Tuesday, May 5, 2009

canon 5D Mk2 testing continues

-----------------------------------------------------

this is a temp blog post to address some global issues I'm working on regarding testing the canon d5 mk2 camera.  We've shot some tests and have seen a film-out to 35mm and are encouraged to continue learning the best way to get the great images this camera is capable of making from the native media files and on to 35mm film.  

I've written a lot of the same information various ways so don't be offended it I repeat myself.  SOme of this is just to keep it straight in my own head.   There are quite a few interested parties willing to chime in and help us solve this problem, or at least refine the shape of the stumbling blocks.....  any input is appreciated, but PLEASE read the links below before making any suggestions.  

Ideally, we will gather a few good folks together and each work on one facet of this puzzle and make headway together, and share what we learn and compare notes without having to deal with a bunch of newbie stuff or duplicating efforts.  


----------------------------------------------------

OUR GOALS FOR POST  TESTS

Searching for a workflow that begins with canon D5 Mk II (w manual iris Nikkor glass) and ends with 24p HD with no crushed blacks.  Hopefully includes Final Cut Pro and works on a mac but willing to explore options.  Opinions appreciated, actual hands-on experience preferred!  

This post at Neo Scene Insider seems to be a great place to start, and if you haven't already read it,  or at least glanced at it to know what the issues are please don't make suggestions.  The camera records info in the shadow detail areas that are usually clipped by most all Non linear editors - Avid and FCP and others.  Retaining this shadow detail is essential, and any workflow solution has to begin with keeping this dynamic range visible.  


relevant points to consider, terms that have been floated, etc. (ie, stuff I wish I understood better and need to test)

different ways to home-convert 30p to 24p - cinetools, compressor, etc
when to do these conversions in a long form project's edit workflow
when and how to color correct
how to get best media for possible high end color work in a full service digital suite
how to do color correction and dynamic exposure corrections ourselves
possible use of Apple's "Color" portion of FCP and its compatibility with neoscene (or other programs that help save blacks)
Apple Color and Quicktime 7.6 vs neo scene
CoreAVC
ffMpeg
Neo Scene HD
Cineform's reputation and tech support in general
bringing Canon D5 mkII clips into ten bit color space
YUV vrs RGB color
broadcast color standards and how they do and don't apply to us

To restate goals a bit:  the basic idea is to shoot a long form project, either doc or narrative feature with canon 5D mk II and do as much editing and rendering and color correction ourselves in a home-built edit studio, with the goal of having a festival-ready project in 24p HD that is very close to ready for x-fer to 35mm film out.  One possible path would be to do no color correction or frame rate conversion at all, but simply work to retain as much info as possible along the path in order to bring the media to a full service digital post facility prior to film-out - while creating a parallel workflow path that builds a festival audience ready HD version that has our own home studio 30p to 24p conversions and basic color correction for shot-to-shot matching, rough color grading etc.  

I am not a post production person, just a motivated creative type who is searching for solutions and beginning to research options on my own.  I wish I could say I understood all this better already, but that is why I am seeking advice.  

thanks in advance.

-----------------------------------------------------------------------

SHORT VERSION

we want to make a sundance-wannabe film with this d5 mk2 DSLR camera.  We've already shot some tests and seen the footage converted to 24p and transferred out to film.  It looks great and the camera works well, after a fashion.  

The issues we can improve on are two

1 - we need to retain the color quality in our workflow - camera very easily creates files that get the blacks crushed and highlights clipped in non linear editors.  We want to stop that from happening, and bring the best color and contrast into final clor correction prior to 35mm film out

there is some discussion of bringing the media into "color" portion of FCP in order to work with the 8 bit color in a 10 bit space, and how this helps even though it doesn't create any new information - is simply leaves room for it to be manipulated and gives the best look but it only works with certain types of files

2 - we will have to convert 30 p to 24p, and it seems like the best method is to use FCP and Cinema Tools to interpolate the data using a method detailed below by Denver Riddle.  but we dont knwo the limits of this, or if it is truly the best, and when to do it in the day to day workflow of a feature production on location.  

-----------------------------------------------------------------------------

great thread about 30p to 24p conversions

listen to this guy DENVER, aka denverkr   he knows what he is talking about

http://www.cinema5d.com/viewtopic.php?f=13&t=1201

THIS SEEMS TO BE THE BEST METHOD

can anyone improve on this?

----------------------------------------------------------------------------



---------------------------------------------------------

color space notes

here's the best intel we're read about color space, and what the camera is putting out:


read the link he starts out with, which is here


but follow the updates....  he corrects his info twice.  

In any case, there is a lot to learn here about what the camera is spitting out as media.  

-------------------------------------------------------

frame rate 30/ 29.97 NOTES


There is also a new note about how the camera SAYS it is 30p but is in reality 29.97, which also seems to be confirmed. 


-----------------------------------------------------------------

how to get to our test clips, and where to post your edited clips or examples


contact me via email to get info on how to reach our server for the first time, and log in to get an applet file that will get you to my server where you can download and upload clips

 clarkleewalker1(at)mac.com

Once on the server, then look for the little red cone icon that says "logon_ to_project" and copy that file to your hard drive somewhere, like on your desktop

Use that method to log on next time, as the password will change soon.  

Click on it and the mac should connect to the server and appear on your desktop.  There are a few clips there already, more to come. 

Thanks again for your help.  


----------------------

end
Posted by at

3 comments:

  1. JeanDodgeMay 6, 2009 at 9:24 AM

    here's a short thread with screen shots about "brightness issues" with QT exporting

    http://www.cinema5d.com/viewtopic.php?f=24&t=143

    this thread links to this article

    http://provideocoalition.com/index.php/cmg_blogs/story/brightness_issues_with_h264_quicktime_movies/

    The articles and problems are circa 2008, so with luck the new QT 7.6 fixes some or all of this, but we should be aware that if you see brightness changing as you go in and out of QT, it may be this gamma tag glitch causing the problem

    ReplyDelete
  2. JeanDodgeMay 7, 2009 at 1:55 PM

    "graham" reports the following


    So I ran some color space tests on the 3214 clip (it has a nice color plate in the FG) and I found something interesting.

    When I transcode the clip via Compressor to ProRes HQ, there is lift and gain clipping, all the data in super white and super black is gone and there are slight, but noticeable shifts in color and saturation. What's crazy is if I transcode via Quicktime, side by side in Color the two clips, the raw and the ProRes transcode, are virtually indistinguishable. There are a few pixels here and there in the waveform that move a bit, but visually they look exactly the same and there is zero clipping.

    If I take these two clips into FCP on the same timeline, the raw experiences clipping and shifting, but if I send the timeline to Color, all the data in the super whites and blacks returns and again, the two look exactly the same. There was some documentation on this specific issue on one of the blogs you sent.

    My big issue is why, when I'm doing the exact same thing in both Compressor and QT that both supposedly use the exact same codec, am I getting two completely different results.

    I'm on a dual 2.7 power pc G5 running tiger 10.4.11 and QT 7.6.

    ReplyDelete
  3. JeanDodgeMay 7, 2009 at 5:11 PM

    an AVID user on a forum told me this today:

    Avid has sorted out RGB to YUV HD/SD (and vice versa) issues very well. If you were editing on Avid I'm sure a solution could be tracked down. And I believe it's as simple as choosing the proper import settings.

    Just as a way of education, digital images are stored between the values of 0 and 255 (assuming 8-bit color). RGB images use this entire range. YUV does too, but the range between 0 and 15 is called super black and between 236 and 255 super white. So in YUV land, 16 is considered nominal black and 235 considered nominal white.

    Depending on your delivery, having values below below 16 or above 235 may or may not cause a problem. For broadcast, it's a big no-no. Most cameras that record YUV record super black and white info. But not all NLE's properly use this info. Some see YUV and just truncate everything below 16 to black and everything above 235 to white. Avid does not do this.

    Cineform has had problems with YUV footage with super blacks and whites. To accurately represent the values between 0 and 15 it sometimes resorts to using negative integers, as per a discussion with David Newman of Cineform. It supposed to work out in the wash, but does not always.

    It looks like from the stills you've linked that the camera is recording YUV from 0 to 255 but the import process truncated everything below 16 and above 235. That's why the darkest blacks have lost their detail.

    ReplyDelete

Subscribe to: Post Comments (Atom)