C100: When you don't need an external recorder and when you do

[RZE]

I've seen some mis information in posts with regard to whether you need an external recorder with the C100. I've seen people say that there is no difference in picture quality. If you are trying to get the most out of this camera picture quality is not the issue. Image fidelity is much more important. The 8 bit structure needs a decent amount of post manipulation to use the cameras best gamma clog. The bitrate is the determining factor whether that manipulation breaks the codec or not. Showing comparisons between eos standard avchd and prores hq is meaningless. Of course when you crush the s*%t out of the blacks what good is recording more information. Its when you crush everything into the mid tones in clog, aka get the most out of the camera, that is where the difference is crystal clear.

Depending on the way you intend to use the c100 dictates whether it is trivial or whether it is a necessity to record externally. If you do not grade your footage, if you are going to use either cine gammas or normal gammas then it is likely that you don't need it. If your project has a budget or a client, if you are shooting a narrative, or if you are using this camera in clog mode and you intend on grading it then it is a necessity. I uploaded a still shot in clog to the internal sd card. All I did was normalize the image and its already collapsing. There is macro blocking throughout, green artifacts, and banding. I could not present this image to a client. The bitrate is far too small for the remapping necessary to normalize blacks and highlights in clog. Also bring the jpeg into any CC software and try to pull a key or make a secondaries correction then you will see how bad the macro blocking is.

 

[LeavingTheCandy]

Some appropriate lighting would be a good start.

 

[AndreeOnline]

I've seen some mis information in posts with regard to whether you need an external recorder with the C100.

Can you detail some of the misinformation you've seen?

Also, can you provide identical frames from internal/external recording, as this is the only way to see the differences.

By only providing a single frame it is impossible to exclude what can potentially be a large number of factors that can result in poor looking footage. There is also nothing saying that an external recording wouldn't break in the exact same way under similar lighting/exposure-->pushing/pulling in post.

Once you take the footage into your editor and start to grade it, any calculations will be made internally at a much higher precision, and it's not limited to your recording bitrate. You will however not be able to escape the fact that you only have an 8 bit color depth to work from, and that will break if you are trying to remap image information. But that is true for both internal and external recording.

Anyway... can you provide direct comparisons to better understand the issue at hand?

 

[Erik Naso]

The best use of an external recorder is for CG and green screen. This is where 4:2:2 will really help. In C-Log you have to expose properly still. If you underexpose you will get noise in the image when you grade it. I think 90% of the time the AVCHD codec is just fine. I think you have other issues other than the codec. The sample is full of noise from what looks like a high ISO and no lighting or very little lighting.

 

[RZE]

Can you detail some of the misinformation you've seen?

Also, can you provide identical frames from internal/external recording, as this is the only way to see the differences.

By only providing a single frame it is impossible to exclude what can potentially be a large number of factors that can result in poor looking footage. There is also nothing saying that an external recording wouldn't break in the exact same way under similar lighting/exposure-->pushing/pulling in post.

Once you take the footage into your editor and start to grade it, any calculations will be made internally at a much higher precision, and it's not limited to your recording bitrate. You will however not be able to escape the fact that you only have an 8 bit color depth to work from, and that will break if you are trying to remap image information. But that is true for both internal and external recording.

Anyway... can you provide direct comparisons to better understand the issue at hand?

In a thread catered for prospective buyers "someone" said: "While it might be comforting to know that you are getting the most out of the camera by going external, several tests have shown that it can be very difficult to tell the difference between internal AVCHD and external recording." This propagates an oversight about which gammas are usable internally. Clog is a main selling point of this camera, and people should know if they intend to use it then 24mbps won't work well.

I'll gladly post a direct comparison when I get a chance, probably this weekend, but since these results seem foreign to you, I have to ask have you actually recorded externally? Because the grain structure is clearly different, avchd=macro blocking, prores=organic grain, theres no other variable.

Also I don't know what "Once you take the footage into your editor and start to grade it, any calculations will be made internally at a much higher precision," means.

 

[AndreeOnline]

In a thread catered for prospective buyers "someone" said: "While it might be comforting to know that you are getting the most out of the camera by going external, several tests have shown that it can be very difficult to tell the difference between internal AVCHD and external recording."

That quote is actually from me. Do you consider it part of the misinformation you mention?

I have not had the opportunity to record externally myself yet, as the Ninja 2 is hard to come by here. I have looked at footage that others have recorded and made available for download.

But more importantly, I have recorded many internal clips in Canon Log during my own evaluation. I do not see macro blocking in my footage and that includes open fire and other demanding scenes. I don't use noise reduction and I see some noise. This noise is more like a fine grain and far from the blotchiness that is otherwise typical for digital cameras like prosumer camcorders and DSLRs.

But it is very important to understand that Canon Log is a very purpose made picture profile. It assumes a certain workflow and an accompanying understanding of some of its limitations in 8 bit color depth. These limitations have been the basis for discussions in other threads.

One obvious one is that if you record an internal scene in low light and expose Log to IRE 32-35ish, you will see quite a bit of noise when you lift your gamma to compensate in post. But since the sensor has no fixed pattern banding, the nature of this noise is more organic.

I am looking forward to your comparisons, as the camera still is young and there is much to learn. Just make sure to upload actual screen sizes and in a lossless format, such as .png. The frame above downloaded as a 1000px wide .jpg.

Also I don't know what "Once you take the footage into your editor and start to grade it, any calculations will be made internally at a much higher precision," means.

What I mean is that once you start to do changes and add effects to your footage in your NLE, the computations will be done internally (in the NLE) at a much higher bit depth than your 8-bit footage and associated bitrate. If the footage recorded OK to AVCHD without artifacts, they won't be introduced in post by the grading process. What can happen though, is that noise that is hidden in the blacks comes out if you lift the shadows too much.

 

[RZE]

The best use of an external recorder is for CG and green screen. This is where 4:2:2 will really help. In C-Log you have to expose properly still. If you underexpose you will get noise in the image when you grade it. I think 90% of the time the AVCHD codec is just fine. I think you have other issues other than the codec. The sample is full of noise from what looks like a high ISO and no lighting or very little lighting.

Color grading is necessary for deloging Canon log, so external recorder is or is not necessary? Also clog was designed to underexpose middle grey, If there is noise when underexposed material is lifted then that has to do with the exposure latitude of the recording format, and my point is that avchd is limited in that facet. I believe that in the same situation, prores would yield cleaner darks with no macro blocking with the same lut and exposure applied. More concrete examples to come. The sample was recorded at 850 ISO with middle grey at 35 IRE. Of course you can put middle grey higher, which I will in new examples to come, but your highlight retention will go out the window and macro blocking will still be present in lutted footage.

 

[AndreeOnline]

If there is noise when underexposed material is lifted then that has to do with the exposure latitude of the recording format, and my point is that avchd is limited in that facet.

There is a bit of mixing and matching of concepts at play in your post above.

Exposure latitude is a consequence of the imaging sensor design and its transfer characteristics via gamma curves. The C100 yields 12 stops at all ISOs, but reaches maximum over exposure latitude at ISO 850 and above.

The 850 ISO is the lowest ISO that gives you 800% over exposure "protection". That doesn't mean that other, lower ISOs would be a better choice in certain situations, such as when a clean and crisp result is desired for green screen work.

I am certainly not making a case that AVCHD is the same as 4:2:2 ProRes. I'm just joining others that have come the same conclusion as me: for "normal" shooting, the differences between the codecs are minute in relaxed viewing, and that is regardless of choice of gamma.

My tip to people starting to shoot with the C100 would be: if you find that you constantly have to lift the image in post when using Canon Log, consider using Wide DR where you still get 12 stops of DR but where the gamma curve follows the rec 709 standard a bit longer before rolling off.

 

[RZE]

Andree I guess we have different understanding of exposure latitude, I'm not talking about total dynamic range. I'm talking about the exposure range in which gamma corrections can occur without degrading the to the image. So the real bottle neck is definitely not sensor design or gamma curves, its the recording format.

In all this talk of dynamic range, you do understand that if canon log is not underexposed properly that you don't get 12 stops right? There are downsides to Wide DR gamma and color spaces that people should be aware of as well, but I guess no one will join me in discussing them, discussing limitations of the camera to be better prepared collectively has be met with more opposition then expected. But use what you want, just try not to go on a tangent with advice.

Also "normal" shooting doesn't mean much, address the situations stated in the original post, if you would like to debunk, just substantiate.

 

[Erik Naso]

Is this sample graded? It looks a lot like an underexposed shot that was pushed in the grade. I had this noise in the shadows as well with a scene that I shot in C-Log, ISO 850 with AVCHD that was underexposed.

 

[RZE]

Is this sample graded? It looks a lot like an underexposed shot that was pushed in the grade. I had this noise in the shadows as well with a scene that I shot in C-Log, ISO 850 with AVCHD that was underexposed.

It was a partial lut. I restored blacks, but gamma is still a stop under because the codec was already breaking.

 

[MPM]

This sample looks underexposed....just sayin.

 

[maarek]

Why don't people ever post 100% samples in this forum (and also anywhere else on the net, like EOSHD always posts lowquality jpegs)? There is enough space on any server to put a couple of 1920x1080 png files, why use this forum mechanism where it's scaled and jpeg?

And just saying you "normalized" a picture doesn't really cut it. Show the original. It looks like it was way underexposed. These are not raw-cameras. Expose properly. Like people used to do.

 

[AndreeOnline]

Andree I guess we have different understanding of exposure latitude

Yes, I have come to the same conclusion.

In all this talk of dynamic range, you do understand that if canon log is not underexposed properly that you don't get 12 stops right?

No.

How you expose your scene will not change the gamma curve associated with Canon Log—a curve designed to yield 12 stops. That curve is constant and maps light values logarithmically in a pre-defined way.

It is true that middle gray will not fall "in the middle" at IRE50, but rather at about IRE33, but you need to decide for yourself what in your scene should be addressed this value. Expose correctly for what is important in your scene.

12 stops is fairly good, but you will often encounter scenarios where the scene itself has a much larger DR. Especially if you are not controlling lighting. If you let the camera decide exposure it might try to save a practical light that is in the scene, and thereby exposing a person's face too low. Then you need to apply exposure compensation.

Ideally, you use a good external monitor that covers sRGB space and just look at that to make sure you're in the right ballpark.

My rule of thumb would be too look at the critical parts of the image and make sure that they don't need to be pushed large amounts in post.

 

[olof]

You may want to look at the tests I did in this thread.
#1

I will post more from those tests soon. Or as soon as I have time.

 

[RZE]

I'll be shooting my tests tonight, then I'm done with this community.

 

[AndreeOnline]

I'll be shooting my tests tonight, then I'm done with this community.

Why the drama?

If you can provide some identical footage from internal/external that would be a great resource to take the discussion in a good direction.

 

[MPM]

I'll be shooting my tests tonight, then I'm done with this community.

Hey mate,

No matter where you go on the interweb, people will have varying opinions of your own, it's no big deal and the world would be pretty boring if everyone agreed.

I'm not a huge fan of forums, because some people do get carried away (did I just contridict my first comment?) but after roaming here for a while I decided to sign up because, on the whole, the people here seem to be willing to share their opinions in a mature way, even if they are a difference of opinion.

We would all like to see your internal/external test of the C100 as we all own (or are wanting to own) the camera and, since we are all camera nerds, like this kind of stuff.

HIt the test up, post it here and like AndreeOnline said the discussioon will go in a better direction.

Yew!

Phi

 

[RZE]

Hey mate,

No matter where you go on the interweb, people will have varying opinions of your own, it's no big deal and the world would be pretty boring if everyone agreed.

I'm not a huge fan of forums, because some people do get carried away (did I just contridict my first comment?) but after roaming here for a while I decided to sign up because, on the whole, the people here seem to be willing to share their opinions in a mature way, even if they are a difference of opinion.

We would all like to see your internal/external test of the C100 as we all own (or are wanting to own) the camera and, since we are all camera nerds, like this kind of stuff.

HIt the test up, post it here and like AndreeOnline said the discussioon will go in a better direction.

Yew!

Phi

I have no problem with different opinions. What I do not like is insinuation of incompetence. I'm totally fine with I think thats underexposed, thats an opinion, but if some one is basically telling me learn how to do it right thats another thing. Why should I spend my own time conducting tests for that crowd.

I found some interesting results while testing. The test that will follow tomorrow morning consists of a subject on a basement background. I flagged off her key to cut the background in half. The flagged off area screen left is 3 stops under middle grey, screen right is 2 stops under. I wanted to see how noise behaved in that range. Middle grey is set to around 45 IRE. I shot at 850 ISO. Interestingly exposed this way, the images are very similar. The key differences is that the external actually has more noticeable noise. (We are talking small discrepancies not big ones). I even tested 6400 ISO. Though the noise is more organic it is more visible. In addition external is definitely sharper. The finer detail in the rock was lost completely in the AVCHD especially at higher isos. NR was at zero. Also theres a plasticy feel to the avchd, because the marco blocking blends so much. So sharper with slightly more noise vs. less noise and plasticy. I would say now that if you have non-discriminating corporate clients, then this onboard can be a really good option as long as you have enough light to overexpose middle grey without clipping. I still would never shoot anything narrative on board though, for me sharpness of the masters is really important. The noise from external can be removed, but at the same time, the biggest thing I learned was that both prores and avchd are not ROBUST codecs. I denoised via neat video and even the prores suffered from some degradation. I'm used to denoising RED and its the effect on the image is barely noticeable. I now see why the c300 is way more expensive. On the c100 you can underexpose the sensor properly to gain the full highlight retention of the camera, but delogging and denoising are going to leave you with a noticeably degraded image even in prores. What I realize now is that you are better overexposing losing highlight retention to get a cleaner image. The c300 requires no denoising , you can expose properly and you weirdly have a robust enough codec to delog without a significant reduction in picture quality, in a smaller file size! When I've operated c300, I haven't had to work as hard to get the most out of it.

I tested how raising middle grey a stop to 45% affects the highlights. If you put middle grey there you loose a stop of detail in the highlights and get one in the darks, IF your noise floor is not already blown. That means that highlights clip 3.5 stops up from middle grey even at 850 ISO. You get more information in the darks, but I hate clipping, 3.5 stops is dangerous in some instances and if you shoot at 320 it will be even less. If you plan to shoot AVCHD you definitely need to overexposed to save your darks, but at the same time hope you don't clip. If you can make your exposure range of the scene 2.5 stops up and down, you can get an ok image with avchd. But when the exposure range exceeds that, external recording will allow to underexpose to a greater degree to protect highlights, I want to do more tests to see how delogging 32% IRE pans out for avchd, it was horrific results my first test, but I'm going to do it again. My prediction is that if you want 4.5 stops of dynamic range in the highlights you're going to need an external recorder.

I also tested some c100 against RED, and sorry to say it, but RED really murders c100. Its not the same.

 

[Matt Turner]

I also tested some c100 against RED, and sorry to say it, but RED really murders c100. Its not the same.

It's also 3 times the price! Looks like you have done some thorough testing RZE, I hope you'll stick around and contribute more to the forums.
Personally I've found the workflow of the Ninja recorder to be worth far more than just the image quality. It's almost always attached to my C100 now.