Canon 1D C vs Sony A7S 4K - dynamic range - preview

[hmcindie]

I have to agree with Andrew here, the 1DC does make a better usable picture than the A7s. The a7s has a lot of features that the 1DC lacks but the 1DC is a powerhouse in those things it does do. The only real problem with the 1DC was always the ridiculous price (and soft HD)

It is also beautifully built. It's a tank. I'm gonna try to get a loaner again just to shoot something with it and compare more to my a7s.

[AccordingToMe]

Higher ISO = more noise. Is it possible the reason they are forcing higher ISO for Slog is to create noise for dithering to prevent banding?

​My own analysis of Canon Log is that when put the 1Dc in Log mode (which is a special MODE in the camera, not just a picture profile) the raw data is pulled by 2 stops. The ISO400 then boosts the image back up again, according to the Log curve.

If you put a 1Dc in Log mode and ISO below 400, you will never reach IRE 100, no matter how you expose. The data is pulled. It is equivalent to highlight protection mode. It makes sense for non raw recording. Take 10 clean base stops and add 2 "ISO stops" and you end up with 12 stops, even if DxO only measures around 11.8 for 1Dx raw.

Sony does the same. But instead of going for 12 reasonable stops, they up the ante and go for 5 stops via ISO3200. 10 base stops and 5 ISO stops. DxO rates the A7s pretty high at ISO100, but raw is one thing and video is another. It assume they are doing something similar behind the scenes. I don't know what the signal looks like on an A7s at ISO100. can you still expose the histogram all the way to the right?

[hmcindie]

I don't know what the signal looks like on an A7s at ISO100. can you still expose the histogram all the way to the right?

​Depends on the picture profile. The 16-235 level cinegammas will go to the right but stop at 235, the 0-255 cinegammas go all the way to 255 but they won't let you go under ISO 200.

[AccordingToMe]

​Depends on the picture profile. The 16-235 level cinegammas will go to the right but stop at 235, the 0-255 cinegammas go all the way to 255 but they won't let you go under ISO 200.

​OK, thanks. The fact that you can't go below ISO200 suggests to me that there is something similar going on.

It dosen't really matter what it is, but my understanding is that Sony recommends quite high ISOs with the log profiles, correct? ISO3200 even? And then the community has landed somewhere below that at maybe ISO2500? I'm grasping here.... not really following Sony...

As anecdotal as it may seem, I also can't help but noticing that for filmning, looking at the "money making" ISOs from ISO400-6400, the DR is for all intents and purposes identical between the A7s and the 1Dc (according to DxO screen). The difference between the "theoretical" 14 stops and the more "rubber hits the road" 11 stops often fall below the forum discussions "noise level".

[Andrew Reid]

Canon modeled Canon Log after Kodak's Cineon system from 1993.

​That's interesting. How did you find this out, do you have any more info?

[Andrew Reid]

I have to agree with Andrew here, the 1DC does make a better usable picture than the A7s. The a7s has a lot of features that the 1DC lacks but the 1DC is a powerhouse in those things it does do. The only real problem with the 1DC was always the ridiculous price (and soft HD)

It is also beautifully built. It's a tank. I'm gonna try to get a loaner again just to shoot something with it and compare more to my a7s.

The 1080p in Super 35mm mode internally is amazing! As is the 1080p HDMI in 1.3x 4K mode. Not a hint of softness or moire, as good if not better than the C300. The full frame 1080/24p is softer - same as the 1D X but better than the 5D Mark III and more dynamic range. The 60p full frame is the softest but then so is the A7S in 60p full frame mode.

[AccordingToMe]

​That's interesting. How did you find this out...?

​It's in the Canon Log white paper. I don't have the direct link, but it should be easy to find with google.

Highly recommended reading if shooting Canon Log. Canon has released many white papers on the Cinema EOS system. Reading them will probaly increase one's understanding of Canon's philosophy towards sensors, cameras and lenses.

Remember, Canon Log has been out for a long time. It's been thoroughly discussed and examined online.

​...do you have any more info?

​In their documentation of the Cinema EOS system, Canon describes the C300 as a quick-to-market product. They had to use Digic DVIII and went with a mpeg-2 codec. Mpeg-2 specifies 8 bit, and so Canon's hand was forced.

The Canon C100 mkII uses Digic DV IV, but again it comes with a consumer codec. I am, however, fully expecting the C300 mkII to be both 4k and 10 bit or more capable. It will be very interesting to see those camera specs. Not in relation to the FS7... but in relation to Canon's own IQ goals/standards.

The C500 ($16.000) already does this:

• 4k (4096x2160/3840x2160) raw 10bit at 60fps
• 4k half raw 10bit at 120fps
• 2k 4:4:4 12bit at 60fps

How about 4k internal to a consumer (think efficient) codec with 4k 10 bit output, 2k 12 bit output, 2k 120fps, Dual Pixel AF?

[Inazuma]

The shadows on that 1DC image just look more crushed to me than the A7S'. I'm not seeing any dynamic range advantage. Maybe different monitor gamma?

[Andrew Reid]

The shadows on that 1DC image just look more crushed to me than the A7S'. I'm not seeing any dynamic range advantage. Maybe different monitor gamma?

Yes that sounds likely. ​Any way to show visually what you're seeing on your screen? Take a photo of your display perhaps?

[Andrew Reid]

I'm not seeing any dynamic range advantage.

​The light rope highlights are burnt in one and not the other. C'mon now get a grip.

[Guest Ebrahim Saadawi]

Let's be a bit more scientific, a subject like this should not include any opinions, just pure facts.

Lets start with resolution/detail just to take out of the way:

1dc



A7s



They're the same resolution. The 1Dc is just not sharpened internally, when you do it in post you get to the exact same a7s level. No opinions just facts.

Shadow DR

1Dc



a7s



Same detail in the shadows, but the 1Dc has lower noise floor, so effectively slightly more DR. Not a worthy difference though. So let's head to Highlight DR

1DC



A7s



Note: A7s more clipped and shows yellow colour shift before clipping. The other example:

1Dc



A7s



Note: noticeably more clipping on the a7s and the 1Dc shows information outside the window. 

1Dc



a7s

​
So based on the samples Andrew is giving us, no opinions, just facts:

Detail: Tie, but the 1Dc starts more organic/softer
DR: Less shadow noise and less highlight clipping on the 1DC and a7s clips the blues extremely strange, all add up to slightly more DR, on the 1dc maybe 2/3s of stop or one stop.
Colours: and here's the biggie for me, the a7s is yellow. Definitely makes people yellow and needs to be corrected carefully. The 1dc is reddish and lively

1dc (red desk)



a7s: (yellow desk)



I've seen the yellow/green skin in practically every a7s video outthere, it's not easy to correct, but I can do it. This issue + the faulty blue clipping gives the 1DCa clear advantage in colour science/rendition. 

[toxotis70]

I dont have any yellow /green tint in my videos.

i always shoot slog2-sgamut and then color grading with right lut in resolve.... Maybe you can try that too.

 

blue color clipping is because of wrong white balance, if you put 5000 kelvin or more, there is no blue clipping

[Inazuma]

Yes that sounds likely. ​Any way to show visually what you're seeing on your screen? Take a photo of your display perhaps?

​Not sure if you will be able to see this here either (photo of the photo attached). But what I'm talking about is that I can see clear tonal differences between the lenses on the left and the wall. I can see more of the sheen on the lens of the camera in the centre. I can see more tonality on the camera at the front on left. It looks more 3D where the on the 1DC image it looks more flat because of the crushed shadows. If I copy and paste the image into Photoshop and increase the contrast in the shadows of the 1DC image to get that same detail out, the image reveals a lot of banding and chroma noise. I will concede that there is slightly more range in the highlights of the 1DC image, but it really is slight.

But anyway, I guess it doesn't matter to me much since I am not in the market for either camera right now. Just thought I'd point out that I can see differences that don't seem to match the OP's findings.

[jcs]

I've seen the yellow/green skin in practically every a7s video outthere, it's not easy to correct, but I can do it. This issue + the faulty blue clipping gives the 1DCa clear advantage in colour science/rendition. 

​If Canon = red bias (R), and Sony = yellow bias (R+G), then taking the difference D = Sony - Canon, D = (R+G) - R, D = G: the difference simplifies to a green bias. The green bias with Sony is mostly with SGamut color mode. Cinema color mode has less green, Pro has even less. I can now fairly well match 5D3 RAW with the A7S skin tones with custom tweaked profiles, never using SGamut. While SGamut make work OK if corrected in post, I suspect without the proper color science workflow in post, when skin tones are corrected to look less green, the background may end up magenta, etc. Some lights have color spikes/bias, especially low CRI LED, FL and CFL, adding to the challenge. More info here: http://www.eoshd.com/comments/topic/7762-lighting-white-balance-and-skin-tones/

 

[toxotis70]

 

 

sgamut is not only a much wider color space than anything else, is also a twisted one.

That means you HAVE to use a proper technical  LUT to bring colors in a REC709 space (SLOG2 -> REC709)

You have to put this node in Davinci in the end , so you can correct in 32 bit space before the conversion.

The green  or yellow tint you see , is only because you see the image in a wrong color space for your monitor (which is only REC 709 compatible) 

 

[sam]

 

 

sgamut is not only a much wider color space than anything else, is also a twisted one.

That means you HAVE to use a proper technical  LUT to bring colors in a REC709 space (SLOG2 -> REC709)

You have to put this node in Davinci in the end , so you can correct in 32 bit space before the conversion.

The green  or yellow tint you see , is only because you see the image in a wrong color space for your monitor (which is only REC 709 compatible) 

 

You're right. I think this is why the fs7 has produced such underwhelming footage for many on youtube and vimeo.  ​Although I don't think he mentioned what monitor he is using and it  could be within a 1% or 2 of encompassing  Adobe rgb  if he owns an Eizo or equivalent.

[Andrew Reid]

That S-Gamut chart applies to the F35 and F65 top end only not the FS7 or A7S.

And in Inazuma's reproduction of my screen grab he seems to think more contrast = less crushed blacks and a flatter image = more crushed blacks when it is the opposite, so make your own view of that people, that's all.

[sam]

 what am I missing?  I see the f65 and f35, and would expect the fs7, f5 and f55 to fall relatively close.  Also edited my above post in which I stated s gamut was close to adobe rgb, but was looking at the wrong line.  Thanks Andrew for all the knowledge shared over the years. 

[AccordingToMe]

​…on the 1DC image it looks more flat because of the crushed shadows.

​An image can't be more flat with crushed shadows. When people say crushed shadows they mean "crushed down", as in turning dark gray areas completely black.

Crushing the shadows makes an image more contrasty and punchy, often with the purpose of removing the dark, noisy part of the image by sending it to black.

[Inazuma]

That S-Gamut chart applies to the F35 and F65 top end only not the FS7 or A7S.

And in Inazuma's reproduction of my screen grab he seems to think more contrast = less crushed blacks and a flatter image = more crushed blacks when it is the opposite, so make your own view of that people, that's all.

Are you trying to discredit me because I'm unfamiliar with the terminology? That's really fine. I never claimed to be an expert of any kind. I'm just pointing out what I see. My professional knowledge doesn't play into that.