C300: from C300 to C300 Mkii or C500 Mkii

To be fair, I only really notice the egregious example so aliasing. Back in the 5D II days, I always had a set of Tiffen old style diffusion filters on hand in order to reduce the fine contrast until aliasing went away. And then I bought Mosaic's filters. After working around that mess, nothing seems that bad.
 
Back in the 5D II days the other issue was lack of 24p. Here's a side by side I did 30p vs 24p (retimed in post):
Canon later added 24p via FW update. Canon keeps playing this game (e.g. RP, 90D).

When the 5D3 was released, I tested it for aliasing- it did really well:

When I saw aliasing with the C300 II, it was a bit surprising and a step backwards. Now they've likely fixed it with the C500 II. This sea-saw pattern of improvements - crippling - improvements appears to be business/sales related and not technical.
 
The 5D III has reduced aliasing because Canon put a softening filter in it to turn it to mush, or it was mush due to line skipping, or whatever. It was basically mush. While the C300 series are way sharper, and if you were to chose to add filtering on set or in post, you could eliminate aliasing while not being forced to shoot mush. It’s like complaining that a car that goes 300 mph is more prone to overheat at its top speed than a car that can only go 120 mph. If overheating is a big concern, just go slower, while still being able to go way faster than 120 mph without overheating.

As it is, I rarely get aliasing from C300 Mark II footage, so it’s not really something I worry about. Are you getting aliasing regularly on actual footage from projects or just from test footage?
 
Eric- did you watch the Bugatti (cover) video I posted- does that 5D3 footage look like mush? Zero aliasing from the camera, even with extremely difficult ground patterns, brick walls, fine fabrics. ML RAW on the 5D3 is pretty decent: the softness came from Canon's crippling before (or during) H.264 compression. While it looked OK post sharpened, indeed it's nowhere near Nyquist limit true 1080p.

The C300 II as you noted didn't alias that often, again only on fine detail (which can be a disaster if the subject is wearing the wrong outfit), however after my experience with the 5D2, 5D3, FS700 (more detail and decent slomo but weak color and it's own image artifacts), I expected the C300 II to be a move forward in all aspects: it wasn't. Before I sold it I was shooting mostly 2K RGB 12-bit and upscaling to 4K and post sharpening. The 1DX II's 4K is more detailed than the C300 II's 4K (it aliases too).

The point about aliasing is that anyone looking to buy a new camera should be aware of why aliasing happens (undersampled sensor and/or OLPF problems). Whatever the camera's stated target resolution, you'll need 1.5x oversampling along with the appropriate OLPF to cut frequencies to prevent aliasing/moire. ARRI gets away with 4.4K => 4K at the expense of less real detail (they're using an OLPF to prevent aliasing). While it meets Netflix's requirements, 6K => 4K cameras can produce full Nyquist limit detail allowing for more flexibility for post work: reframe / cropping etc.
 
jcs said:
Eric- did you watch the Bugatti (cover) video I posted- does that 5D3 footage look like mush? Zero aliasing from the camera, even with extremely difficult ground patterns, brick walls, fine fabrics. ML RAW on the 5D3 is pretty decent: the softness came from Canon's crippling before (or during) H.264 compression. While it looked OK post sharpened, indeed it's nowhere near Nyquist limit true 1080p.
Click to expand...
I'm not seeing that video in this topic so I'm not sure what video you're referring too. I was talking about H.264 5D. The vast majority of people who shot video with the 5D shot in H.264, myself included, so I'm not counting a hacking feature that records in a not very post friendly (such as for weddings) format as something that makes the 5D not mush. 5D H.264 is mush, which is what most people used, ergo, 5D is mush. But yeah, I've seen some sharp and good looking 5D RAW samples before.

Back in the 5D and 7D days I was a prime guy, because I felt I needed to get every bit of sharpness out of the cameras that I could, because they were mush. Then I switched to a C100 which came with a stock 24-105 which I figured I'd never use, but instead found the sharpness and low light ability of the camera resulted in better and sharper images with the often considered lackluster 24-105 than say a 5D Mark III with a 50 f/1.2, so I have since embraced zooms, and the cameras keep getting better. One side of me really likes sharpness, but the other side of me often thinks, "This is sharp enough," especially when you consider the devices that most people are watching content on. Like, is the small increase in sharpness of a high end prime over a high end zoom really worth it? Hollywood uses zooms a fair amount, so apparently they're not too worried about zooms providing a dull image, even if they often prefer primes.

One thing I haven't figured out is why my images have varying degrees of sharpness. I'll be shooting with the Amira and 30-105 say for a short film, and some shots look sharper than other shots. Often all at 800 ISO, similar aperture, in focus...I don't know. It's not camera or lens specific; I notice the same with the C300 Mark II; some interviews may be tack sharp, others not as much so, again, not because it's out of focus. Maybe it just needed more light hitting the sensor? Maybe there was too much harsh light hitting the sensor causing flaring which reduced sharpness? Maybe the flange distance is changing?

Is the C300 Mark II's external RAW much sharper than the internal image? I've often heard that C200 RAW-lite is sharper than C300 Mark II 10 bit, but C300 Mark II RAW is uncompressed unlike RAW-lite, so in theory it should be better than the C200, aside from other factors at play such as the C200 having a different sensor I believe and perhaps different processing. Like the 5D, though, and probably even more so, it seems no one really uses RAW on the C300 Mark II.




I guess my point with aliasing is if a camera has issues that rarely happen, they typically aren't a big concern. Plus, with aliasing you can always do on-set or post softening to eliminate it, so it's not entirely incurable. To me it's kind of like people who complain about most cameras not being global shutter. If they shoot fast moving objects on a regular basis that actually needs global shutter, I can see their point, but if they shoot what most of us (such as myself) shoot, then they're rarely going to notice rolling shutter issues on high end rolling shutter cameras which have pretty fast readout speeds.

On the other hand, issues that I'd be more concerned about are issues that affect all footage, such as h.264 turning a 5D to mush or an FS7's color science affecting the color from all of its footage.
 
Last edited:
Eric- 8 years ago I shot this H.264 5D3 footage which I sharpened in post:


Is this mush? Pretty much zero aliasing on very difficult material (and I was trying to make it alias).

If your subject matter has infrequent fine detail, then the C300 II aliasing might not be much of a concern for you. In 2020 I don't recommend any cameras with this kind of fundamental flaw that should not have been there when the camera launched. IMO the C300 II's color science was sub-par compared to Canon's DSLRs. Sure the FS7 has worse color science, yet that camera did far better in the market- probably because it was less crippled. Even though you don't personally like that camera, more people in the market liked it vs. the C300 II.

Varying apparent sharpness: it's all about contrast: from micro to macro. Was a lens hood / matte box used (flaring is a good guess), was it really in focus, etc. Could be related to sensor temperature and in-camera NR: hotter = more noise = softer after NR.

C300 II RAW vs. C200: don't recall any examples showing the C300 II being any better.

I'm only interested in stills + video cameras at this point- will research carefully the 1DX III and EOS R sequels when they launch. Any major crippling / aliasing = pass.
 
The C300 MKII is sharper in RAW output, but with increased aliasing. It actually has what appears to be bayer pattern induced color fringing in high contrast edges. This happens when you record either prores or prores RAW on an Atomos.

 
jcs said:
Eric- 8 years ago I shot this H.264 5D3 footage which I sharpened in post:


Is this mush? Pretty much zero aliasing on very difficult material (and I was trying to make it alias).

If your subject matter has infrequent fine detail, then the C300 II aliasing might not be much of a concern for you. In 2020 I don't recommend any cameras with this kind of fundamental flaw that should not have been there when the camera launched. IMO the C300 II's color science was sub-par compared to Canon's DSLRs. Sure the FS7 has worse color science, yet that camera did far better in the market- probably because it was less crippled. Even though you don't personally like that camera, more people in the market liked it vs. the C300 II.

Varying apparent sharpness: it's all about contrast: from micro to macro. Was a lens hood / matte box used (flaring is a good guess), was it really in focus, etc. Could be related to sensor temperature and in-camera NR: hotter = more noise = softer after NR.

C300 II RAW vs. C200: don't recall any examples showing the C300 II being any better.

I'm only interested in stills + video cameras at this point- will research carefully the 1DX III and EOS R sequels when they launch. Any major crippling / aliasing = pass.
Click to expand...
Yeah, I'd say particularly the wide shots in that 5D video are very much lacking in detail. I'm also seeing aliasing on the grill of the car.

It's funny how on DVXuser I always hear how the FS7 is so popular, but when I shoot big events, I see way more C-series cameras there from other crews, and I know more C300 owner/ops than FS7 owner/ops. Maybe it's a regional thing. I know Dan has also said the C-series are more popular in L.A. as well. So, I don't know where the FS7 is so popular. It's very popular for reality TV where the production companies often buy their in-house FS7s, which may be a factor in more owner/ops I know having C-series while the companies bought the FS7s.

One of the things I really like about the C-series compared to the FS7 is you can charge more for it. FS7 clients are often lower budget clients; I get an FS7 request, and they often expect lower rates. I find that higher end producers with higher expectations (e.g., not reality TV) often are willing to pay a premium for Canon colors over the FS7.


You're saying that if a camera sensor temperature gets hot the camera applies noise reduction? I've never heard of this.

I'm curious how mattes attached to a matte box affect contrast and sharpness. I have a basic Bright Tangerine Misfit Atom matte box; it only has a top flag, so I wonder if a more featured matte box with side flags would also help with maintaining consistent contrast and sharpness.
 
Eric Coughlin said:
I'm curious how mattes attached to a matte box affect contrast and sharpness. I have a basic Bright Tangerine Misfit Atom matte box; it only has a top flag, so I wonder if a more featured matte box with side flags would also help with maintaining consistent contrast and sharpness.
Click to expand...

If light is hitting the lens(not to be confused with the lens gathering light), the image can be negatively effected(like robbing contrast or creating an ugly, unwanted, distracting artifact). Most modern lenses have excellent coatings and designs to mitigate and deal with some of these bad things, but they can still occur. I almost always run a matte box when shooting sit-downs. My Arri is fairly deep compared to say my BT Misfit, but there are still times when I need the top flag and even side flags. And there are times when we need to set external flagging up, too. Unless you're purposefully trying to create flair, ideally you would have the lens in a black hole, so to speak.
 
Camera sensors get noisier as temps go up- that’s physics. Add noise reduction and detail goes down (lower SNR).
 
You must log in or register to reply here.
Back
Top