C200: What stops the C200 from being a near perfect camera?

Perfect is a strong word. How about really, really great? When I write 'you' below, I'm targeting a generic reader. Not directing this at anyone special.

When I look at this camera, I only look at the internal 12bit raw. Why on earth would you look at anything else? It's actually pretty much the ultimate codec. The data rates fluctuate around ProRes 4444 equivalence. It's 4x-5x compression. ProRes is actually a larger file at around 3.11x compression for the 12bit 4444.

Had it been ProRes you would get slightly larger files, but be able to go straight to post. With Raw Light you'll get some minor benefits of being able to adjust white balance and whatever else Canon will allow, but you'll probably want to do a transcode to a ProRes master for post. But it's a reasonable trade-off.

This camera is for people who will appreciate that stand out feature, together with all the other Cinema EOS line features we've grown accustomed to by this point. This is a camera that combines ultimate usability* with highest end image fidelity. If someone looks at this camera and can't see that, I know I can't help them.

If the 10bit XFAVC codec was something 'you' needed—please don't bang your head in the wall over this camera. Canon has made several business decisions that makes sense to them, in order for them to launch the C200. That said, I consider this an aggressive launch that 'includes' more than it 'excludes'.

Just be happy about that. Be very happy about that.


* = for some, the C700 represent ultimate usability. For others the Amira might be that. For other still, it might be a RED or a Sony or a BMD. It's the whole point. Figure out what YOU need and don't bother with the rest.
 
Perfect is a strong word...
* = for some, the C700 represent ultimate usability. For others the Amira might be that. For other still, it might be a RED or a Sony or a BMD. It's the whole point. Figure out what YOU need and don't bother with the rest.

100%! this is for those Canon users who want to jump into raw on suitable projects, I know I always wanted to go Red for some of my projects but the C300 MKII has been indispensable and I love all the benefits of low light and built in Nd's etc.

The C200 would be a good A or B cam to the C300 mkII, the best of both or go C700 if you need it all, so many options out there and with Red Epic Dragons hovering around $15K on reduser and eBay, there are some very affordable and amazing tools on the market right now...
 
There's a lot of talk about the C200 being only 8bit when recording internal MP4.
I think if you're absolutely sure you need 10bit and that 8bit is detrimental to the type of work you earn your cash with - then just buy a C300mII
if you want a Canon.
Because your absolutely sure about the 10bit, you must be absolutely sure to earn back the couple grands difference in no time.
So, sure bet.

Everyone else could approach the issue like this:
C100 and C300 are 8bit and widely accepted. Yes the C200 is 4K, hence the mp4 has 4 times the C100mII's max data rate.
(Again, for a couple of grands more, you can buy the C300mII, if you need/want even more.)
So, apart form having a different sensor (from the C700) we can expect a similar pleasing picture from the C200, like from the older Canon Cs.
So, most people (again, not those "I'm absolutely positive I need 10bit or I'll end up under the bridge and I don't mean the RHCP music vid" people) should be fine.

THEN add internal RAW with a mild compression, DPAF with a touchscreen, a probably better AVCHD codec for free in 2018 and last not least a sub 8K price tag
Near perfect at least for me.
 
why only 2k?? when converting 4k rawlite you'll get 4k 4:4:4

That's not possible on a single sensor. The C200 has a single 4K Bayer-pattern sensor. As such, it could down convert to make 2K 4:4:4, but in 4K it would be equivalent to about 4:2:0 at the most.

(This is an oversimplification of course; we had the extensive discussion when the Red One first came out).
 
Perfect is a strong word. How about really, really great? When I write 'you' below, I'm targeting a generic reader. Not directing this at anyone special.

When I look at this camera, I only look at the internal 12bit raw. Why on earth would you look at anything else? It's actually pretty much the ultimate codec. The data rates fluctuate around ProRes 4444 equivalence. It's 4x-5x compression. ProRes is actually a larger file at around 3.11x compression for the 12bit 4444.

Had it been ProRes you would get slightly larger files, but be able to go straight to post. With Raw Light you'll get some minor benefits of being able to adjust white balance and whatever else Canon will allow, but you'll probably want to do a transcode to a ProRes master for post. But it's a reasonable trade-off.

This camera is for people who will appreciate that stand out feature, together with all the other Cinema EOS line features we've grown accustomed to by this point. This is a camera that combines ultimate usability* with highest end image fidelity. If someone looks at this camera and can't see that, I know I can't help them.

If the 10bit XFAVC codec was something 'you' needed—please don't bang your head in the wall over this camera. Canon has made several business decisions that makes sense to them, in order for them to launch the C200. That said, I consider this an aggressive launch that 'includes' more than it 'excludes'.

Just be happy about that. Be very happy about that.


* = for some, the C700 represent ultimate usability. For others the Amira might be that. For other still, it might be a RED or a Sony or a BMD. It's the whole point. Figure out what YOU need and don't bother with the rest.

I agree completely with this. 10-bit 4:2:2 XFAVC would have made the camera near perfect, but I'm more interested in shooting 4K raw for narrative films and shooting 4K 8-bit 4:2:0 MP4 for fast turnaround clients with no color grading requirements (which is most of what I do). I guess my only wish is that it had dual CFast slots in addition to the dual SD slots. But I can live with it as is. I'm saving up for it now, and hopefully by the end of the year I'll get one.
 
That's not possible on a single sensor. The C200 has a single 4K Bayer-pattern sensor. As such, it could down convert to make 2K 4:4:4, but in 4K it would be equivalent to about 4:2:0 at the most.

(This is an oversimplification of course)

Hi Barry,

Even with your 'catch all' in parenthesis, I think your post is a bit unfortunate.

"4k 12bit 444" is possible. It's simply the technical specification of how the data is acquired on the sensor level. Indeed, each of the pixels, or photosites, will only represent one of the RGB colors according to the Bayer pattern, but that is generally what we're all referring back to when discussing sensors here. On a forum like this when people talk loosely about 444, 422 or 420, I think the Bayer pattern is implied.

I know you're talking about 'net fidelity', but I think to a lot of people it will be confusing to read that 4k 12bit 444 is equivalent to 420. The knowledge you assume for someone to understand you correctly, and fill in the blanks you leave out in an effort to 'simplify', is often lacking.

Homer Kelly, the author of 'The Golfing Machine' wrote: "asking for something complex to be 'made simple', doesn't make it simple – only incomplete".

Anyway, essentially it comes down to this: when opting for RawLight on the C200 you get equal image fidelity with your entry level Canon Cinema camera as you would with a number of high end, often many times more expensive camera systems. All this in a package that already has a tidy power solution and lots of assist features. And what I really like about it is that I could remove everything on the top and just use the optical viewfinder and essentially pass as a 'nobody' with a somewhat large, but still "normal" camera.

I'm just a bit surprised that so much attention is paid to a feature of the camera (internal LongGOP) that isn't very exciting, where there ARE exciting things to talk about.
 
I think there are two camps..

those that think that 444 or raw are the minimum to make good work.. especially in non controlled non set environments

(often these people transition to motion from raw stills photography)

and those that actually think the current generations of cameras actually deliver something that is a joy in post.

Mix in of course the data/media costs and that most editors are technically behind the curve and you have complex views on this camera.

Personally it seems the most exciting sub 10k camera ever.. but also a big sell to my clients who are, unlike me, happy with FS7 files.

S
 
Hi Barry,

Even with your 'catch all' in parenthesis, I think your post is a bit unfortunate.

"4k 12bit 444" is possible. It's simply the technical specification of how the data is acquired on the sensor level. Indeed, each of the pixels, or photosites, will only represent one of the RGB colors according to the Bayer pattern, but that is generally what we're all referring back to when discussing sensors here. On a forum like this when people talk loosely about 444, 422 or 420, I think the Bayer pattern is implied.

I know you're talking about 'net fidelity', but I think to a lot of people it will be confusing to read that 4k 12bit 444 is equivalent to 420. The knowledge you assume for someone to understand you correctly, and fill in the blanks you leave out in an effort to 'simplify', is often lacking.

Homer Kelly, the author of 'The Golfing Machine' wrote: "asking for something complex to be 'made simple', doesn't make it simple – only incomplete".

Anyway, essentially it comes down to this: when opting for RawLight on the C200 you get equal image fidelity with your entry level Canon Cinema camera as you would with a number of high end, often many times more expensive camera systems. All this in a package that already has a tidy power solution and lots of assist features. And what I really like about it is that I could remove everything on the top and just use the optical viewfinder and essentially pass as a 'nobody' with a somewhat large, but still "normal" camera.

I'm just a bit surprised that so much attention is paid to a feature of the camera (internal LongGOP) that isn't very exciting, where there ARE exciting things to talk about.

Reminds me of all the negativity that I had seen in these forums when the original C300 was launched. At that time one camera was touted to be the C300 killer. Now also, one competitor has announced a camera whose specs are still being tweaked. Historically people have a way of hating Canon in the internet boards. Guess it will continue for sometime.

Morgan is right. I come from still photography background. With the introduction of DSLRs many people were hesitant to buy CF cards and wanted to record the still shots in jpeg. I remember buying a 1GB card for 250 US dollars in those days. Soon everybody and their nanny started shooting in raw.

The future is RAW. RED was way ahead of the curve in introducing RAW in their cameras. Now with the C200 the RAW floodgates for the unwashed masses will open. :)
 
First and foremost, the RAW workflow really needs to be tested.

The available footage from almost everyone looks like typical, eh, Canon footage...while the Cinema5D RAW video samples look amazing to me.

5d is fullframe so it will always have a huge advantage in look compared to 1.5x crop sized sensors.
 
I think there are two camps..

those that think that 444 or raw are the minimum to make good work.. especially in non controlled non set environments

(often these people transition to motion from raw stills photography)

and those that actually think the current generations of cameras actually deliver something that is a joy in post.

Mix in of course the data/media costs and that most editors are technically behind the curve and you have complex views on this camera.

Personally it seems the most exciting sub 10k camera ever.. but also a big sell to my clients who are, unlike me, happy with FS7 files.

S

Real 4:4:4 camp comes from laser film scanners and days when you had to be really really smart and talented to work on ~ $1M finishing systems.

The second camp (ex. photographers included) must really enjoy those current generation of cameras with halved horizontal chroma resolution and intra frame h.264 compression in post.
In fact, they are enjoying it so much that they are spending 20x more time then what it would take with 4:4:4 footage doing the same task.

4:2:2 has little to no visual difference, but the difference is significant in image analysis.
 
Real 4:4:4 camp comes from laser film scanners and days when you had to be really really smart and talented to work on ~ $1M finishing systems.

The second camp (ex. photographers included) must really enjoy those current generation of cameras with halved horizontal chroma resolution and intra frame h.264 compression in post.
In fact, they are enjoying it so much that they are spending 20x more time then what it would take with 4:4:4 footage doing the same task.

4:2:2 has little to no visual difference, but the difference is significant in image analysis.

Sorry you lost me there. Im not a tech guy, but my understanding is that 444 or raw are full colour 12 (or 14 or 16 for raw) files, 444 being as good as 12bit raw but 50% less efficient with in camera data.

The basic point is that some struggle with compressed codecs from fs7/c300 and will see the raw as good news, other see the fs7/c300 as adequate and will be non plussed to eat data on set and be forced to buy expensive media.

S
 
5d is fullframe so it will always have a huge advantage in look compared to 1.5x crop sized sensors.

lol. Talking about sample footage from the C200. Cinema5D is a website.

What I said had nothing to do with the 5D camera or FF sensors.
 
Sorry you lost me there. Im not a tech guy, but my understanding is that 444 or raw are full colour 12 (or 14 or 16 for raw) files, 444 being as good as 12bit raw but 50% less efficient with in camera data.

The basic point is that some struggle with compressed codecs from fs7/c300 and will see the raw as good news, other see the fs7/c300 as adequate and will be non plussed to eat data on set and be forced to buy expensive media.

S

What do you mean by struggle? Image manipulation isn't just colour correction.

If it would be just that, here is an example what you can do with poor HDV 1440x1080 4:2:0 taken with Canon HV series (20 or 30?):


 
"4k 12bit 444" is possible. It's simply the technical specification of how the data is acquired on the sensor level.

I'm having trouble understanding your post. 4:4:4, 4:2:2, and 4:2:0 mean something very specific.

They aren't "how the data is acquired on the sensor." They are how the signal is recorded, after all processing is done --- literally the opposite end of the whole process. The sensor is the beginning. The recording is the end.

Indeed, each of the pixels, or photosites, will only represent one of the RGB colors according to the Bayer pattern, but that is generally what we're all referring back to when discussing sensors here. On a forum like this when people talk loosely about 444, 422 or 420, I think the Bayer pattern is implied.

What does this part of your comment mean? I can't decipher it. It sounds like you are saying, "On a forum like this, people bat around 4:4:4, 4:2:2, 4:2:0, and Bayer, and we mean roughly the same thing by all four terms."
 
There's a lot of talk about the C200 being only 8bit when recording internal MP4.
I think if you're absolutely sure you need 10bit and that 8bit is detrimental to the type of work you earn your cash with - then just buy a C300mII
if you want a Canon.
Because your absolutely sure about the 10bit, you must be absolutely sure to earn back the couple grands difference in no time.
So, sure bet.

Everyone else could approach the issue like this:
C100 and C300 are 8bit and widely accepted. Yes the C200 is 4K, hence the mp4 has 4 times the C100mII's max data rate.
(Again, for a couple of grands more, you can buy the C300mII, if you need/want even more.)
So, apart form having a different sensor (from the C700) we can expect a similar pleasing picture from the C200, like from the older Canon Cs.
So, most people (again, not those "I'm absolutely positive I need 10bit or I'll end up under the bridge and I don't mean the RHCP music vid" people) should be fine.

THEN add internal RAW with a mild compression, DPAF with a touchscreen, a probably better AVCHD codec for free in 2018 and last not least a sub 8K price tag
Near perfect at least for me.

This is about the most sensible thing I've ever read on this board. You're right; if you absolutely need 10-bit you can easily make the extra few grand to buy a C300/II
 
This is about the most sensible thing I've ever read on this board. You're right; if you absolutely need 10-bit you can easily make the extra few grand to buy a C300/II

Or just get an external recorder Atomos/7Q and you have 10bit 4:2:2 out of the SDI...
 
I'm having trouble understanding your post. 4:4:4, 4:2:2, and 4:2:0 mean something very specific.

They aren't "how the data is acquired on the sensor." They are how the signal is recorded, after all processing is done --- literally the opposite end of the whole process. The sensor is the beginning. The recording is the end.




What does this part of your comment mean? I can't decipher it. It sounds like you are saying, "On a forum like this, people bat around 4:4:4, 4:2:2, 4:2:0, and Bayer, and we mean roughly the same thing by all four terms."


CIE XYZ raw capture from a Bayer sensor is not straight red pixel, blue pixel, green pixel spatial sampling. It is a 4 vector yrgb interpolation red samples over full resolution luma, blue samples over full resolution luma, and green samples over full resolution luma. If I shoot a standard resolution chart with my D16 there is no visible difference in RGB channel resolutions when debayered to a 16 bit tiff. If I take the tiff file and encode that to 4:2:0 video, the red and blue channel spatial resolution are halved both horizontally and vertically. 4:2:2 halves resolution horizontally but not vertically. There is absolutely no direct correlation between Bayer sampling and any form of encoded video sub sampling regarding chroma resolution or color gamut.
So yes you can get full 4:4:4 4K color encoding from a 4K Bayer sensor. Depending on the quality of the Debayer algorithm there will be minor artifacts, but loss of color gamut or chroma resolution are not among them. Those are limited by the choice of video codec and degree of compression. Not sure how lossy raw compression would affect this.
 
My point is that when you have unknown variables - "interpolation, halving resolutions, minor artifacts, lossy compression", there is a room for error.
When there is an error, it can be corrected but an intervention must be done, usually by hand. Interventions take time.
And human time is the most expensive resource, otherwise we wouldn't "strugle" to automate all aspects of our lives.

(For those who can't read between the lines, storage and infrastructure are many many times cheaper then constant interventions in workflow that has inadequate pipeline.)
 
Hi,

I'm having trouble understanding your post. 4:4:4, 4:2:2, and 4:2:0 mean something very specific.

I can see where some of the confusion comes from. I wrote my post quite clumsily. It's like two trains of thought got merged into one sentence. What I set out to say is actually exactly what you say: 444, 422 and 420 are just technical specifications that say something explicit about the encoding (chroma subsampling). And yes of course: the chroma subsampling is not on the sensor level. That sentence came partly from me thinking about stuff related to the Bayer pattern.

So when Canon says the C200 captures 4k 12bit 444—that is exactly what it does. There is no need to slander Canon in this instance by saying "in reality it's equivalent to 420, since it's a Bayer sensor".

This whole 'equivalence' discussion is a much lager topic and more often than not it's centered around resolution: to actually truly resolve 4k, a sensor needs to be of higher resolution in order to effectively over sample the inherent resolution loss that will happen in the debayer process. A subsection of this topic discusses if 4k 422 can be downscaled to 2k 444 and so on…

In this thread, what happened was a few posts kept feeding off of each other. I actually think the original comment was concerning the weaker internal Long GOP codec and that, even if it's weak, it can be put to effective use by the way of downscaling from UHD to HD and increasing quality that way (for HD delivery). From that, I think there was a carryover to the Raw Light codec and now I'm not quite sure where we all land. It's not important.


What does this part of your comment mean? I can't decipher it. It sounds like you are saying, "On a forum like this, people bat around 4:4:4, 4:2:2, 4:2:0, and Bayer, and we mean roughly the same thing by all four terms."

Here I mean to say: when we talk about chroma subsampling and use terms like 444, 422 and 420, we all mean the signal that is captured in a Bayer pattern. We normally don't refer to a "true net RGB signal equivalence", as if every photo site was full RGB (like Foveon). So again, I think we're on the same page here...

Short version:

When we talk about camera resolution, we normally refer to captured pixels, even if it's "just" a Bayer pattern. Same with chroma subsampling. We all mean the de facto spec the camera manufacturer lists and not a theoretical, true net value.

Hmm.... I'm not even sure this post makes more sense. But honestly... it doesn't matter - it's all good. I'll make my own decision and let others make theirs. =)
 
Back
Top