Canon EOS 5D Mark II – good for chroma keying?

[Anthony Bari jr.]

H.264 is a finishing codec, when you edit/work with it. Its like eating a candybar with the wrapper on.

When you transcode to ProRes, you will open up the file and optimize it for editing.
All that info is stored in the original H.264 file, just zipped up and not as easy on your machine.

*if you are working with just 1 single H.264 clip you may not notice as much.

You can use the H.264 but when you bring it in to FCP it will need to render constantly.

*Production*Post-Production*
Apple Certified Instructor (Final Cut Pro 7)
“Semper Fi USMC”

[Thomas Morter-laing]

That said though, you cant replace information. It wont be any better for green screen, it will only be better for rendering and it’s effectiveness within the machine.

😀
Tom Morter-Laing
Freelance Editor
Certified Apple Product Proffessional, 2010
http://www.depictproductions.co.uk

Sony Z5, with Rode NTG2.
iMac 27″ intel i7 2.93GHz, 12GB RAM, ATI HD5750 [1GB GDDR5], 2TB Int. SATA with 2TB External HDD; (FW800), with Elgato Turbo H264HD.

[Anthony Bari jr.]

H.264 off the camera has a 4:2:2 color space
just like the ProRes 422 codec

*Production*Post-Production*
Apple Certified Instructor (Final Cut Pro 7)
“Semper Fi USMC”

[Thomas Morter-laing]

I thought Canon’s H264 profile was 420 (as suggested here: https://www.cinema5d.com/viewtopic.php?f=14&t=22347&start=0&st=0&sk=t&sd=a) although that is only a forum so it may well be inaccurate. If it IS 422, is that the same as the 7D/ 550D(T2i) ?

😀
Tom Morter-Laing
Freelance Editor
Certified Apple Product Proffessional, 2010
http://www.depictproductions.co.uk

Sony Z5, with Rode NTG2.
iMac 27″ intel i7 2.93GHz, 12GB RAM, ATI HD5750 [1GB GDDR5], 2TB Int. SATA with 2TB External HDD; (FW800), with Elgato Turbo H264HD.

[Clyde Villegas]

Same with Thomas, I would also like to confirm if the video clips recorded in the SD/CF cards from the Canon cameras (the 5D, 7D, 60D, 550D) are recorded with 4:2:2 color space.

And since I’m working in AE on a PC, there’s no reason to convert from h.264 to ProRes 422. Do I understand you guys correctly?

ut in omnibus glorificetur Deus

[Bill Davis]

You guys are largely MISSING THE POINT.

Pulling a good key TODAY is LESS sensitive to color space issues and MORE sensitive to RESOLUTION.

Think about it for a second. When we were all working with SD analog video, then the amount of information contained for EACH pixel – particularly the OFFSET between the color info pixels and the luminance value pixels was CRITICAL for keying. This is because the WHOLE signal was working with VERY limited resolution. (640×480 rasters)

Now, suddenly, you can work in a 1920 x 1024 raster – that’s an INCREDIBLE increase in the ability of the signal itself to DEFINE EDGES between the INDIVIDUAL PIXELS in a picture.

So suddenly, instead of gross fuzzy edges between things – you’ve got TIGHT well-defined edges. And VOILA – you can KEY based on that much tighter difference!

You’ve got to STOP thinking you can apply SD concepts directly to to modern HD video signals such as those generated by a modern high-rez DSLR. The historical “Chroma schmutz” (my term) of SD is dwindling as resolution increases. Yes, if you want to key for a big screen MOVIE, you should consider chasing these distinctions. But for common, garden variety video – a DSLR image will RUN RINGS around ANY classic SD source for keying.

And I can tell you directly, after keying a LOT of DSLR footage – that the keys I can pull now, are ORDERS OF MAGNITUDE better than I could EVER pull back in the day of SD video.

Simple as that.

[Thomas Morter-laing]

Lol; hey Bill, try telling what you just said to my recently aliased looking key, shot on HDV with perfect 6 point lighting and subject distancing set up by a pro DOP…

😀
Tom Morter-Laing
Freelance Editor
Certified Apple Product Proffessional, 2010
http://www.depictproductions.co.uk

Sony Z5, with Rode NTG2.
iMac 27″ intel i7 2.93GHz, 12GB RAM, ATI HD5750 [1GB GDDR5], 2TB Int. SATA with 2TB External HDD; (FW800), with Elgato Turbo H264HD.

[Thomas Morter-laing]

Cheers Dave, yeah this one happened a year or so back and I’ve learnt since! Saving up for a nanoflash as we speak… Although this new atomos ninja looks tantalising. (not worried about timecode you see, idont usually need it cos of my workflow 🙂

😀
Tom Morter-Laing
Freelance Editor
Certified Apple Product Proffessional, 2010
http://www.depictproductions.co.uk

Sony Z5, with Rode NTG2.
iMac 27″ intel i7 2.93GHz, 12GB RAM, ATI HD5750 [1GB GDDR5], 2TB Int. SATA with 2TB External HDD; (FW800), with Elgato Turbo H264HD.

[Bill Davis]

Sorry for being late getting back to this thread, but I’ve spent the past six days shooting what is likely around my 300th corporate video – including incidentally – shooting green screen keys once again with my 5DMkii.

Dave, you seem to take issue with my contention. Causing me to wonder precisely what experience YOU have with pulling keys on 5DMkii footage yourself? Have you actually DONE it, or are you relying on your technical understanding of the underlying THEORY for your contentions?

Because my experiences are NOT theoretical – they are real world practical.

5DMkii files – transcoded to ProRes 422 LT keys WORLDS better than the (admittedly difficult) keys I’ve been pulling for 10 years from 4:1:1 DV. And are generally worlds SUPERIOR to keys pulled from the BetaSP sources I used in my early career. Transcoded to ProRes 422 (Standard) the keys are substantially better. And while I haven’t needed it for my work – I have friends who regularly transcode to ProRes 444 – allow them to regularly do fine quality effects work.

I never argued that 5dMkii files were the “be all and end all” for keying. I simply argue that the excellent pixel density of these files allow for better quality keying than any solution available at a similar cost.

YMMV – and perhaps does. But I stand by my direct experiences.

“Before speaking out ask yourself whether your words are true, whether they are respectful and whether they are needed in our civil discussions.”-Justice O’Conner

[Terence Kearns]

The atomos Ninja is awesome…. if you don’t have a 5DII. I believe that the EOS cameras in video mode do not output a signal that is free of overlays (at least the red record dot). They indirectly mention it in their FAQ on the atomos website.

On on this thread because I am looking at setting up a videography/photography business. I’m already an experienced photographer who owns a 5DII and a 7D. I’m currently doing some research with a view to getting a more purpose-built video camera, and colour sampling capability is way up on the priority list for me. I imagine a video camera and a hdslr would compliment each other well.

I looked at this video on Up-Converting EOS video footage
https://vimeo.com/19908622

Seems like a lot of extra work, but it does seem to be effective. I’ll probably do mostly wedding work, but I have also taken on a couple of other projects already. I’ll probably do greenscreening at some point.

I have an IT background (software development) and with respect to “how can up-sampling add information that’s not there”. Think of it this way, the information is there, it’s just not accounted for in every frame (due to the use of tweens) with the h.264 codec. It’s been suggested to use a codec (like ProRes) by getting a transcoder to account for the colour information in every frame (or at least a lot more frequently). I don’t know ProRes that well, I’m speculating. So that addresses the perception of “adding” information in the temporal dimension.

In terms of the 1:4 ration of luminance info to chrominance info, this would logically and clearly create problems when chrominance info (accuracy) is relied upon to perform some kind of mapping – ie, chroma keying. As some of you know 4:2:0 and 4:1:1 are all examples of a 1:4 ratio. 4:2:2 is an example of a 1:2 ratio and 4:1:1 is 1:1. Ideally, this is what you would use for accurate colour mapping/translation in the 2D plane.

Once again, by Up-converting, you may not be “adding information”, but you’re giving the file a finer (chromatic) resolution which can be used by the mapping software (effects software) to chop the image up into smaller blocks – and hence having finer edges for chroma keying.

Seems to me that the software is somewhat “lazy” (in terms of processing work) about “accounting” for every pixel in every frame on the fly. Up-Converting is just a way of “pre-accounting” to save the software from needing to rely on the CPU for extra real-time calculations.

So Thomas, Up-Converting is not adding information, it’s about making the mapping task easy enough for the Effects software to do a “propper” job of slicing up the pixels.

I suspect that as CPU processing power increases, the better video editing software will have a processing engine that “accounts” for every pixel on the fly regardless of input codec. Right now they need to be programmed to be “lazy” about it for reasons of practicality.

Some contents or functionalities here are not available due to your cookie preferences!

This happens because the functionality/content marked as “Vimeo framework” uses cookies that you choosed to keep disabled. In order to view this content or use this functionality, please enable cookies: click here to open your cookie preferences.