Visual comparison between the 7D, 5D, RED ONE

[Ryan Orr]

IMO, the Red One will always win with quality of footage, simply due to the codec that is used. Files are created RAW…which means virtually no compression, and extra data/info that let’s post production have control over the image dramatically. Plus it’s 4K…not 1080p

Yes, technically the MK2 had a much larger sensor, but it will never give you RAW data/footage. The H.264 codec is nice, and the camera still gives very nice images even at 1080p size, but it doesn’t compare to the Red One.

The 7D is nice, and it can do over/under crank in post (plus it’s sensor is very near in size to the Red One), but still, won’t compete.

Kinda disappointed in people when they try to compare these cameras. Hands down, the Red One will win every time in my books…IF YOU CAN AFFORD IT. But simply, all 3 are just tools in your arsenal. Get what you can afford and do your job!

If you shoot crap, no matter the resolution or DOF, or any specialness you may try to apply, you are just polishing a turd 🙂

[Noah Kadner]

Also it depends where you’re comparing. Looking at something on Youtube or Vimeo makes the quality comparisons a lot harder to gauge. See it projected on a 60 foot screen and all those differences are shockingly apparent. You get what you pay for with the RED. But of course if you’re going out to DVD, online or even over the air HD, you can get away with a *lot* with the lowered codec quality and frame resolution of the Canons.

Noah

Check out my book: RED: The Ultimate Guide to Using the Revolutionary Camera!
Unlock the secrets of 24p, HD and Final Cut Studio with Call Box Training. Featuring the Sony EX1 Guidebook, Panasonic HVX200, Canon EOS 5D Mark II and Canon 7D.
Watch Formosa- My indie movie shot with the SDX900 and finished with Final Cut Studio.

[Robbie Carman]

[Ryan Orr] “which means virtually no compression,”

This is simply untrue. RED footage is hugely compressed-big time! Because its wavelet based it gives it several advantages in quality compared to codecs like H.264 and flexibility like extracting proxies and things like like being able to support RAW data RGB data from the same scheme but it is still lossy. In fact I think RED themselves saw its visually lossless but mathematically there is still loss.

Don’t get me wrong I think the brilliance of the RED system is all in REDCODE. Sure the bodies and gigantic sensors are cool but REDCODE is very neat indeed – but it is compressed.

Robbie Carman
—————-
Colorist and Author
Check out my new Books:
Video Made on a Mac
Apple Pro Training Series DVDSP
From Still To Motion

[Noah Kadner]

I wouldn’t characterize it as ‘hugely’ compressed. In terms of file size compared to say 4K uncompressed- sure the size is a tiny fraction. But we’re talking about a very efficient and totally optimized codec. It’s far better than just about any other compression codec in use in cameras today.

Personally I care less about the file size as I do about the resulting image quality. REDcode is a nice, clean 4K signal that’s visually close to a nice 4K film scan. Only with a lot less of the attendant expense and time to acquire it.

Noah

Check out my book: RED: The Ultimate Guide to Using the Revolutionary Camera!
Unlock the secrets of 24p, HD and Final Cut Studio with Call Box Training. Featuring the Sony EX1 Guidebook, Panasonic HVX200, Canon EOS 5D Mark II and Canon 7D.
Watch Formosa- My indie movie shot with the SDX900 and finished with Final Cut Studio.

[Ryan Orr]

Okay, in the spirit of the subject of this thread (comparing the Red One to the 5D and 7D), the Red camera is virtually uncompressed compared to the H.264 codec.

I’m not 100% sure, but when rating a codec lossy or lossless, is it in the terms of mathematical data, or picture quality? If it’s picture quality, I consider it virtually lossless/uncompressed. If it’s in mathematical terms, then yes, it is lossy/compressed.

Maybe it would have been best to have said visually instead of virtually to avoid the confusion?

[Robbie Carman]

Don’t get me wrong Noah and Ryan I agree with both of you 100% about the quality of REDCODE – 100%. It’s just that I see so many people talking about RED RAW saying that its the same thing as RAW still photography – and its not that was really my point.

Robbie Carman
—————-
Colorist and Author
Check out my new Books:
Video Made on a Mac
Apple Pro Training Series DVDSP
From Still To Motion

[Noah Kadner]

Yeah I’m with you- it’s RAW in terms of giving you the ability to manipulate the original sensor data i.e. not baked in gamma/color like a JPEG based or MPEG based compressor has. But it’s not RAW in the sense of being without compression. I would call it visually lossless/lossy depending on how well exposed it is.

Noah

Check out my book: RED: The Ultimate Guide to Using the Revolutionary Camera!
Unlock the secrets of 24p, HD and Final Cut Studio with Call Box Training. Featuring the Sony EX1 Guidebook, Panasonic HVX200, Canon EOS 5D Mark II and Canon 7D.
Watch Formosa- My indie movie shot with the SDX900 and finished with Final Cut Studio.

[Neil Abeynayake]

Ryan,

The primary role of a codec is to take large data sets presented to it in a logical array, deduce and compile that data to be used by an end user more economically. The RED One uses codec by the name “RED code”.

(I quote);

Redcode RAW is a variable bit rate wavelet codec which allows raw sensor data at resolutions of up to 4096 x 2304 to be compressed sufficiently for practical on-camera recording. Two variants were offered previously, one with a maximum data rate of 28 MB/s (224 megabits), and one with a maximum data rate of 36 MB/s (288 megabits), but the camera was recently updated to record with an additional data rate option of 42 MB/s (336 megabits). Compared with the uncompressed data captured by the sensor, these bit rates represent compression ratios of about 12:1 and 9:1, respectively. Because Redcode is a wavelet codec, similar to CineForm RAW and JPEG2000, the blocking artifacts associated with other digital video compression algorithms are absent. However, sample images detailing Redcode specific artifacts have been posted on the manufacturer’s forum.

Redcode is a mathematically lossy codec, meaning that decompression does not fully restore the original image data captured by the camera. Red does claim the codec is “visually lossless”, meaning that the information loss is not visible to the naked eye when images are viewed; however this is a subjective statement.

Unlike cameras that record RGB data, the camera records raw data similar to the Dalsa Origin and Silicon Imaging SI-2K. Recording raw data allows white balance, gamma and other image processing parameters like sharpening to be set during post production. Adjusting these settings directly on camera does not impact the raw data that is actually recorded. Such adjustments only influence live monitoring outputs, but are attached to the recorded data as metadata.

(End quote)

Recently, I had an interesting conversation with a Canon Engineer who is involved with the current incarnation of EOS DSLR line. His and Canon’s view about the DSLR’s could be summarize as; The DSLR form factor is designed to complement established ENG trade. However, the camera (is designed and) could perform much more than just ENG work (Yes, we know this!).

The Canon 7D’s image sensor outputs 5,184 x 3,456 pixel (RAW and Large JPEG) images. As you know, it can capture 8 frames a second. This is happening at 12% of the capacity of its dual DIGIC 4 processors. In reality those processors can handle over 40 frames a second. There is couple of drawbacks to this however. The main issue at hand is the enclosure. Because of the form factor of the DSLR, there are restrictions in dissipating heat. This could lead to serious over-heating of onboard electronics and associated parts. The second is the ingest of RAW data by transport” media (CF cards). According to Canon 7D owner’s manual, the camera can output a RAW image using 17.90 (Megapixels) which corresponds to about 25.1 Megabytes + 17.90 Megapixels JPEG image @ 6.6 Gigabytes, 8 times every second. This corresponds to a data transfer around 253.6 MB/s every second.

Now if you want to capture at 5K resolution for digital cinematography (according to the Canon Engineer, theoretically the 7D can resolve at 24 frames @ 5K), get just the RAW data file (i.e. 25 MB/frame) and multiply that by 24 frames a second = 600 MBytes/second. Not very many flash memory cards available today can handle that kind of ingests unless, they are custom made for that camera.

Also, you would have all kinds of work-flow related problems within Canon’s current user base. Therefore, Canon uses a h264 based codec with a QuickTime wrapper (at least for now [they may add other PC based file options later because half of the Windows based clientele is pretty upset about the current QT only ingest]).

We work with a couple of RED One’s and use Assimilate Scratch work-flow. I agree about the aesthetic look of RED output. We also, work with 1D, 5D and 7D footage as well.

About “RAWworks” comparison of 5D and 7D to RED, the company and the people who are associated with them are well qualified to do so. Some may not like it but, there is a lot of merit in what they have done.

“Always remember that you’re unique. Just like everyone else”.

Confucius

[Uli Plank]

While I second everything that was said about the high quality of the codec and the advantages of having more latitude for color grading with the RED One, let’s not forget another problem of the current DSLRs.

As confirmed by the Canon engineer, they can’t handle full rez at motion frame rates. So they are ‘binning’ (sampling down groups of pixels before further processing the image). This introduces another big problem: aliasing! The OLFP in the camera is optimized for photography – as it should, or all photos from the Canon’s would be soft, but they are tack sharp with good lenses.

So it can’t be optimized for 1920 x 1080 at the same time. This is why you get all kinds of aliasing effects, moiré and the like. Shoot a rez chart with the Canon and you’ll see that it resolves around 720, all the rest is fake detail produced by stepping over the Nyquist/Shannon border…

Now I admit that the shallow focus will help you keeping regular patterns (like brick, fences or the like) out of focus most of the time. But if the person in focus has the wrong clothes, you’ll the horrible aliasing all over critical patterns. Been there, bitten by that, since you don’t see that when shooting, Plus, you don’t see it on the web. BTW, the Canon was initially seen as a tool for web jounalism by Canon.

The OLPF of the Red One is optimized for 4K – which they scan and record. Smaller sizes are produced by windowing, not binning, so the OLPF is still right.

Uli Plank

Director of the Institute of Media Research (IMF) at Braunschweig University of Arts

[Joel Mielle]

Just the fact that an under $2,000 camera body (7D) can be compared to a red, leaves little room for argument. Just go out and shoot, the average punter won’t even know the difference even on a big screen.
Again, buy what you can afford. These are exciting times!