April 19, 2011

First Sony F3 Uncompressed S-Log Footage Hits the Web

While I was catching some Zs at NAB, the guys from Next Level Pictures and a number of shooters who have appeared on this site in the past (Vincent Laforet, Jared Abrams, Timur Civan, Tyler Ginter) were off testing the Sony F3's S-Log firmware upgrade. This is the first time the uncompressed outputs from the F3 were enabled outside the walls of Sony (in this case, they were recording to a Cinedeck), and the footage should demonstrate greater latitude than the default F3 settings. Here's the test:

http://vimeo.com/22540499

Initially, it's hard to judge the differences between the two, because the original XDCAM material is so much brighter and more saturated. S-Log by default outputs a flatter, grayer image, and the extent to which you retain any of that look has to do with the Look-Up Table you use and how you grade the footage. To better match the two, they could have added some more saturation to the S-Log footage, but that isn't the most important thing to pay attention to -- the key thing is the blown-out highlights lost in the XDCAM footage that are retained in the S-Log material. To better demonstrate this, I pulled a couple of framegrabs from identical shots and darkened them. This way we're not looking at a brighter and more saturated image compared to a darker, flatter image -- here, we're looking at two shots that have been darkened (roughly) the same amount, to see which version retains more highlight detail. Take a look at this comparison:

Look at the reflection on the glass in her hand. You can probably guess which side is the S-Log version -- the left. The highlight has significantly more gradation and detail than the XDCAM version.

S-Log gives much greater flexibility in post, though exactly how much, it remains to be seen. Sony claims an extra stop of latitude, and the participants in this test seem to think it'll be up to two stops. Without the firmware upgrade the F3 gets 11.5 stops, so with S-Log that would presumably take it at least to 12.5 -- better than any other video camera out there save the ARRI ALEXA and RED EPIC (both of which cost significantly more than the F3). Yes, for an HDSLR shooter the F3 is a far more expensive camera with a street price of thirteen grand. But it's far less than an ALEXA, which is $75k +, and the S-Log upgrade has the potential to turn the F3 into a mini-ALEXA. Unless Canon announces a large-sensor professional camera in this range, and until the EPIC-S actually ships, Sony has carved out their own pricepoint for a very capable camcorder.

Here's much more on the test in the form of a behind-the-scenes video:

http://vimeo.com/22576231

Details from the release:

New York City, NY (April 18, 2011) Next Level Pictures in association with Tstop Cinema teamed up with Abel Cine, Kessler Crane, Wide Open Lens, Angenieux, Carl Zeiss, Cooke, Tyler Mounts, and Miller to unveil the worlds first ever camera test using the Sony F3 with S-Log. S-Log isn’t scheduled for official release until late summer 2011 so we are proud to get our hands on the camera and put it through several latitude tests to include helicopter aerials by Vincent Laforet, vibrant Vegas city lights, and skin tone tests of two models in ten different locations throughout Las Vegas at NAB 2011. The crew of twelve had only ten hours with the camera to produce and film the entire project. The purpose of this test is to compare the latitude and image quality of the new S-Log firmware to native EX codec.

Your Comment

22 Comments

The footage looks great, but I have one huge problem with the video: Why would you compare graded S-Log footage to ungraded xdcam footage? Obviously the graded footage is going to look better. It would make a LOT more sense to compare the graded S-Log footage to GRADED xdcam footage.

Just my two cents...

But, like I said, the footage looks great. The F3 is pretty impressive.

April 19, 2011

-1
Reply
Trevor

Perhaps I'm mistaken, but wouldn't the XDCAM footage be comparable to graded S-Log footage in the sense that the XDCAM footage would already have "look" applied to it, which would need to be added to the S-Log footage through grading?

I'd be interested to see how the ungraded S-Log footage from the F3 compares to the 4:4:4 footage that can be output from the NEX-FS100. I doubt it would retain as much latitude otherwise Sony would seemingly be cannibalizing F3 sales with the FS100. However the sensor in the F3 is the same as in the FS100, so the potential is there.

April 19, 2011

0
Reply

The NEX-FS100 doesn't have a 4:4:4 output. Correct me if I'm wrong...

April 19, 2011

0
Reply
avatar
Ryan Koo
Founder
Writer/Director

April 19, 2011

0
Reply

Really, Sony will give their $5k camera a 4:4:4 output when their $13k camera requires a $3k upgrade to enable 4:4:4? No way.

Spec page says "uncompressed 4:2:2 HDMI output." Which I believe is 8-bit. The F3 does 10-bit 4:2:2 HD-SDI and can be upgraded to 4:4:4. Sony's not stupid -- regardless of what rumors have been reported, 4:4:4 on the FS100 ain't gonna happen.

April 19, 2011

0
Reply
avatar
Ryan Koo
Founder
Writer/Director

How do you explain this?

Doug: “Yes, I’ll confirm that the HDMI out on the FS100 is 4:4:4 8bit. That is what the FS100 product manager at Sony told me right to my face."

I agree that it doesn't make sense for Sony, but why would the product manager for that specific camera be wrong about that? Of course B&H would say 4:2:2, it's what was written by Sony when they sent over the PR weeks ago. If they decided to change it in the interim, wouldn't the product manager know that?

April 19, 2011

0
Reply

I'll believe it when I see it...

April 19, 2011

0
Reply
avatar
Ryan Koo
Founder
Writer/Director

S-Log footage has to be graded in the sense that it's just data and you need to apply a look. They weren't grading the S-Log stuff other than to apply a similar look to the XDCAM footage's default appearance.

April 19, 2011

1
Reply
avatar
Ryan Koo
Founder
Writer/Director

Nice footage for sure. Just wondering how the single link 4:2:2 10bit HDSDI (non-S-Log) footage would compare recording to something like the KiPro or Nanoflash...

Shooting dual link to a $10k Cinedeck isn't going to fit a lot of indie budgets. But obviously makes a huge difference for those that can afford it.

April 19, 2011

0
Reply

Agreed -- the Convergent Gemini at least brings the price of an uncompressed external recorder down to $6k (plus media), at least.

April 19, 2011

0
Reply
avatar
Ryan Koo
Founder
Writer/Director

How's $345? http://www.cinema5d.com/news/?p=5917

Note: I have nothing to do with Cinema 5D, I just happened to see both these articles there and they're relevant to the comments here. In case you were wondering.

April 19, 2011

0
Reply

Yeah, I'm getting to that one too... saw so much stuff at NAB! It won't do 4:4:4 and it won't do S-Log, but you can't beat the price.

April 19, 2011

0
Reply
avatar
Ryan Koo
Founder
Writer/Director

Yeah forgot about the Blackmagic dealio... again still uncompressed so tough workflow for post - probably still have to transcode to ProRES or DNX or similar for any longform projects. I do features so uncompressed is out for me.

April 19, 2011

0
Reply

wow, that hyperdeck looks amazing!!
specially because the price opens it to a much wider market: I would expect it to become huge if the next generation of canon DSLRs keeps the H264 codec but offers proper HDMI out

April 20, 2011

0
Reply

Is the F3 rated at 11.5 stop!!! On provideocoalition they tested the f3 at 12+ stop of dynamic range and in some case (cine-gama) a hint of a 13 step. Abelcine did the same thing and had the same result on backlit charts. Provideocoalition as a set of test of the Alexa, Redmx and even 5D for comparison.

April 19, 2011

0
Reply
Daniel

At the Single Chip Camera Evaluation (more on this soon), which I watched on two different screens at NAB, Bob Primes, ASC and a crew of around 100 rated the F3 at 11.3 stops. The ALEXA scored a 14.5, so I wouldn't say those tests were low. Either way, seems the F3 gets around 12 stops depending on the test.

April 19, 2011

0
Reply
avatar
Ryan Koo
Founder
Writer/Director

Was that 11.3 with Slog? I believe most are saying that with Slog they are confident the F3 is getting 14 stops.

April 20, 2011

0
Reply
MadP

No, that 11.3 was in REC 709. Tests on S-Log should be out soon.

April 26, 2011

0
Reply
avatar
Ryan Koo
Founder
Writer/Director

I am asking myself if the problem is not about how they measured it their. There are many sensors where the dynamic range is in the shadows. It might seem strange what I am saying, but from my experience mostly with photographic equipment (lot of camera I have used during the years and also raw output) most of the info was in the shadows. You could underexpose quite a lot to protect the highlight, which would look very dark and just do a boost in gama in photoshop to get a perfect image and compress the highlight with a curve so that it does not blow. There are even extreme to that in the Nikon D7000 and Pentax K7 sony equipped 14 stop dynamic range camera. They call it the iso less camera... the reason is that you can take completely black pictures (to the eyes) at low iso and boost the signal so much to get a good image.

The provideocoalition test between the alxexa and red mx emphasize ita lot. How the Alexa has more highlight and how you can dig very far into the shadows of the red mx. In a dvxuser thread commenting on the footage, one of the participant in the above test is claiming 1.5 to 2 stop gain in dr and about 14 stop dr, a hair blow the Alexa.

April 20, 2011

0
Reply
Daniel

September 22, 2011

0
Reply

Truly inspired

September 25, 2011

1
Reply
Eddie T.

Is this truly the only S-log footage available. It's being beat to death. Please post other footage.

November 25, 2011

-1
Reply
joe lipari