As are most things. :thumbsup:
C500: C500 Mark II: Observations from a C200 point of view
- Thread starter chris f
- Start date
As are most things. :thumbsup:
Grug
Veteran
Correct. I wrote up a bit of a blog post about it (that you might find helpful) after conducting a quick test with the Arri LF when that first came out: http://dreamsmiths.com.au/blog
rob norton
Veteran
Great write up Mark.
c500 ii looks awesome for documentary, high end corporate, personal projects or if you’ll always be able to use your camera eg Barry G.
But for a traditional production (TVC, short film, feature) budgeting to shoot full frame, it seems like it’ll be either fx9 hire/owner operator or Alexa mini LF hire? Unless c500 ii owner ops match the fx9 rates?
Me using a camera is one thing (I like c500 ii better from only viewing not actually using) but if I was in charge of a short film I’d definitely go with the fx9 if it meant significant savings or step up to mini LF if the funds were there. This is a ridiculous hypothetical because I’d more likely take the even larger savings shooting on S35 cameras.
Jaime Valles
Veteran
The other advantage of the C500 II over the C200 is greater dynamic range. That's not something to be discounted easily. But it's up to you to determine if it's worth buying a $16,000 camera for a couple of extra stops of DR.
The format differences were more/almost linear with film. In digital, there are far more variables to take into account.
Oddly enough, for Alexa's, it's also more linear since Alexa, Alexa LF and Alexa65 all share the same sensor (with LF having two of them and 65 three). But the rest depends on the processing power, the inate sensor design, codecs, etc..
Oddly enough, for Alexa's, it's also more linear since Alexa, Alexa LF and Alexa65 all share the same sensor (with LF having two of them and 65 three). But the rest depends on the processing power, the inate sensor design, codecs, etc..
I keep hearing this but I've yet to see it actually proven. I have a hunch they're very, very close to each other. They're cut from the same sensor
chris f
Veteran
I don't think there is, or if there is a difference it's pretty minor. Referencing my side by side screen grabs from before, I also ran some tests using all the same settings in raw for both cameras and looking at all the waveform monitors, etc. in Premiere side by side and I'm seeing the exact same thing, maybe a 1% difference but my guess is that was probably just a 1% change in light levels coming from outside the window. Maybe a proper test with charts would show something different, but as far as usable DR I think it's the same.
chris f
Veteran
One thing I've been trying to figure out is would it have been possible for Canon to offer a Super 35mm option within the C500MKII that utilized the whole 6K sensor? In full frame raw it can shoot 6K and in full frame XF-AVC it can shoot 6K that is down sampled in camera to 4K, but when shooting Super 35mm in all modes it's just a 4K image that is created by cropping into the sensor instead of down sampling.
The BMPCC6K camera is a super 35mm crop and also 6K so it seems this is possible, no? I guess I can't wrap my head around what the difference between a 6K full frame sensor and a 6K super 35mm sensor is....
Liam Hall
Veteran
If you are on a budget-conscious short, you can shoot on a DSLR/Mirrorless. Given you can shoot anamorphic in RAW on the C500MkII, and swap out the EF mount for PL, it makes it a good option if you can't quite stretch to an Arri LF. Not sure the FX9 comes into the equation.
Last edited:
Liam Hall
Veteran
You are confusing resolution with format. The BMC is S35 not FF. It can record in 6K or it can downsample to 4K. The C500 MKII does the same trick, but with larger photosites, you can shoot RAW 5.9K using the whole sensor or you can downsample to 4K in XF-AVC using the whole sensor. You can of course, choose to shoot S35 on the C500 MKII.
chris f
Veteran
Copy that. But why can't the C500MKII do 6K or down sample to 4K in super 35mm? Is that a limitation of Canon's processor or an internal choice? or is it physically impossible to do in the camera?
NorBro
Major Contributor
Companies can do whatever they want. Some do down-sampling really well (Blackmagic) and some do not (ones not to be mentioned).
The cameras are in their control and they can provide us the purest, highest image quality possible from any of them if they wanted to, but they have to make choices.
The cameras are in their control and they can provide us the purest, highest image quality possible from any of them if they wanted to, but they have to make choices.
chris f
Veteran
So it's one or the other? No such thing as a 6K scan from a super 35mm crop off a larger FF sensor?
Liam Hall
Veteran
Think about it, if the FF sensor is 6K, how can an S35 crop also be 6K?
It'd have to offer more pixels in full frame in order to have a crop 4K or 6K. So, if 6K is defined (roughly speaking) as 17.6 MPX, then the full frame would have to be 1.5 higher (26.4). Technically, it can be done (Venice is ~ 24) but it will probably have to wait a few years for the proper implementation.
It's physically and mathematically possible to downsample any resolution to any resolution. There are many ways it can be implemented: https://en.wikipedia.org/wiki/Image_scaling
CamCabal cameras are generally crippled and/or built with the cheapest tech to maximize profit, e.g. using really old larger fab tech when making chips. Currently only phone makers can afford the best fab tech (e.g. 5nm fab process). Sony tends to release the more advanced tech (e.g. FX9 producing superior IQ vs. the C500 II while missing other features such as no touch AF, menu/control/usability issues etc.) however there are rumors that Canon may leapfrog Sony in the mirrorless market (EOS R5 etc.). Which would mean Canon is using more advanced (smaller) fab tech than Sony: more powerful processing & features while also using less power. The 1DX III is definitely using smaller fab tech than the prior gen.
Using separable function downsampling, meaning two passes, horizontal followed by vertical, RAM buffers wouldn't need to be very large, and if a system can do 8K => 4K (simple box filtering or averaging would work there), or 6K => 4K, any subregion of the image sensor can be downsampled to any desired resolution. Mathematically it's relatively easy (in the realm of computer imaging). If the hardware is fixed function to reduce cost, then that could make arbitrary sampling/filtering not possible for that specific implementation. If an implementation can sample the entire sensor fast enough (and/or windowed sub-regions), and store the result to a RAM buffer, then arbitrary resampling can be done, just like we can do in post or with Photoshop etc. Smaller sub regions won't be able to get significant resampling to help reduce aliasing and/or maintain high quality 4K, though sometimes this is good enough re: slomo. A work around in post would be to try AI upsampling, which is getting better and better with lots of competition. So with a bit of post work 6K => 4K (in camera) => 6K upsample (or even 8K) is possible.
Best to review each piece of gear and/or test/rent and make sure it does what you need, as we've learned there are rarely firmware updates which provide new features (on the high end, this can happen, e.g. more frame rates).
CamCabal cameras are generally crippled and/or built with the cheapest tech to maximize profit, e.g. using really old larger fab tech when making chips. Currently only phone makers can afford the best fab tech (e.g. 5nm fab process). Sony tends to release the more advanced tech (e.g. FX9 producing superior IQ vs. the C500 II while missing other features such as no touch AF, menu/control/usability issues etc.) however there are rumors that Canon may leapfrog Sony in the mirrorless market (EOS R5 etc.). Which would mean Canon is using more advanced (smaller) fab tech than Sony: more powerful processing & features while also using less power. The 1DX III is definitely using smaller fab tech than the prior gen.
Using separable function downsampling, meaning two passes, horizontal followed by vertical, RAM buffers wouldn't need to be very large, and if a system can do 8K => 4K (simple box filtering or averaging would work there), or 6K => 4K, any subregion of the image sensor can be downsampled to any desired resolution. Mathematically it's relatively easy (in the realm of computer imaging). If the hardware is fixed function to reduce cost, then that could make arbitrary sampling/filtering not possible for that specific implementation. If an implementation can sample the entire sensor fast enough (and/or windowed sub-regions), and store the result to a RAM buffer, then arbitrary resampling can be done, just like we can do in post or with Photoshop etc. Smaller sub regions won't be able to get significant resampling to help reduce aliasing and/or maintain high quality 4K, though sometimes this is good enough re: slomo. A work around in post would be to try AI upsampling, which is getting better and better with lots of competition. So with a bit of post work 6K => 4K (in camera) => 6K upsample (or even 8K) is possible.
Best to review each piece of gear and/or test/rent and make sure it does what you need, as we've learned there are rarely firmware updates which provide new features (on the high end, this can happen, e.g. more frame rates).
Last edited:
chris f
Veteran
Yes.
Only with post processing upsampling (with AI). This will be possible in future cameras. Not possible yet: downsampling goes down in resolution, using a math thing called a filter (just a bunch of multiplies and additions). A FF 6 or 8K sensor that crops to a 35mm region will be less than the full sensor in resolution. However, AI upsampling can help with getting back to 6 or 8K while also reducing aliasing. Amazing tech is coming in the near future.