7d vs 5d vs Red Scarlet (Ultimate Question!!)

[Brian Drysdale]

Increased resolution and therefore more detail in spectacular scenes was main reason that people have traditionally shot with 65mm and Vista Vision cameras, it wasn't for the shallow DOF effect. Having to use this effect in order to reduce detail is a limitation that the film maker has to bear in mind, but they're missing out on the major reason why you'd shoot on a larger format camera in the first place.

You have to decide if the current DSLR cameras are appropriate for the story you're trying to tell. Unfortunately, the lower the price of the camera the more compromises are usually built in, especially if it's also aiming at two markets. The RED brain seems to be a major cost factor and that could be the limitation for the stills cameras going for a higher quality film making product.

[Eric Darling]

Honestly, the codec isn't the biggest problem with the camera. I've done perfect keys with the 5D, after converting the H.264 to ProRes. And the bitrate is not low for H.264. As a matter of fact, it's quite high. H.264 is a low bitrate codec by its nature. The problems in the footage generally do not occur because of the compression. Aliasing typically occurs due to the way the cameras (5D Mk II and 7D) "bin" pixels in the downsample from the full image coming off the sensor. That's going to exist no matter what codec the camera uses.

Other issues have often been attributed to using non-UDMA compact flash media. Canon's Tim Smith explicitly said back in July that they have seen this kind of an issue with the 5D in testing. Banding in high detail areas can be attributed to the codec being instructed to drop the rate to accommodate slower cards. The bitrate is variable in the camera, and it will respond accordingly if you are cheaping out on your media. Because I have exclusively used UDMA media, I've never seen such banding or any quality issue that I can't attribute to the binning problem.

[Bill Pryor]

This is sort of "deja vu all over again." When HDV cameras came out everybody said the codec was horrible and you couldn't key it and there were artifacts. Yet people have made feature films with it, keyed it, and shot without noticeable problems. I do it all the time. It's not as easy to work with as some codecs and you can't screw up and expect to fix it with Color--you can tweak it a bit but there are strict limits on how far.

I'm not saying the 7D or 5DMKII are perfect cameras; they have limitations we have to work around, just as HDV does. But work within them and we can get great looking footage. It's not going to be as good as a Red that costs way more, but you can intercut it with Red or most other cameras, if you shoot it well, and not have any problems. One person on this board has even shot Steadicam footage with a 5D that intercuts with 35mm film on a production, and no doubt others have done similir things. There's always a better camera and always one that'll look better, but that doesn't diminish the one you have.

[Benjamin Eckstein]

Very well said. I think sometimes these discussions, while useful knowledge, can be a downer.

[Chris Hurd]

Originally Posted by Bill Pryor

There's always a better camera and always one that'll look better, but that doesn't diminish the one you have.

I should probably take this statement and put it in bold text across the top of every single page on this site. And that "better camera" not only doesn't diminish the one you have, it doesn't diminish the potential of what you can do with the one you have. It doesn't diminish the fact that image quality is affected by human factors more than anything else.

[Joe Carney]

On a more positive note...
With this camera you can do preproduction with the same lenses you will shoot with.
(Site scouting, storyboarding...). David Mullen uses a DSLR and comic book software to do his storyboards.

[Daniel Browning]

Originally Posted by Mayer Chalom

I personally think there is way too much commotion over the scarlet it will probably suck in low light because squeezing 4k into a 2/3 chip will lower its low light sensitivity...

Smaller pixels do not mean worse low light sensitivity. The 7D pixels have over two and half times less area than the 5D2 pixels, yet the performance per area is the exact same. (The reason for this is that Canon scaled read noise with pixel pitch, so that 7D has 1.7 e- read noise and 5D2 has 2.6 e-. After scaling for spatial resolution, both come out to the same 2.6 e- per square micron.)

[Brian Drysdale]

I think that can be a problem with how RED market their products. They announce them at a relatively early stage of their development, which something most manufacturers don't do. Indeed some well known manufacturers are incredibly secretive (and I don't mean Sony), so you only hear about the products when they're about to be sold in the very near future. The only industry that I know works in a similar way to RED is aerospace, when you often hear about products long before they've even flown and then add years before delivery.

The time scale for the Scarlet's development seems to be about normal for these types of product. The best thing is to buy (or rent) what is available when you need it, otherwise you'll be for ever waiting for the next best thing.

[Barry Green]

Originally Posted by Bill Pryor

Somebody said above that H.264 is a "terrible codec." Why is that? What specifically about H.264 is bad?

Hoo boy.

Okay, well, let's start at the beginning -- h.264 is the most advanced codec out there right now. It is the most efficient, operating at 2 to 2.5 times the efficiency of MPEG-2. It's very processor-intensive, it's a very complex codec. According to Sony, 9 megabits of AVCHD matches the visual quality of 25 megabits of MPEG-2 HDV.

But -- before we get into the whole "what's bad about h.264", let's first spell out that there are a bazillion different profiles and levels and implementations of h.264, and they are most decidedly not equal. So you can have AVCHD, which is long-GoP 4:2:0 8-bit h.264, or you can have AVC-Intra, which is intraframe, 4:2:2, 10-bit h.264. Or, you can have the kind of h.264 which is in the 7D, 5D, and GH1, which is the weakest of all the h.264 implementations.

There is a large toolbox of available options for h.264 developers to use, and depending on how many of these tools are employed, that changes the performance of the codec significantly. Hugely. In fact, I will go so far as to say that the very best high-definition video codec out there (AVC-Intra) and the very worst high-definition video codec out there (GH1's AVCHD) are both h.264! The best, and the worst, are both h.264.

So -- the point is, you cannot, should not, and must not lump all h.264 together. You have to know what it is you're talking about when the term h.264 is bandied about. AVC-Intra is the best-quality implementation out there, it's the main codec the BBC chose to use to archive all its footage under their Digital Media Initiative. It's good stuff. And in the middle of the spectrum you've got (what I call) "real" AVCHD, such as is found in the HMC150. That's a long-GoP 8-bit 4:2:0 implementation, with an IBP GoP structure, which performs on par or better than XDCAM-EX.

But then, at the bottom, you've got the half-implemented h.264 versions. Those are what the 5D/7D and the GH1 use. In those versions we're also talking about long-GoP and 8-bit and 4:2:0, but the GoP structure is not IBP, it's IPP. Meaning, there are no B-frames. And that's a major huge quality hit. Much of h.264's power and efficiency come from the B-frames, which allow it to predict a frame based on up to five surrounding frames. But the GH1, 5D & 7D don't have B-frames, they have only an intraframe (I frame) at the beginning of the Group of Pictures, and then all P frames afterwards. So in "proper" AVCHD, your GoP might look like IBBBPBBBPBBBPBBB. Heavily reliance on the B frames is what gives it its incredible efficiency. But in the GH1 or 7D, the GoP looks like IPPPPPPPPPPPPPPPPPP. It's weak. It's handicapped. It's not a "full" implementation.

Now, the GH1 suffers more than the Canons do in this regard because the GH1's maximum bitrate is 17mbps, whereas the 5D goes to about 40mbps and the 7D goes to somewhere around 48mbps. They attempt to make up for the inefficiency by throwing bandwidth at the problem. And it helps, but... I can't help but think that the net result isn't really any better than we would have gotten out of HDV, as far as compression efficiency goes (excepting, of course, that all these flavors of h.264 are recording full-raster 1920x1080).

So -- what about h.264 is "bad"? Not a thing, if it's fully implemented. Full h.264 is the best video codec going. Fully implemented AVCHD is the best long-GoP codec going (in 4:2:0; XDCAM MPEG422 is the best long-GoP codec out there AFAIK). But a bad implementation of AVCHD or h.264, such as the GH1 uses, can cause serious artifacts and loss of detail in the picture. It can be stellar, or it can bite you.

As far as keying goes, the codec is only one part of the equation; what the camera head delivers to the codec is equally or even more important. The SLRs do some weird and funky things when scaling their huge chips down to HD-sized frames, and the 7D and 5D in particular produce a very high level of chroma moire on fine detail. It's the kind of thing that could make a chroma key difficult to pull; it won't always happen, but it'll happen on areas of fine detail with a repeating pattern (so certain fabrics, like maybe a knit shirt or a tweed suit, might cause chroma moire which would seriously impact the ability to pull clean edges).

In the end summation, about h.264 as an acquisition codec -- I don't think long-GoP is the best idea for acquisition; long-GoP is optimized for delivery. Long-GoP works great as a delivery codec, where the compression engineer can optimize the footage and use massive horsepower to run multiple passes and really squeeze amazing performance out of low bandwidth. But as an acquisition codec, you're at the mercy of the power of the camera's built-in encoder. And on a low-cost camera like the 7D, 5D, or GH1, it's not likely that they're going to have the most powerful chips in there, so you might end up with (and, in the case of these three products, you DO end up with) mediocre quality from the h.264 format. Even though h.264 can be the best and most advanced format out there, it all depends on the particular implementation of it.

[Chris Barcellos]

Barry:

I will put that post in my primer notes on this stuff. A great general explanation of a topic that has been all mysterious to me.

So, all that said, why are guys like Phil Bloom and other pros all so happy with the images they are getting out this camera, over our prior camera-- in my case, the the FX1 and HV20-- and others over the HVX200 ? As far as a pleasing image in general, despite the codec's limitations, I think camera can overcome some of the things you mention.

If you "fix" the image into an intermediate format for editing, and shoot at lowest sharpening settings, and keep the image on the flat side, do you not help the situation ?

[Jon Fairhurst]

Barry,

You can't request Intraframe, and bemone the fact that DSLRs don't use B-frames.

The thing about high-end h.264 implementations is that they can have very high compression ratios, but they are computation intensive. DSLRs need to run for hours on a small battery. Using a simple implementation and high bitrate is the most effective solution.

Long GOP is fine for acquisition. But it stinks for editing. So we transcode. This has little to do with quality. It's a workflow thing.

The real limitations are 4:2:0 and 8-bits. Nail those exposures, people!

[Chris Gotzinger]

Originally Posted by Barry Green

(and, in the case of these three products, you DO end up with) mediocre quality from the h.264 format.

"Mediocre quality from the h.264 format" suggests that the footage shows codec-related issues so severe they make the image quality "mediocre".
Please show me a piece of original, ungraded 5D/7D footage (the GH1 is off-topic - see thread title) that shows these codec-related issues.

[Mikko Topponen]

Originally Posted by Barry Green

It's weak. It's handicapped. It's not a "full" implementation.

That doesn't make ANY sense. With a high enough bitrate, those b-frames are pointless. The entire reason to use b-frames is to get the bitrate low enough and for it to still look good. That's why AVC-Intra is only keyframes. It doesn't make sense to praise AVC-Intra as the best codec in existence (It's just the I-frames) and then say that the coding mechanism in the 7D is bad because it doesn't have b-frames.

The less motion estimation there is, the better the image. That's why they don't use b-frames when the bitrate is high enough. That's why AVC-Intra is only keyframes.

[Kain Yun]

Originally Posted by Chris Gotzinger

"Mediocre quality from the h.264 format" suggests that the footage shows codec-related issues so severe they make the image quality "mediocre".
Please show me a piece of original, ungraded 5D/7D footage (the GH1 is off-topic - see thread title) that shows these codec-related issues.

Download the raw video midway down the page:
Canon EOS 7D Hands-on Preview: 14. Samples: Digital Photography Review

Every frame of 5D/7D footage is full of codec related issues.
Whether people have the eye to see it is a different question.

It probably depends on the quality people are used to looking at. Any colorist would call the footage complete garbage from a technical point of view.

It's better than an HV30, but not as good as a prosumer camera like the HVX. But the HVX wont do shallow DOF without an adapter, to there's a trade-off.

[Alex Leith]

Originally Posted by Kain Yun

Every frame of 5D/7D footage is full of codec related issues. Whether people have the eye to see it is a different question.

I agree that the particular clip you linked to is not the greatest piece of cinema, but can you say exactly what the "codec" issues are that you see in every frame of 5D/7D footage?

Are you saying these issues are things that prosumer cameras do not suffer from?

Am I also detecting a Panasonic vs the rest of the world subtext to this thread? I don't know if we're descending dangerously close to an unquantified "favourite brand" mud-slinging match?