support the high speed standard (called UHS – for Ultra High Speed).  http://www.canonrumors.com/tag/5d3/

[nanomad]

Yes, it wasn't (unless you're writing to both cards)

[CaptainHook]

My first speedtest was with ML on an SD card testing my lexar 1000x


I just tested again with ML on the lexar card and no SD and got the exact same results...

[bumkicho]

- Clean HDMI out with pillarboxes

What would be the resolution of actual image?

[Christian]

Been waiting for Alex to crack the 5D mkIII...   ordered one immediately on release of Alpha 1!  Thank you Alex and whole crew for all your hard work!!!!! Can't wait to start testing!   

[sicetime]

This is all very exciting! I can't thank you guys enough for all this incredible work.

So knowing that these write speeds are around 89 mbs on lexar 1000x, what would that theoretically mean, I have read on here some pretty incredible numbers, but how feasible are they?

What would it take (monetarily, teamwork help, devs) to get these things up and running? What road blocks are you hitting?