The power of sLog - and why i'd pay $6,000 for it.......

TimurCivan

Director of Photography
Staff member
I got sLog yesterday, and i finally have had a chance to test it out a bit.

Below you will see, sLog raw Tiffs, and graded Tiffs. Feel free to DL and grade. For reference, i shot the skin tones at about 35 IRE to Uncompressed Quicktime - 3G 444 sLog, @ 0dB F16, with a 2k Par with double diffusion, and in the shot is a 40w Compact fluorescent. no Fill.

The skin tone is metered at a T16, with the highlight +1 stop, and the fluorescent bulb itself is +9 stops, and the top of the puppy's head is +3 stops over key.

I am flabbergasted that i can own a camera that has this much imaging power.

me_log_1.1.3.jpg

Uncompressed TIFF http://www.timurcivan.com/downloads/me_log_1.1.3.tif

Quickly graded on a Davinci Resolve to put some contrast and secondaries in the image so you can see what its supposed to look like.

me_delog_1.1.2.jpg

Uncompressed TIFF http://www.timurcivan.com/downloads/me_delog_1.1.2.tif



By the way..... 444 IS everything its cracked up to be.

Some extreme lowlight grain testing...

This was lit with 1 single candle. Cooke Panchros @T2.8 @ 0dB on The Sony F3 with sLog.

MIND YOU: when in sLog your sensitvity doubles. So 0dB, is now 1600 ISO, 6dB is 3200 ISO etc.... 18dB is 12,800ISO

They do get grainy, but its REMARKABLY clean considering the ISO.

sLog Raw Out of camera (DL the 1080p TIFF)
0dblog_1.1.jpg

TIFF: http://www.timurcivan.com/downloads/0dblog_1.1.1.tif

0dB Graded
0dbdelog_1.1.jpg

http://www.timurcivan.com/downloads/0dbdelog_1.1.3.tif

6dB
6dbdelog_1.2.jpg

6dB Delogged TIFF http://www.timurcivan.com/downloads/6dbdelog_1.2.2.tif
6dB RAW LOG TIFF http://www.timurcivan.com/downloads/6dblog_1.2.1.tif

12dB
12dbdelog_1.3.jpg

12dB delogged TIFF http://www.timurcivan.com/downloads/12dbdelog_1.3.2.tif
12dB RAW LOG TIFF http://www.timurcivan.com/downloads/12dblog_1.3.3.tif

18dB
18dbdelog_1.4.jpg

18dB deLogged TIFF http://www.timurcivan.com/downloads/18dbdelog_1.4.2.tif
18dB RAW LOG TIFF http://www.timurcivan.com/downloads/18dblog_1.4.1.tif


Also, just for kicks we recorded log to xdcam, to see if we could grade it. Did not work one bit. Log has to go offboard. Too bad.

-----------------------------------------------------------------------------------------------------------------------------

Todays Test:


1 Sony F3 w/ sLog update
1 Ki Pro Mini
1 set Cooke iPanchros
1 DaVinci Resolve

Objective, to discover where sLog should be exposed, How it should be treated in post, and what the difference between 709 standard Gammas and sLog in post, and acqusition.

From previous testing the F3 w/ sLog clips to white @ 6 stops over key. Using this information i metered the brightest highlights in each scene to +6 stops, JUST touching the clip point ( or what i thought was the clip point). The dark end of the scale fell to where it fell.

The goal being to see how far down you can drag up the image, and how conservative i have to be with using fill light to maintain an image in the black section of the scene.

OK, So here's the skinny. You have to expose just as carefully as you would with any other medium. The sLog w/ Offboard capture gives you incredible room to adjust and manipulate later, however like every camera, the closer you get to perfect in camera, exposure wise, the more Oomph your image will have after delogging. That said, 10bit 422 ( and especially 444 capture) gives you and INSANE amount of room to fudge.

For kicks we recorded the REC709 180% LUT as well and tried to match them to see if matching was possible. The Footage can be massaged into a similar looking image, but the highlight quality, grain structure, and shadow retention goes out the window. If you need a very high dynaic range shot with and F3 and dont have sLog, if you at least have a KI Mini, you can get SOMETHING.... it wont be as pretty, but it'll be there.

These compositions are boring, very boring, but they are carefully chosen. to reflect situations i've run in to the last few years, on various shots that have always bothered me and made me wish i had the "Infinite Dynamic Range" as MacGregor so eloquently put it. Well, for the most part, we have that now, at an indie price point. Yes, the Alexa is better.... by about a stop. (when you hit the 13.5+ zone 1 stop actually becomes a lot) but at 1/3 the price, this is terrifying performance. And the fact that i own it, and its paid it self off in 3 months, and i dont have a 70k mill stone around my neck is simply phenomenal.

The thing when shooting sLog thats tricky; it goes against every instinct you have as a DP. You have to under expose, so far down that you think "Theres NOTHIGN THERE!!!!! WHAT THE #$%^ I'M DOOING!?!?!?!?" but use your exposure tools, use your false color, use your wave form. Skin tones stay at 35IRE, High lights on skin @ 40-45 IRE. Dont worry, its ALLLLL there.... The trick is to use spot meter to see how many stops you are from one point to another in the scene. Get a Grey card, light your set, get your highlights to about +6 stops over key, then set the Grey card @ 35IRE with the waveform. You should be set. ( i think it goes without saying TRY IT OUT for yourself first, see how you like to use it...)

The truth is i dont know why, but for some reason, color information, because of the sLog, is more accurate, pleasing, and is retained all the way up to the clipping point. Even 10bit capture on regular 709 doesnt have the same amount of detail in highlights. Its just not retained in the same way. Pay attention to the highlights in the forearms in the "dramatic" lighting shots.

I highly reccomend downloaing the 2gig low compression version to really see in full resolution whats going on, and how Frigging sharp the F3 really is.
 
Timur it would be nice if you can do a small test (yeah righhhht!!!) with regular 444 without slog and another one with Slog (Same set and lighting). You can over impose all 14 f stops over the areas where you measured every different exposures. Kind of what Zacuto did, but it doesn't have to be a nice set with models.
I would like to help you, but I am 1500 miles from NY.
 
i didnt have time. My dit had work early in the morning. This weekend i will continue.
 
been working with s-log for the past 2 days on a commercial shoot with Ed David, and his f3 with s log as B cam, and A cam as Alexa. my consensus is, the F3 has a fusion powered reactor of a sensor, in a camera body that can barely handle it :) and that the extra money on a Alexa isn't for nothing... BUT

the f3 performed great, kept up 80-90% of the time with the Alexa, though there were a few scenes where it was extremely high con, using the full 14 stops of the Alexa, and the F3 was short just that little bit rendering the highlight detail out the windows coming into hallways with little to no fill. It looks like the Alexa has literally, a full stop over the F3, but the F3, has a full stop more sensitivity than the Alexa.

Also the s-log curve is nowhere near as aggressive as c-log off the Alexa. You really have even more room to play with when you stretch the contrast out on a Alexa cause it compresses the gamma so much. the F3 doesn't do it nearly as much, so on really high con scenes, you might not even be able to tell your on s log if you don't have a proper waveform to look at. F3 s log is more contrasty, and more saturated too in comparison to the Alexa. It's something that should be said because they are too different gamma curves, and should be exposed differently for best results.

the inability to do custom white balance on the f3 in s log mode kind of bites, there was a lot of mix color temp situations that the Alexa ate alive, but there was plenty of room to dial back in the right white balance off the f3 with ki mini. The one lights matched wonderfully, and minus some of the really high con scenes where we had clipped windows, most people probably would have a hard time telling the what was Alexa and what was f3 if you intercut them together.

I really wish Sony would take the F3 sensor and jam it in a more powerful body just to see what happens... Hopefully we can get more done this weekend, not much time for anything these days though :p
 
Given the price points of the two camera systems, Tom, it's amazing you can even compare them as closely as you did.
I'm not surprised the F3 fell a bit short of the Alexa... I'm surprised it was that close.
That's pretty impressive.
 
Given the price points of the two camera systems, Tom, it's amazing you can even compare them as closely as you did.
I'm not surprised the F3 fell a bit short of the Alexa... I'm surprised it was that close.
That's pretty impressive.

it is, but that extra stop of highlight retention really separates the men from the boys. we had a hallway shot with 3 windows that was clipping the hell out of the F3, but Alexa ate it up. It saved a ton of time not having to net or bring down the windows in other ways, and we barely had any fill too. plus there is the more polish in the actual physical hardware of the Alexa that makes you kind of sigh when you pick up the F3 and have to frankenstein it a bit. but again, talking about 60 thousand dollar price variance. they A and B cam very well, and the F3 def feels like the baby brother to the Alexa, but the baby is still a baby, people shouldn't forget that.
 
the inability to do custom white balance on the f3 in s log mode kind of bites, there was a lot of mix color temp situations that the Alexa ate alive, but there was plenty of room to dial back in the right white balance off the f3 with ki mini.

:p
Was the F3 footage from the Ki mini? Was the Alexa also Ki mini?
 
So there was a little bit more to be had out of the F3 at 444 with a Gemini or similar.

Not bad for a camera at this price point.
 
10bit capture is more important by a factor of 10, than 444 over 422.

Honestly, the best way, is to record to a piq240 to dnxhd. It uses the full 110ire of the signal. Pro res only holds up to 100ire. That extra stop is clutch.

We loaded some of that prores footage from the F3/alexa shoot into FCP and were able to pull that extra stop out of the +100 ire sections of the signal. The alexa/F3 matched even more, but the real difference is what happens at 109IRE. The alexa rolls into white a little more smoothly than the F3 does @ its clipping point. That is not to say that it looks ugly, infact when it does fnally burn out, the F3 turns white, and turns white smoothly. Its just that the when the alexa turns white that 1/2 stop of the (14.5 stops) gives it a smoothness cushion. Its just extra silky and beautiful.

The alexa is remarkable.... its got so much range, that, all the work we do, flagging silking, netting, filling, etc to control contrast down to a workable i-10stops, becomes almost moot. It can see 14stops... so you kinda don't want to do as much to the scene because it just looks right when you turn the camera on.

Its hard to explain, but seeing it in action, you'll understand. I truly feel the Alexa is the best digital camera in the world right now. The fact that the F3 gets within a breath of it.... just remarkable.
 
Timur, Do you have a Pix240 yet? I was promised an eval after IBC.

I may have mentioned this but at the Abel F3 training I was able to capture a brief S-Log clip to my BMD HyperShuttle. While just4:2:2 (albeit uncompressed) just some very simple grading with Colorista II was beautiful.
 
Timur, Do you have a Pix240 yet? I was promised an eval after IBC.

I may have mentioned this but at the Abel F3 training I was able to capture a brief S-Log clip to my BMD HyperShuttle. While just4:2:2 (albeit uncompressed) just some very simple grading with Colorista II was beautiful.
I don't, but its on order. Pro res DOES record the superwhites, but it doesn't display them in DaVinci. You have to go to FCP bring it down then rencode, wich can be destructive. Recording to dnxhd, or uncompressed, preserves the signal in its glory. Without the need to recompress. I even think it records to .mov files when doing dnxhd.

Not sure ned, but does the hyprdeckling record 3G444? Cause the f3 can out put 3G444 and its beautiful.
 
it is, but that extra stop of highlight retention really separates the men from the boys. we had a hallway shot with 3 windows that was clipping the hell out of the F3, but Alexa ate it up. It saved a ton of time not having to net or bring down the windows in other ways, and we barely had any fill too. plus there is the more polish in the actual physical hardware of the Alexa that makes you kind of sigh when you pick up the F3 and have to frankenstein it a bit. but again, talking about 60 thousand dollar price variance. they A and B cam very well, and the F3 def feels like the baby brother to the Alexa, but the baby is still a baby, people shouldn't forget that.

Well, it does outperform the Red One MX in dinamic range, doesn't it? How is the sharpness of the footage? Definition compared to Alexa?
 
Back
Top