Messages

Be careful with the video, the menus doesn't correspond to the latest build version available that you can find here.

The proper description of the preview menus being documented in the original updated post of this forum thread.

You can select independent preview options during idle and recording, then check:
- set "crop rec preview" to "OFF" (not "auto mode")
- set "Preview" to "Framing" or "Frozen LV" (not "auto" nor "real-time")
- ensures "Kill global draw" is "OFF"

12bit lossless is working very well, but the menus are confusing: you need to keep 14bits lossless settings in the RAW video menu and use the "bitdepth" option in the main video menu (OFF means "14bits", then you can select 12, 10, 8bits).

THERE IS a realtime liveview during anamorphic/crop mode (some people are actually using it via HDMI to deal with focus while recording), just not the one you're expecting as it's the Canon-native 1080p hardware one when you'd prefer a replication of the ML software based framed preview (yes, the grayscaled half-resolution, doing its best fps-wise) which is actually displayed as a replacement of the hardware one, dealing properly with crop modes and anamorphic features by design.

There's currently no such thing for the 5D3 as in Bilal's crop mood for EOS-M that tweaks directly the hardware liveview, solving then both LV and HDMI displays at the same time (which will be the best solution, let's continue to hope Bilal will unlock it one day).

I tried to overwrite the current HDMI output with the framed-preview buffer but I currently get synchronisation issues and/or crashes so I give up, maybe I'll try again later.
Anyway I will prefer to solve this better than rely on an USB stream.

HDMI DOES HAVE realtime liveview when recording, the Canon-native hardware-based one (limited to 1080p), what it doesn't have currently is a replication of the Magic Lantern software-based framed preview used on the 5D3 when dealing with any crop/anamorphic modes.

mmm don't remember if it's cropped but I think its' not, I'm not sure... as I remember 1280@60fps is just a specific preset requiring first to switch to 1080p@30 in Canon menus.

To answer your last question, you always need to apply the following write speed (MB/s) rule of thumb:
( width x height x bit_depth x frame_rate x compression_rate ) / bit_to_MB_conversion
where lossless compression rate is variable depending of light conditions, but you can assume it will be 60% in the worst case.

So if dealing with a 1920x1280 12bits image at 60fps you'll get:
( 1920 * 1280 * 12 * 60 * 0.6 ) / ( 1024 * 1024 * 8 ) = 126.5625 MBps

Generally, a good CF card got a record speed of ~90MB/s, which is enough in most of the cases, but here you can see that you'll have to deal with card spanning and potentially SD overclocking depending of the SD write speed to get proper continuous recording.

Keep also in mind the maximum possible speed of the 5D3 (hardware limitation) is around ~148 MB/s (card spanning + SD overclocking) even with two SD+CF cards reaching ~90MB/s each.

In terms of 5D3 ML builds for doing raw videos, the most up-to-date is currently this one which is basically Danne's build with some additional features, like the ultrafast framed preview for crop modes (and anamorphic).

Then:

1. no matter if cropped or not, Danne's build is the best one for any raw-videos purposes on the 5D3, feature-wise.
In terms of frame rate you can get stable 23.976/24fps with almost all modes (even hi-res), 30 with some and 60fps is possible in 1080p (raw) like in this video:

(warning: both 4K and 120fps outputs here are due to upscaling/frame interpolation in post)

2. 5.7K mode is misleading (and a little bit oversold/trendy imho, compared to 3.5/4K modes) as it's a so-called "anamorphic" preset, meaning in fact a video resolution of 1920x2340 that you will have to desqueeze 3 times horizontally in post to get 5760x2340 (yes, it means a huge loss in quality), the advantage being the usage of the whole full-frame sensor (where other modes are cropped).
It means also that it may not necessarily requires card spanning technique (SD+CF recording) depending of the record bitrate (resolution+framerate+bitdepth), so maybe a CF alone is enough or even maybe a SD supporting overclocking (please refer to this up-to-date known compatiblity list)

Actually card spanning and overclocking are required only if you want to shoot very-high resolutions videos like 3.5K or 4K continuously, which is working pretty well.

Card spanning will alternate writes on both CF & SD to avoid write buffers saturation, meaning in the end you will get multiple chunks of video scattered on the two cards that you will have to join in post using MLVApp or MLVFS (quite transparent).

"live view" (the realtime/hardware-based one) is only available when dealing with native camera resolutions, any crop mode/anamorphic relying then over a quite slow software-based preview, which is at least properly framed in most of the case.
My specific build (with ultrafast preview) speed up things but it's not perfect (please check this video:

) and far from hardware-based achievements like in Bilal's crop mood build for EOS-M.

3. overclock is very reliable if compatible, I'm living in the south of France where the summer is HOT and I never get any overheat issue on the 5D3 even when spending a whole shooting day using it with 3.5K mode, card spanning and overclock quite intensely :-)

Hope it'll help!

You can use MLVFS to deal directly with MLV files in Premiere/After Effect/Resolve (create a virtual directory of cDNG files)

bravo!

Note anyway there are already a bunch of feature films shot partially or entirely using a Canon EOS (generallly the 5D mark II or III, the latter being able to record 4K RAW 14bits, which is more than the FX3) and ML, e.g. Rubber

shotonwhat is a great resource to find out: 5D2, 5D3

Get creative with your mind, a camera is just a camera and ML is just software.

As stated by The Creator DP team itself, the fuzz about the Sony FX3 is just unbelievable: the decision was mostly driven by a practical question of size and freedom of movement, the camera being maybe the less important ingredient to create a "cinematic movie" if you consider the whole equation "scenario+talents+production design+lighting+lens+camera+postprod".

Then think about how good a FX3 is compared to the cameras used to shoot movies in the last 50 years and imagine what Spielberg would have done back then with any potatoe cam or an iPhone...

So if you got 4K$ left you'd better invest them in good locations, costumes, lighting or maybe lenses if you want to achieve a "cinematic" movie.

Note anyway the biggest difference between any ML-augmented camera and the FX3 is the reliability: ML is basically a hack, meaning you WILL struggle with unexpected camera stops, limited compatibility with external monitors and recorders, limited focus capabilities, a painful post-production workflow due to MLV format, card spanning etc.

The main difference between a feature film and a short being money and time: it's ok to spend time with technical issues and to workaround limitations when alone without external/money pressure but barely possible when dealing with a full team of 50+ people and deadlines.

Not a specialist but I assume microSD is perfectly fine using a SD adapter.

For the "real" write speed of the different cards, please refer to this chart, showing what's really possible when the card is compatible with 240MHz overclocking (available with Bilal's "Crop Mood" Magic Lantern specific build) and it's activated.

Anyway I don't know if it's part of the 650D capabilities nor if it's really possible to shoot 4K with it... as far as I know, only the 5D3 is really capable to shoot (continuous) 4K at a decent framerate, mainly thanks to SD overclock and CF+SD card spanning capabilities.

@tofu007 basically you just have to select 720p video @50 or 60fps (PAL or NTSC) in regular Canon menus, then you can select 1920 50fps or 1920 60fps preset in ML menus (Danne's 5D3 build) and that's it, if needed you can override the record fps to 50 or 60 but as I remember it's already fine using the presets.
Then you generally want to use a proper shutter speed matching the target fps, like 1/100 or 1/120 to match the 180° shutter rule.

Just double-checked on my side and it's ok using the following procedure:
- exFAT (quick) format the SD card using Windows (explorer/right-click on SD driver/format)
- download the latest ML build from this repository (1.1.3 target here) then extract and copy the whole zip content to the empty SD card root
- put the SD card in the 5D3 and startup the camera
- go to the regular Canon menu/firmware update and do the update, you may see "Magic Lantern is installing"...
then restart the camera as asked by ML at the end of the firmware update and that's it

Code updated today to deal properly with lenses that doesn't report focus distance and/or are stucked at initial calibration step, e.g.: the Canon EF 50mm f/1.8 II lens, as reported by some user on YouTube.

why are you specifically searching for LJ92 compression?
does any alternative lossless compression algorithms supporting 12bits depth fits your initial need?
do you really need to work with lossless media or a professional lossy compression enough for you post-production requirements?

depending of these answers, I guess you can maybe avoid a lot of struggle just by using standard tools like ffmpeg/Resolve, alongside JPEG2000/x265/AppleProRes4444 algorithms?

why do someone say the free version is only 8 bit?

People usually say a lot of shit on the Internet
Free & Studio version of Resolve relies over the same GPU engine, Studio just got more available advanced features.
As explained by @masc, an 8-bits engine will imply very visible banding issues.

Don't get how MLVApp would have "better color" than Resolve in the end, a 14-bits integer bit depth footage fit perfectly in 32-bits float without any loss, and we're talking here about one of the most renowned color-grading software with bullet-proof color science (DaVinci Wide Gamut & ACES colorspaces etc.), so eventually the only weak point may be the RAW (DNG) debayering algorithm (personally prefer the one used by Adobe in CameraRaw) but the differences are quite indistinguishable...

@mlrocks DaVinci Resolve engine (both free & studio version) works internally using 32bits float (single precision GPU float value).

Superb, bravo!

( 3840 * ( 3840 / 2.5 ) * 23.976 * 14 ) / ( 8 * 1024 * 1024 ) = 236MB/s
  ^ width       a.r. ^   fps ^  bit dpth.^      ^---^--------^ bits to Bytes to KB to MB conversion

x0.4 in @WalterSchulz computation is the lossless compression ratio, which seems quite optimistic to me depending of the overall video exposure and the ISO (close to 100=less noise pattern=better compression): if you just increase it slightly to x0.5 then you get ~118MB/s, which sounds more realistic in real conditions with potential partial overexposures.

Sorry, have really hard time to understand your issue/question...

At least, for the last one "but if you find a way to present card data like in legacy mode", is it related to this other change?

[discontinued]

Please note this module is now discontinued, being fully replaced with the focus sequencing module.

[UPDATED 2024-01-23]

UPDATED 2024-01-23

Hello,

Just released a new focus sequencing feature allowing us to create and manage a list of focus point created via autofocus with specific transition durations, then easily replay the sequence step-by-step while recording a video, using a single button push.

Demonstration video:

Documentation:
The complete documentation (explaining both why it was a challenge regarding the lens calibration and how to use the feature) is available online using this markdown link.

Download link:
5D3 users, 1.1.3 & 1.2.3 pre-builds are available in this GitHub repository, which is basically an up-to-date fork of Danne's repository, including all the modules but also the ultrafast framed preview feature and the SD/CF dual slots free space display.

FAQ:
- does this replace the previous cinematographer module?
yes, it's a complete replacement of it, including a way more user-friendly GUI, a lens calibration process and above all an autofocus-based sequence creation which was not possible before

- why is it running on a 5D3 only as it's mainly a ML module? does a port to other cameras possible?
basically, because a change in the lens.c ML code was required first to get a reliable lens_focus function used by the focus_seq module (explained in the documentation)...
this is why it's currently embedded in the only specific Magic Lantern GitHub repository that provides this function, anyway as the change doesn't impact yet the legacy code a port in other repos for other cameras remains highly possible (pending)

Greetings:
Thanks a lot to @names_as_hard for code review, support & advices, much appreciated!


23-Jan-2024 update:
Following a user (@bigbe3) comment on YouTube, I just fixed the module to deal properly with lenses that doesn't report the focus distance and/or that are stucked at initial calibration step due to a missing forward lens limit detection, typically like the Canon EF 50mm f/1.8 II: the download link above was updated accordingly.

an easy one to port, small non-5D3-specific code addition and a minor change of menu priority, all done in the same single file (lens.c)

those drone shots are excellents!

must be totally independent of SD-overclock / Card Spanning / preferred card options, so yes, my guess is you don't exceed 1GB of write
(tell me otherwise)