This morning I came across Rodney Charter's tweet about Rarevision's 5DtoRBG converter for Canon 5D Mark II and Canon 7D H.264/MOV footage to Apple ProRes. When Rodney speaks I tend to take note. So I was curious about this new (at least to me) tool for converting Canon EOS footage that he tweeted about.

I briefly used the Canon E1 movie plugin for Final Cut Pro, but soon went back to using the ever faithful utility belt tool every Canon DSLR editor has - MPEG Streamclip. It's familiar workflow, easy batch list, versatility, and speed are hard to match. However, MPEG Streamclip, the Canon E1 movie plugin and Compressor uses Quicktime to handle the YCbCr to RGB color conversion. Unfortunately Quicktime does not do the best job possible. Adding insult to injury, QT adds noise to mask the compression artifacts.

Rarevision explains:

"The main problem with all this is that you have to trust your NLE or compositing app to do a good job of performing the YCbCr to RGB conversion. Many programs use QuickTime internally to perform YCbCr to RGB conversion which, according to our testing, does only a mediocre job. I suspect this due to a performance compromise, in that "well enough" is suitable for most users (remember, QuickTime is designed for real time playback of those 1080p movie trailers as well). This is probably OK for general use, but unacceptable when trying to maximize the quality of highly compressed H.264 footage for visual effects work, green screen compositing or film outs. To add insult to injury, QuickTime adds noise to its H.264 output (and so does any program that uses QuickTime to decompress H.264) in what looks like an attempt to cover up H.264 compression artifacts. And guess what? There's no way to disable this. You're stuck with it if you've converted your footage with Final Cut Pro, Adobe After Effects, MPEG Streamclip or Canon's Movie Plugin-E1 for Final Cut Pro -- all of them use QuickTime to decompress H.264."

Rarevision makes a rather bodacious claim:

"5DtoRGB gets you as close as possible to the original data off the camera's sensor while putting the brakes on any additional quality loss. In short, it'll make your footage look just plain amazing!"

 5DtoRGB's features include:

• Panasonic Lumix GH1 support (video only)
• Full support for Canon 5D Mark II and 7D
• Higher quality output than Canon's E1 plugin [compare]
• Higher quality output than QuickTime Player or FCP [compare]
• ProRes 4444, 422 HQ/LT/Proxy Support
• DPX output for visual effects / green screen work
• Raw YCbCr output option for unprocessed luma channel extraction
• Output luminance matte as an alpha channel (ProRes 4444 only)
• Timecode support (extracts Exif timestamp from THM files)

Rarevision 5DtoRBG

I ran a few clips through 5DtoRGB in an informal test. Pixel peeping is not my specialty so please don't expect this to be an exhaustive or empirical test. Is there a difference between Quicktime conversion and Rarevision's 5DtoRGB conversion? Yes. Of course, your mileage may vary. Still, my first tentative experiments with 5DtoRGB are interesting.

 Original Canon 5D Mark II footage.

Original Canon EOS 5D Mark II H.264 Movie File

Original Canon EOS 5D Mark II H.264 movie

600x close-up of original Canon EOS 5D Mark II H.264 movie file

MPEG Streamclip (Quicktime Conversion) to Apple ProRes 4444 

Apple ProRes 4444 output from MPEG Streamclip. Note that the output gamma is noticeably darker than the converted footage output from 5DtoRGB

600x closeup of footage from MPEG Streamclip output to Apple ProRes 4444

Rarevision 5DtoRGB Conversion to Apple ProRes 4444 

Output from Rarevision 5DtoRGB to Apple ProRes 4444

600x enlargement Rarevision 5DtoRGB to Apple ProRes 4444

Rarevision 5DtoRGB Conversion to Apple ProRes 4444 with Add Noise Option

Rarevision 5DtoRGB Conversion to Apple ProRes 4444 with Add Noise option

600x enlargement Rarevision 5DtoRGB to Apple ProRes 4444 with Add Noise option

The first thing I noticed right away is that the output from 5DtoRGB is not as dark or contrasty as the MPEG Streamclip output. The Apple ProRes 4444 footage from 5DtoRGB is more faithful to the gamut of the original footage. 5DtoRGB does allow you to select either a 1.8 or 2.2 output gamma. By the way, you know all those .THM files associated with your Canon footage? Don't throw those away! 5DtoRGB needs those files to get the start timecode values.

One other observation: the conversion process is much slower than what you get out of MPEG Streamclip. Rarevision acknowledges this and says it's by design. You cannot batch the files for processing either. This is what they say about the conversion process:

"5DtoRGB takes a no-compromise approach to quality, ignoring any concerns about speed. 5DtoRGB bypasses QuickTime decoding altogether, works internally at 10 bits and uses double-precision floating point math for its YCbCr to RGB conversion. It also recognizes Canon's full range 8 bit YCbCr values (0-255), avoiding clipping and the resulting loss of picture information. The resulting files are the absolute highest quality you'll ever get out of the camera. In fact, you could argue that they're even better than the camera originals since they've undergone high quality chroma smoothing (which you can disable if you want, but you shouldn't)."

I'd like to hear from other editors and DSLR shooters out there that have used 5DtoRGB. I'll do more extensive testing in the near future. I want to get your opinion on this. So, what do you think? Is 5DtoRGB a better way to convert 5D Mark II or Canon 7D footage?

Update on Thursday, September 2, 2010 at 8:38AM by Carl Olson

Comparisons are a tricky thing. My impromtu and informal "testing" sure triggered a lot of interest that I did not anticipate. In retrospect I wish I had had more time to do more rigorous testing. Ben Richardson makes a good point about people possibly drawing the wrong conclusion from my impromptu testing. Lesson learned!

Fortunately, other pixel peepers (and I mean that affectionately) have taken up the noble task. I imagine we'll see more results come in. For starters, here's one on Cinema5D's forum posted by member 555Productions with a followup test by member 5David.

http://www.cinema5d.com/viewtopic.php?f=24&t=19876&start=0&st=0&sk=t&sd=a

Thanks to Jim Bruno (@jimbruno08) and Brian Dickman (@b_dickman) for the tweet pointing to this interesting Cinema5D forum post.

Certainly 5DtoRGB has raised the bar in HDSLR shooter's quest to squeeze and preserve every milligram of quality out of the flawed footage coming out of these cameras.

Update on Thursday, September 2, 2010 at 3:08PM by Carl Olson

Another interesting test, this time from Robin Schmidt (a.k.a El Skid), with his results and observation in his blog article: 5D To RGB The Follow Up: Bigger Comparison Yes Yes (Prores 422 SQ)