Magic Lantern and Canon's 2.0.4 firmware
 

After two days of hacking on the 2.0.4 firmware with IDA Pro and patchdiff2, I have Magic Lantern booting and running on 2.0.4. Hopefully I'll be able to declare a stable release sometime soon (like later this week after more testing).

The new features in 0.1.8 are (besides 2.0.4 firmware support) autobooting, waveforms, spot-meter and USB control (via PTP). There isn't really that much new code, so hopefully it won't take too long to decide that the stability is unchanged from the previous release. The ML audio control doesn't interoperate with Canon's, so you have to use the ML menu and config file to set the gains. 1080i HDMI continues to be problematic and there is no estimate for when/if it will be functional.

Outstanding work as always, Tramm... thanks for the report; much appreciated!

Wonderful!

Regarding the audio, I've been doing some testing with 1.1.0-ML compared to 2.0.4 and have found that 1.1.0-ML has less noise than the new Canon code when used with an identical, strong external signal.

That said, the Canon code looks to have a flatter frequency response, and there is no DC offset. For whatever reason, 1.1.0-ML has a DC offset on the left channel (I've now seen this on two cameras), and a slight rolloff at high frequencies. The DC offset limits headroom and can cause premature clipping.

Hopefully, Tramm will be able to dump the contents of the audio registers. I'd love to analyze the differences. It would be great to get the improved behavior of the Canon code with the lower noise offered by Magic Lantern.

I also have the juicedLink DN101 (defeats ALC externally) and a Tascam DR-100 to compare with the Canon and Magic Lantern firmware approaches.

Congratulations on the breakthrough, Tramm!

Thanks for your efforts!

I think a lot of us would be willing to pay for the HDMI to work. A Nanoflash with the 5D would solve all of the audio issues and improve the video.

HD monitor while recording? I'd definitely pay a little for that.

So would I (pay for full HDMI output) .. or at least no further resolution drop between Live View and Recording...

I guess many many people would...

Count me in on the HDMI/Pay For thing - if that helps...

Except that I don't think Tramm's talking about full screen, full res output over hdmi - just the equivalent of what you can get when you're not recording. Much better for focusing, certainly, but not much use with an external recorder.

Nice work
 

Thanks, this 1.8 update looks very promising!
I see two little dots in the centre of the screen I assume this is for the spot metering? is there going to be a way to turn them on/off in the next release?

Canon 2.0.4 concert recording...
 

I recently recorded a 4 hour live concert of a singer-guitar player. I downloaded the new firmware (Canon 2.0.4) and used a CX431. I used a Whirlwind mic splitter and DI Box that allowed me to intercept the mic and guitar feed before it went to the house soundboard. IThis allowed me to have balanced XLR feeds directly into the CX431 and Canon 5D MKII. I panned the vocal mic left and guitar right in the CX431.
As for settings in the Canon 5D MK II, I set the audio manual input 1 click right of off (all the way left).Of the 64 settings the new Canon firmware allows, I chose setting 1 of 64. At this setting, I had no hiss or noise and plenty of headroom. I had the CX431 pre-set to medium with the trim at the 1:00 position. What I got was a clean recording of the vocal and guitar on each channel. Once in FCP, I can take the audio to Soundtrack Pro for mixing and mastering. I anticipate this will turn out good. When I set the Canon 5D MK II audio to 1/3 or 1/2, I got a ton of noise. Then I called Robert Rozak at Juicelink and he convinced me to experiment... So, 2 hours before stage time, I started to experiment and came up with this extremely low setting on the 5D MK II that seemed to work. Why, I don't know but the final result was clean and quiet and with plenty of headroom. Anyone have similar success??? Pleace, Joe Ray Skrha at: joeray@alaska.net

By lowering the MKII audio levels, you effectively took the in-camera preamps out of the equation and let the Juicedlink preamps do all of the work.

If Canon was more serious about making this more of a video camera (or allowed to) they would have added a line level audio setting for external devices.

I posted my initial look at the noise levels here:

http://www.dvinfo.net/forum/canon-eo...ml#post1501780

I just completed some noise tests with the juicedLink CX231 and FW 2.0.4 with and without Magic Lantern's 0.1.8 alpha code.

Here's what I found:

* Magic Lantern with m_gain set to +17dB and d_gain set to 0 is within 0.1 dB of 2.0.4 firmware at three ticks above the minimum.

* Using a Rode NT-1A (an exceptionally quiet mic) in a dead space, Magic Lantern shows 8dB less noise than the Canon code, statistically. Most of ML's noise is at the low end. It's MUCH cleaner at high frequencies.

* The previous Canon code had a DC offset in the left channel, with and without Magic Lantern. The new Canon code has fixed the DC offset issue with and without ML.

The bottom line is that the Canon code is nice and clean. With the juicedLink, it's just a bit noisier than the H4n or DR-100, but it's solid. With the Magic Lantern 0.1.8 alpha code on top of 2.0.4, the noise is exceptionally low.

More testing to follow...

Just a quick few words of thanks to Tim Polster, Chris Barcellos and Jim Fairhurst... I really appreciate your comments on the noise issue and am in awe of your knowledge and experience. Thankyou for sharing your thoughts with me. JoeRay

Although recording out over HDMI would be nice someday, I'm much more interested right now with HD monitor out while recording. That would be huge and would be in line to drop some money on getting that working.

Funny thing is, most consumer/prosumer cameras that have "line-in" just pad the inputs and send everything through their preamps anyway. Of course feeding a hot signal into something like that is better than nothing, but you're most often still stuck working in a signal path that goes through the (usually) inferior preamps. That said, it's almost entirely acceptable to all but the true audiophile snob, LOL.

I'll bet good money this is a direct result of Canon using some combination of digital and analog gain where with ML, we can bring everything down low and only use analog gain in many cases (with an external preamp). It's pretty clear to me that this in concert with no live levels and no monitoring leaves plenty of room for ML to still add a lot of value in the workflow (despite it's extra steps). I was kind of hoping for someone like me (who mostly has been using ML for the audio features), the new firmware would obsolete ML, but obviously not. Although I guess the other really nice feature is zebras, got to like those zebras. And if we could get auto-restart on 4 gig limit, THAT WOULD BE HUGE IMHO (even if it's not a seamless split, which I'm sure it wouldn't be, it would still be AWESOME).

Can I get access to ML 1.8
 

Dear Jon. Bill, Chris or Tramm,
I am going on a private vintage train (sleepers, dome car, diner, etc) this Saturday for a week. I will be traveling with many fellow musicians. I will be filming with my 5DMKII and CX431. While I have Neumann and Schoeps microphones, I anticipate using line feeds from soundboards.
Would ML 1.8 be available by Saturday to download? I want it for the 8dB of cleaner sound. I shot a concert last week using 2.0.4 with the Manual audio on Click 1 and was very impressed at the improvement over previous firmware. Still, I can't turn down an extra 8dB of less noise. If you think it's stable enough for the audio benefits of ML 1.8, is it possible to access it? If so, where can I find a link??? I appreciate all the help and will be making a donation. Peace, Joe Ray Skrha at: joeray@alaska.net

Alfa version is here. Go down and take the latest. Be careful and after filming take battery out, always.
0.1.8-alpha for 2.0.4 - Magic Lantern firmware development | Google Groups

So far, ML 0.1.8 has been solid for me. I haven't done extensive tests though. As recommended, remove the battery often, just in case.

Sounds like a fantastic train ride. Is it the one in the Pacific Northwest?

Given that description, why run audio into the 5D2 at all? I do a lot of music, and the thought of recording it in camera never even crosses my mind, heh. Way easier, and depending on gear, way better quality, to just go double system sound for anything music related -- at least that's my two cents.

That is if you have a sync to avoid audio drift. It is not as simple as just recording to your sound device of choice and letting it and the video run separately.

I use a digital timepiece for larger music jobs, and it keeps the time pretty close, but it is not practical for all recording situations.

Have you actually experienced any drift recording dual system with these cameras? Remember you're never shooting more than 12 minutes of video in a single take - I've never seen a clip up to that length drift noticeably.

No, I am still in a testing phase with the vDSLRs. But I will say I have never known any video camera and an external recording device to be in sync. They all use different crystals.

I have seen footage drift after five minutes and would prefer not to sync and re-sync in post.

Are you implying that a few frames out of sync is o.k.? Maybe for a motion picture with ADR and a lot of cutting but not for music recording.

I have never used the sync software people talk about but I wonder if audio integrity is retained after the stretch.

Hey guys,

I know video peaking has been on the list for Tramm for awhile but has he ever made any post about whether or not it was really possible to do this through firmware?

Thanks

You will find it in menu with name "Edge detect". Edge detect and zebras are working only with camera. External monitor (Marshall hdmi) is showing them in false places.

Canon 5D MKII with ML 1.8 and DN-101
 

Today I installed the DN 101 0n my CX431. Let me review: New firmware 2.0.4 with manual-1 click... With ML 1.8 the audio is much clearer and it has less noise. I then add the DN-101 and I used the headphones to monitor the sound... The honeymoon stops here... According to what I read, JuicedLink recommends we pan all feeds to the left channel. All unwanted noise is put on the other channel. When we save the video on Soundtrack Pro, we are to delete the noise channel and convert the audio to mono giving us sound on both channels. I did it and it seems to work...but what a pain. Additionally, when using DN-101, you cannot rely on their meters. I suggest you rely on ML meters only. Second, the sound you hear in the headphone jack is also very distorted and not reliable. Hence, the only use of the DN101 is to separate the sound channels which will give us MONO at best.
Any suggestions or have I screwed up somewhere... Appreciate anyones thoughts. Joe Ray at: joeray@alaska.net

DN101 with 2.0.4 and ML 1.8... Is it beneficial or not !!!
 

I continued working with ML 1.8 and CX431 (with mic input setting and gain high) with a DN101... I set the ML analog gain 10dB and Digital gain to 0. I compared the sound with and without the DN101. Honestly, I cannot see an advantage using the DN101. The meters on the ML are much better and accurate, the headphone output on A/V is much better than DN101 and the audio on the ML remains in stereo compared to the DN101 which is mono.
Does anyone have a different opinion? Also, Jon, Chris, has anyone had better results with different audio settings with ML? I am running soundboard unbalanced RCA to XLR but because of the low volume out of the soundboard, I had to have the CX431 set to mic instead of line. Again, any thoughts??? Peace, Joe Ray at: joeray@alaska.net

Joe:

DN101 is not for Magic Lantern. It is for those who don't use ML, and want to shut down the AGC. And Yes, in that case, you are giving up one channel, and your recording will be mono. Last year in April /May, before ML, I had used an MP3 player with a 16kz tone to do the same thing they are doing with the DN101.

What are you using to get sound to headphones from the 5D ? Are you boosting with an amp ?

I use pluraleyes which doesn't stretch the audio at all - it just determines the sync between camera and the external audio and lines them up. You can get bad drift in FCP if you get the sequence settings wrong (far less likely with the latest firmware upgrade) but as long as settings are correct there's no drift that I notice. I'm not doing a lot of music, primarily interviews, but if it is drifting by a frame or so over a 12 minute clip it doesn't seem to be enough to notice with a person talking. Maybe more so with some music performances, just haven't really tested that.

I'm confused here. With or without ML we've moved past the point where you need to disable AGC with any external device. In other words the DN101 isn't "for those who don't use ML"...no one needs to disable AGC, Canon just did. Obviously I know you know that Chris but the post is confusing.

Well, unless you own a 7D of course...

Here's the audio hierarchy...

With a T2i/550D, 7D, 1D4, use a JL preamp and the DN101. The sound will be almost as good as an external recorder like the Zoom and you won't have to sync.

With a CX231, 5D and FW 2.0.4 but no Magic Lantern, your sound will be at a similar level to a Zoom or H4n. You won't get a headphone output though. You could consider the DN101 for monitoring, but I don't know if the sound quality will be as good. I haven't tested that combo.

With a CX231, 5D, and Magic Lantern, you get the cleanest sound without going to a high-end recorder, and you can monitor with headphones, given the right cable. No DN101 is necessary.

Thanks for your help Jon.

Magic Lantern is still going to give you better control over gain, plus out of camera monitoring, and thus I still consider it the best resolution.

Chris is right. By "better control over gain" that means that you can adjust the analog and digital sections separately. Canon sets a fixed analog gain and lets you adjust digital only. When you crank the Canon gain all the way down, it's not as quiet (noise-wise) as turning down the ML gain.

Sure it is.

I'm using an SD302 into a Marantz PMD661 with both the 7D and 5DII. I haven't had any drift problems in the 12 min. clips. Though most songs never run a full 12 min.

Pluraleyes does a great job of lining clips up if you don't care to do it manually to a clap or board.

Singular Software

Best audio settings on the Magic Lantern 1.8
 

Dear Jon, Chris and Tim,
Thankyou for your input. It confirms my research and experimenting with these various matters.
Now that we use 2.0.4 on manual (click 1) and ML 1.8 with a 5DMKII and CX431, what levers on ML can give me the cleanest signal to noise ratio? ML has three settings that I can adjust... Output Volume, Analog Gain and L & R Gain... How do you recommend I set them (assuming I have a good clean input from a mic or line input into the CX431) ???
I read that you recommend I keep the analog gain at =10dB... how about the Output Volume? or L & R Gain? Is one of these the Digital Gain??? Where should these levels be set at to get the cleanest and best sound??? Sincerely, Joe Ray Skrha at: joeray@alaska.net

Hi Joe,

I run ML as follows:
* m_gain = +10 dB or +17dB, depending on how loud/quiet things are
* d_gain_r = d_gain_l = 0dB
* output_gain depends on the how loud you want your headphone output.

With m_gain = +17dB I get the same level of recording as FW 2.0.4 at 3 ticks over the minimum. That's where I've been running my tests, since I can match levels and compare apples to apples.

With the JL input set to line-level, I get it to calibrate with the gain on "H" and the knob at about 2:00. Switching to a mic-level input boosts the gain in the JL by 40 dB. Of course, with a microphone input, there's no such thing as calibration. You just need to adjust for proper levels depending on the loudness of the source, the distance from the mic, and the sensitivity of the mic.

external recorders as pre-amps
 

I was just wondering, would it be possible to use external recorders like the H4N or the DR100 as a pre-amp? So using their headphone or line output with a short cable as a feed to the camera? Or would that result in inferior audio (compared to using them as standalone recorders, or a JL as pre-amp)?