CION still in the running vs FS7

Im just a bit lost here. Those girls shot with the Coin - they look like humans - stuff from the sony cameras leaves the humans looking like grey androids made of wax.

I think the reel is nice.

Highlights - yes we can see the clip point may not be too wonderful but Id take that over waxwork people.
 
I absolutely don't think so.

Have you really seen the skintones in the CION video?

The footage is much more contrasty than for example in the FS7 video, but you can change this with a click
in your NLE. But can you do this the other way round? No. You can't and that's the dilemma here.

Neither Sony nor the CION seem to produce nice skintones under the circumstances the videos were produced.
Nobody knows, how they treated the footage before putting them online.

At that point, we can really say this one's better than that one or however you wanna compare these two.
We just have some heavily compressed online footage showing the work of somebody, of whom we don't know
how he works. These videos say more about the authors than about the camera.

For me, I'D prefer the footage shown from the FS7 over the CION any day. The CION footage is more or less
destroyed for me. Some of the contrasty parts of the video are really blown from the highlights.

The FS7 skintones can be "waxy", but this depends very much on the post handling of the footage. As I can pull
even nice skintones out of my trusty FS100, I think it's possible with the FS7 and the CION as well.

The CION is just a completely different camera. You shoot under controlled cirumstances and want RAW or ProRes
in a very nice package . CION.
You want a one-man-band machine with small codec and low weight? FS7.
For me, it's that easy in the end. If the footage looks bad, it's your responsibility, not the cameras...
 
The footage is much more contrasty than for example in the FS7 video, but you can change this with a click
in your NLE. But can you do this the other way round? No. You can't and that's the dilemma here.

Well Im feeling that Slog is very hard to unwind and add suitable contrast to. The coin has the right contrast and yep might need a 10% swing in colour to get right.

I understand the logic of flat profiles but the sony ones are not singing to me at the moment.

Ive been using bmc4k recently (is that the same as the Coin?) the colours are great and the DR is somewhat challenged.

S
 
Not that it means anything as it's all personal taste of course, but I'm with Erik on this one. BUT I do think that a lot of reels and hobbyist films - and even some broadcast one actually - are over-doing the flat look, it's like a contest, each person saying "hey, look how washed out and flat our camera can be, it's the greatest"! And I do even think there's ignorance in some quarters that this flat shooting isn't actually supposed to be a "look" in a final film, it's just shot to capture the dynamic range of sensor!
Steve
 
I get the theory but Im just not personally (or online) seeing a lot of footage that is delogged nicely - The F65 looks nice (it was at the sony centre next to the FS7) so it is possible. The varicam looks nice too.

DR is an exchange for tonal capture so maybe the cheaper cams/codecs cant actually jam that much DR into the codec while retaining tonal detail?

SLog was of course developed for F35 12bit capture which has way more tones than 8-10 bit.

S
 
DR is an exchange for tonal capture so maybe the cheaper cams/codecs cant actually jam that much DR into the codec while retaining tonal detail?

You know what, you may have something there! It would be nice to think that there is a reason why some of these cameras are relatively cheap and that it would still be worthwhile spending serious money on some of the alternatives. It's something that wouldn't show up in the bear headline specs.
Steve
 
I thought this was well known? My understanding. Im going to round down and simplify the maths.

8 bit - 250 tones
10 stops =25 tones per stop
5 stops = 50 tones per stop

10 bit = 1000 tones
10 stops = 100 tones per stop
5 stops = 200 tones per stop

12 bit = 4000 tones
etc.

Now we all know that 8bit Slog falls over
Slog may allocate '2 stops actual' to 'one stop recorded' (hence 'flat') so 25 tones is spread over 2 stops, after delogging we have 12tones per stop, clearly not enough-hence banding and general file breakage.
To some extent I wonder if 10bit can handle 14 stops?

Using the BMC and seeing its 'flat' look it is clearly about 60% less flat than F3 Slog... and feels about right to grade but does have a pile less DR.

Of course our final delivery is 8bit in the main, so we only actually need to get 256 tones in our graded image - I can certainly see that a 10bit(1000tone) camera can accept a flat profile of a 1;4 ratio, which in reality I would suggest should be 1:3 to account for other things (compression). So with a 10 bit camera that has a flat curve less steep than 1/3 of a linear/709 curve one could expect a loss of tonality after delogging?

S
 
Last edited:
SLog was of course developed for F35 12bit capture which has way more tones than 8-10 bit.

That pretty much nailed it!

Logs isn't a "use me and you're happy" solution. It's a gamma profile that tries to put as much information into the
stream as possible. But the tonal range esp. with SLog2 is reduced drastically therefore in some parts of this curve.
Exposing correctly is very important here, otherwise you reduce tonal range in skintones instead of highlights for
example. It's a science of it's own and before using it, people need to understand it.

SLog3 is less drastical in it's gamma curve and uses a different gamut too. I hope that SLog3 is for the FS7, what
it should be. I saw some nice images made with SLog3 in comparison with SLog2 and SLog3 is much easier to
grade/delog. You'll loose some tonal range in the highlights, but the important part of the picture is not the
highlights but the skintone area. I'd like to use SLog3 with S-Gamut3.cine which is a smaller gamut than the huge
S-Gamut" of the SLog2 profile.

This is what Art Adams said, when testing it first with a F55:
"SGamut3.cine is meant to capture the same depth of color as a film scan, which is plenty for modern needs, and as it’s closer in size to both Rec 709 and P3 it’s much easier to grade."


Further on he makes the following statement, which is interesting for the FS7:
"The F55 and F5 use different color filter arrays, so even though these LUTs and MLUTs will be available for the F5 there will be differences. I don’t know what those are yet."

More to find in this very interesting article, which misses skintones:http://www.dvinfo.net/article/acquisition/sonyxdcam/sony_sgamut_vs_sgamut3.html


The CION and the BMPC4k don't use log-profiles, but RAW, which is (close to) linear and therefore much easier to cc or grade. There's no complicated gamma
curve, but a linear response. It uses much, much more memory as it isn't compressed.

Edit: Changed "exposed right" to: "exposed correctly"!
 
Last edited:
By "exposing right" I guess you mean exposing correctly.

My experience of the BMC is the 'film look' which is semi flat. We are shooting 4k422prores which is as much data as we can handle.

I did test it in hurriedly in raw but, found it a bit weird!

S
 
By "exposing right" I guess you mean exposing correctly.


S

I think he means over-exposing. This will give most detail in the shadows as there's tons of headroom to play with before you clip - if you under-expose you're not using a lot of this headroom and therefore not all the DR.
Steve
 
I think he means over-exposing. This will give most detail in the shadows as there's tons of headroom to play with before you clip - if you under-expose you're not using a lot of this headroom and therefore not all the DR.
Steve

But as the log curve flattens you are giving less tones/meat to each stop so really log curves dont work with ETTR.

S
 
I think he means over-exposing. This will give most detail in the shadows as there's tons of headroom to play with before you clip - if you under-expose you're not using a lot of this headroom and therefore not all the DR.
Steve

No.
That would be totally wrong for SLog. You have to think different.

SLog2 already has already a 2 stops lifted (shadow) curve. That's why its ISO is rated higher.
But the IREs above 60 are very flat, which means a heavily reduced tonality.

So what means overexposing then?
It means you put important parts of your picture where it shouldn't be because of the flatter log-curve
and therefore the reduced tonality.
Overexposed skintones in SLog2 often look very strange because of this.

Sony wants you to expose middle gray (18% grey) in the 32 IRE area. The image will look very dark and flat
sothat you want to lift exposure up to 50 IRE if judging by an untrained eye.
Here we have the LUTS for the monitor, they "delog" the gamma curve, sothat your exposure LOOKS right
and middle grey is where it belongs in a "normal" picture. Now exposing is much easier and works better
for your and the directors eyes.

What's going on with noise?
Because of the lifted shadows, you'll see more noise than without SLog. But you still have a tonal range in
your shadows, that should be brought back down. Your noise will be brought back too.

Why is SLog so popular?
People want to impress by the huge dynamic range, their camera can capture and most of them even use it,
when it doesnt make any sense. It came with the big sensor cameras which are so popular today.
When recording in 8bit in a more or less ordinary situation you're good with 8 stops of DR.
It's just idiotic to risk loosing the tonality of skintones to capture some extra stops of DR.
Who's gonna look at the lightbulb in the background during an interview? Nobody!
But we can see that the skintones are so strange.

Some additional info's about SLOg2 here: http://blog.abelcine.com/2013/01/18/sonys-s-log2-and-dynamic-range-percentages/


The BM RAW cameras are completely different as is the CION.

The RAW response is linear (at least close to) and 16bit. It has to be interpreted before you can see the tonal range on
a display first. What you see on any display is not what you get when recording linear RAW. It's just one possible
interpretation of the data for your display.
But as it is linear, thereÄS no magic in the gamma. It more a line than a curve a and the tonal range is more or less
independent form the exposure. Its the same tonal range everywhere.
Very easy to grade and cc wihout the need of special delog-profiles.

The biggest downside is the amount of data that's needed for RAW. If you record the BM and the CION in ProRes your
tonal range is reduced to 10bit. If you don't do heavy grading it should be enough but you loose the ability of the
linear RAW that can be manipulated in every way.
 
Since the beginning AJA has said that they were working to build a camera that has "a look." All cameras do whether they like it or not. If you shoot an Alexa ProRes to SxS cards that's the look. Whatever profile you choose or in camera changes you may choose to make the baked in image is just that - baked in. I am SO glad that CION has chosen this path. It is the reason I like the camera, plus I happen to like that high contrast, deep inky blacks, nicely saturated look.

Then if you want to fiddle with it from scratch - boom, take the RAW.

This may seem like heresy but there is a subtle tyranny to how slog seems to work. Expose in a way that is unfamiliar and unsuitable for viewing, fix later. Deconstructing and reconstructing log footage may provide better DR is some areas of the spectrum if you manage to get it right, but it seems, at least to me, error prone and horrendously fiddley.
 
Last edited:
The difference is, Arri sorted out how to work with LUTs right from the beginning sothat
the display looks as it should.

Sony still is trying to find out, how to combine RAW, LOG, cinegamma and LUTs at the same
time. A camera that does everything does nothing easy and perfect.
Alexa is for cinema, no need for Rec.709...

The CION should have had a "Look" but if it's what we saw in the promo video. Thanks no.
There's no subtle highlight rolloff (Alexa) or a tendency to smooth skintones (Canon).
It looks like a 709 compatible S35 camera...
 
No.
That would be totally wrong for SLog. You have to think different.

So when you said "expose right" you did mean "correctly" then?
I was just remembering what Alister Chapman's write up on log exposure was saying - that if you over-expose you get more of the DR. Maybe I read it wrongly or not in enough detail.

Steve
 
I absolutely don't think so.

Have you really seen the skintones in the CION video?

The footage is much more contrasty than for example in the FS7 video, but you can change this with a click
in your NLE. But can you do this the other way round? No. You can't and that's the dilemma here.

I defer to your encyclopedic knowledge of slog but why couldn't you take that fat, true 4K 12 bit ProRes 4444 into your NLE, lower the highs, raise the mids and desaturate it a ton to get a flat look? I would bet it would hold up pretty well.

Neither Sony nor the CION seem to produce nice skintones under the circumstances the videos were produced.
Nobody knows, how they treated the footage before putting them online.
AJA stated clearly that they did not manipulate the latest teaser in post. You can see it in the description on Vimeo.
 
The difference is, Arri sorted out how to work with LUTs right from the beginning sothat
the display looks as it should.

Sony still is trying to find out, how to combine RAW, LOG, cinegamma and LUTs at the same
time. A camera that does everything does nothing easy and perfect.
Alexa is for cinema, no need for Rec.709...

The CION should have had a "Look" but if it's what we saw in the promo video. Thanks no.
There's no subtle highlight rolloff (Alexa) or a tendency to smooth skintones (Canon).
It looks like a 709 compatible S35 camera...

I have seen other footage from the CION that may prove otherwise but there isn't a lot out there publicly yet. We'll see.
 
So when you said "expose right" you did mean "correctly" then?
I was just remembering what Alister Chapman's write up on log exposure was saying - that if you over-expose you get more of the DR. Maybe I read it wrongly or not in enough detail.

Steve

Yes, sorry. I meant correctly - not right.

If all you want is DR, expose to the right as much as possible.
But As I've stated above, it completely makes your tonality and colours freak out
if you go to far from the 18% middle grey to 32 IRE rule of SLog2. It's that simple.
 
I defer to your encyclopedic knowledge of slog but why couldn't you take that fat, true 4K 12 bit ProRes 4444 into your NLE, lower the highs, raise the mids and desaturate it a ton to get a flat look? I would bet it would hold up pretty well.

That would be the best way to get to what we want without all that SLog stupidity.
I would like that very much!
But log-profiles do exist for a reason.

If there's enough information in the ProRes 4444 footage, it would be pretty terrific.
But the codec can just show, what the sensor delivers - not one pixel more.

The promo video showed that there's a harsh highlight rolloff and strange skintones.
These skintones might come from LED lightning and wrong WB - who knows.
But when I show a promo video of a new camera from a new manufacturer in the
camera circus - better let somebody work, whon knows how it looks good.

If it's manipulated or not - who cares?
We see the footage, the first footage and make our very own conclusions.
They may be wrong, but that's part of the game AJA is playing.
 
The Cion is the first camera whose "ungraded" footage already looks right to me, since I began in 1990. I had started thinking I would always need to shoot raw or log, because cameramakers don't seem to share my tastes. If Cion's footage truly looks like this out of the box, it will certainly save time in post.
 
Last edited:
Back
Top