[Ryan Holmes]

You’ll just need to lower your data rates. You can lower the video data rate or the audio data rate.

Moving down to 0.4 or 0.5Mbps for your target video data rate will probably achieve your goal (depending on how long your source file is).

Ryan Holmes
http://www.ryanholmes.me
@CutColorPost

[Ryan Holmes]

[JP Pelc] “But with your whole second paragraph, I think you are equating ML to RAW recording. You can use many features of ML and record to the base h.264, and therefore make no increase in workflow whatsover. In fact, this is the only way I have used ML thus far. Even when using it to record to H.264 the screen overlays that ML provide are enormously helpful”

Right on. I really only use ML to shoot RAW, and thus have a more complicated workflow. I’m not sure if I’ve ever even shot h.264 using ML…that would be interesting!

Ryan Holmes
http://www.ryanholmes.me
@CutColorPost

[Ryan Holmes]

[JP Pelc] “Really? These are issues with RAW, but that can be easily switched off with ML. Otherwise ML is rock solid, easier workflow than built-in firmware (at least as far as on set goes), no additional storage or heat. You can’t possibly argue that the default firmware is in any way better than ML.”

It probably depends on your shooting environments. Here in Texas it’s hot outside generally when you shoot. When running ML software here my average temperature ranges from 113-125 Fahrenheit. Needless to say DSLR’s don’t dissipate heat well, so on several occasions the camera has just shut off while recording. Pushing RAW frames through the 5D3 creates considerable heat internally.

Again, we need to define “better.” The workflow is certainly more cumbersome when using ML compared to shooting h.264 (which I’m no fan of as a codec). But h.264 is drag and drop and can be played from nearly any modern computer regardless of storage. Shooting ML creates a .RAW file which has to be unpacked into DNG files. Those files are big and to be played at realtime demand some level of fast storage (which not everyone has). The benefits to shooting ML are enormous though – dynamic range, color fidelity, sharpness, etc. I shoot ML often, but it’s not a simple, fast format for the average user to use. And many low budget productions don’t have the time or storage capacities to shoot with ML. My main point is that “better” is dependent on your workflow, scenario, and budget. But from a technical standpoint, yes, ML is “better.”

Ryan Holmes
http://www.ryanholmes.me
@CutColorPost

[Ryan Holmes]

[JP Pelc] ” Now you can spend $12,000 on a C300 which still shoots h.264″

Just to correct some information here. The C300 shoots a MPEG-2 based codec called Canon XF which is wrapped in a MXF format. The codec is 8-bit, 4:2:2. It does not shoot to and use h.264.

[JP Pelc] ” just switch from Canon and spend significantly less on cameras that shoot 4K, RAW, and higher frame rates, with focus peaking, zebras, and other features. “

The push towards RAW and 4K seems very analogous to run towards HD a decade ago. Most in our industry were clamoring for more and more resolution, but having the whole pipeline in place to handle HD took years and years to arrive. New cameras, faster on board camera storage, faster recorders attached to your camera, faster computers, bigger SAN, larger LTO archive solutions, etc. 4K is great, but everything is still delivered in 1080i or 720p resolutions. I don’t see broadcasters moving to 4K for quite some time, specifically because it’s expensive (and I’m not talking about Blackmagic cameras here). Personally, I would forgo the whole race to more pixels and prefer more dynamic range. That could be achieved through better sensors, better compression, etc.

[JP Pelc] “Magic Lantern, here we have a group of hackers who in their spare time create better firmware than this multi-billion dollar company!”

I would put a caveat on this and define the term “better.” Better in terms of dynamic range, compression, artifacting, and moire. ML is not better in terms of stability, ease of workflow, storage, heat issues, etc. I love my 5D3 with ML, but there are situations where it’s not going to perform well because of other considerations – heat, storage, workflow, etc. RAW is powerful, but it’s rarely practical and quote challenging on small budgets.

But like I said, the 7D2 is meant for event photographers. It’s not a video camera. Canon doesn’t want to play in the low-margin HDSLR market. They will continue to push people towards the C-series of cameras if they want an interchangeable, Super-35 sensor type camera. Ultimately, I would expect Sony and Panasonic to do the same as they develop their other cinema lines. The race-to-the-bottom in tech pricing rarely ends well for the equipment manufacturer.

Ryan Holmes
http://www.ryanholmes.me
@CutColorPost

[Ryan Holmes]

What’s your budget? Shooting style? Type of work? Do you want to buy new or used? Do you have a large invest in lenses? Do you need to consider additional accessories (screens, audio, lighting, etc.)?

What camera should I buy is similar to asking, what car should I buy – it depends on what you want it to do and how much you’re willing to spend to achieve that goal. All cameras have tradeoffs…there is no perfect camera.

For some jobs a t3i is fine. For others you’d want the full frame of the 5D2 or 5D3. For other jobs the 4K capable GH4 may fit better. And for other jobs you may want a Blackmagic Pocket/Cinema/Production camera. So what camera depends on you and what you do.

Ryan Holmes
http://www.ryanholmes.me
@CutColorPost

[Ryan Holmes]

Andrea – yep….that’s pretty much standard moire. Nothing is wrong with your camera….it’s just the way many DSLR’s handle sampling the sensor for video. Like I said, some of the newer Canon DSLR’s do a better job at it (though it’s still present), and some other companies like Panasonic and Sony have pushed harder on the moire front to reduce or eliminate it. And Canon now has their cinema line (C100/300/500) which does a good job of handling fine patterns.

Ryan Holmes
http://www.ryanholmes.me
@CutColorPost

[Ryan Holmes]

Welcome to the world of DSLR shooting, specifically the 5D2. What you describe just sounds like good, old fashioned moiré – the problem of the sensor resolving a fine pattern cleanly. The best options are to avoid those types of patterns (bricks, vents, etc.) as much as possible…given wedding environments that just not always possible. The 5D3 is less prone to it, but it still happens there too.

Canon’s approach to DSLR video has been to line skip on the sensor in order to go from 24MP to 2MP. As a result of that (and the h.264 codec) you typically get left with soft video and a really good shot at some crazy moiré.

If you’re shooting in situations like that I would dedicate one of your 5D3’s to the angles that would be moiré prone, and use the 5D2 on less challenging angles. The other (more expensive) option is upgrade cameras. DSLR form factor cameras like the Panasonic GH4 or Sony A7S tend to do better handling fine patterns. Or if you want to stay with Canon, something like the C100 is also a very good option.

Ryan Holmes
http://www.ryanholmes.me
@CutColorPost

[Ryan Holmes]

If final file size isn’t a big deal (since you’re playing from a hard drive or thumb drive) then get rid of 2-pass VBR (Variable bit rate). Do a CBR and set the data rate at 16mpbs.

You can also shorten you keyframe distance down to your frame rate (or even half your frame rate). Typically I do keyframe distance at the frame rate – 24, 30, 60, etc. That way every second I’m guaranteed an I-frame as it encodes.

Ryan Holmes
http://www.ryanholmes.me
@CutColorPost

[Ryan Holmes]

[Brandon Eastman] “I don’t see a ProRes selection in Final Cut Express…”

I’m sorry…for some reason I missed this in your original post. FCE can decode ProRes (I think), but it can’t encode it upon export. You’d need to step up to Final Cut Pro (6/7 or X).

If you want to deliver 4:3 SD then you picked the right format. If you want to deliver 16:9 SD then you need to export the sequence as Anamorphic (with non-square pixies in a Pixel Aspect Ratio of 1.2121). You shouldn’t need to go to h.264 and then to MPEG-2. If possible go straight from your timeline to MPEG-2 (this cuts out the middleman so to speak). You may want to export a master of some sort that you can save and reuse in the future, and MPEG-2 is not a good format for that.

Ryan Holmes
http://www.ryanholmes.me
@CutColorPost

[Ryan Holmes]

[Andrei Popa] “No matter if I use Adobe Media Encoder or Compressor, and I select these high bitrate settings, my computer ignores them and makes a file that is usually 3 to 5mbps…”

Are you using current versions of Media Encoder or Compressor? Are you using a VBR instead of a CBR?

VBR = variable bit rate, so the software will use only the data rate it needs to create the picture. Typically this mode will have a “Maximum” setting and a “Target” setting, with target being where the software tries to peg the data rate at but it could go up to the maximum or down below the target if the extra bits are necessary
CBR = constant bit rate, so the software will use all of the available data regardless of whether it needs to or not to define the picture

Ryan Holmes
http://www.ryanholmes.me
@CutColorPost