FS7: Slog2 or 3

No, I have not tried Glimmerglass but I've heard a lot of good things about it. It would be fun to shoot a bunch of test footage with classic filters and today's modern 4K cameras and cine lenses. Maybe even take that to the next step by then comparing the results to Tiffen's post software. Not enough time in the day!
 
Doug Jensen said:
Why do you say S-LOG3 is easier? In my experience they are both take exactly the same amount of effort. Why is one harder than the other for you to deal with?

BTW, if someone clips specular highlights, back-lit hair, sides of faces, etc. at 94 IRE with S-LOG3 there's no getting it back in post.
Click to expand...

LennyLevy said:
Greg, what do you mean exactly when you say Slog2 is harder to grade. I've heard this many times so I'm not arguing with you just wondering what this means in practice.
I won't be grading this , the editor will be and I believe she is working in Premiere not Resolve. Will that make a difference?

I am concerned abut the clipping that Doug mentioned since that's what I seem to see in my tests , but as I mentioned I need to look at them more carefully
Click to expand...

I think it honestly has more to do with Sgamut3.cine vs Sgamut than it does with the gamma curves themselves, Sgamut is so broad that sometimes it can take quite a bit of work to wrangle intense greens or reds down to where you want them sitting in the grade. Whereas Sgamut3.cine's intentionally restricted gamut, gets you to a happy place with a minimal of fuss.

I think to date, the best results I ever got out of the F5(5) came from shooting raw, then processing the footage as SLOG2 Sgamut3.cine in Davinci. Unfortunately this isn't a combination that's available for compressed recording - in which case, you're forced to make a choice, and given that choice I think Sgamut3.cine is of greater value than SLOG2.

The slightly-earlier clipped highlights with SLOG3 are only the absolute top-end of the highlights, so I don't think it's such a big deal honestly - certainly it's less of an issue than Sgamut can present in grading.

I personally never had any great issues with SLOG2 compared to SLOG3, but empirically, it would seem that the vast majority have had them. So unless you're the only person who'll be touching the footage in post, I think the SLOG3/Sgamut3.cine combination remains the safest (by a fair margin) for achieving consistently strong results under a more conventional post workflow.

And a simple workflow matters. Switching between gamuts and gammas mid-project strikes me as sheer madness - any minute gains in image quality would be instantly and painfully offset by all kinds of messiness trying to match things in the grade.
 
Doug Jensen said:
Clipping is absolutely a concern with ANY camera when you have specular highlights, backlit hair in the sun, sparkling water, sun low on the horizon at sunset, etc.
Click to expand...
Of course it's a concern, but it's so easy to spot, as long as you're not relying on an mLUT, that you should be able to take appropriate action or let it blow if that's what you want.
 
It's clear you don't understand my point. Just because it is easy to spot does not mean there is something that can be done about it. Not everything is within our control when shooting outdoors, thus, my choice to use S-LOG2 most of the time outdoors.
 
Back in ther day there was an image of data levels per stop in 2 and 3 - 3 has consistent data levels per stop which means that it's gonna grade more intuitively - 2 is going to force you to put your important bits at certain levels which may not be where you want them just like the old days of keeping skin out of shoulder when shooting a sony baked look?
 
A bar in this chart is data levels allocated to a stop of light level below clip. The ideal is to have a high "bar" in each stop. Bars of consistent height make a file that grades in a linear manner aka like a raw still in Lightroom

https://goo.gl/images/Mo3x58 (Pic)

This shows 2 to have about 7 excellent stops while 3 has about 9 adequate ones?

Hence 2 would be better in a narrow dr scene and fully ettr.

The image may of course be hocum.

The fact each stop in 2 is quite different means that the file will not grade well under mild variations in exposure because the a subsequent added curve will do wonky things.
 
Last edited:
Intuitive grading

We see that many stops in 3 have a value of about 80.

If one wants to lighten or darken by a stop one adds or subtracts 80 to the input value. A linear action.

But with 2 one must add or subtract a different value for every input value - not linear or intuitive
 
Doug Jensen said:
It's clear you don't understand my point. Just because it is easy to spot does not mean there is something that can be done about it. Not everything is within our control when shooting outdoors, thus, my choice to use S-LOG2 most of the time outdoors.
Click to expand...
Of course I understand your point, and I know you shoot slog2 outside. I'm sure that works really well at the air shows and nature parks you spend your time in. Lenny here is talking about a controlled lighting situation. And in a controlled lighting situation you can always do something to address exposure issues.
 
Only somewhat controlled, Not as much as I would like though. it will be a struggle to get as much fill light as I need with my budget and the nature of the situation. However I have lots of good suggestions here and just need to do a a bunch of testing. All stuff I've been meaning to do for ages anyway.
 
LennyLevy said:
They are often under lots of tall trees so there is lots of mottled shade and sunlight and occasional patches of bright sun. In the worst scene,much of the background is getting direct sun and has both light green foliage and lots of tall dried out grasses that are bright as hell. Likewise the ground around and behind us is often dirt and dried (i.e. bright) grasses , so the spots that the sun hits are quite bright
Click to expand...

@ Liam

If that's your idea of "controlled" lighting I'd sure like to see what you define as uncontrolled. :-)
 
Hey OP Lenny - if you have plenty of time to experiment - I might suggest shooting a test using both choices of the actual scene and then perhaps try creating a LUT in lut calc to do the rec709 conversion. You could bring a preview of the actual shot into LutCalc and it now offers some interesting options under "highlight gamut" , "knee" and "highlight level".
Lot of it goes over my technical knowledge, but seeing a real-time preview might be quite helpful ahead of actual shoot. And of course this might even give you a "get in the ballpark" LUT to hand off to whomever does the grading. Just a thought... I plan to do some of my own testing on this soon enough...
 
Thanks Mark, excellent idea, I didn't know about those additions to LUTCalc.
This thread is a model of what happens when people don't hijack DVXUser with nonsense. Sometimes I think "why am I wasting my time on this site" and am even embarrassed to recommend it, but when it works it can be very valuable. Thanks for all the suggestions everyone.
 
Gentlemen,

However interesting in many ways, this thread is escalating. And measured by the numbers of up and downvoters two or more camps arise. Will this end in a slog3 vs slog2 war?

Did anybody try to find the solution with other options than slog2 or 3, lights, diffusors, more lights, links to courses, making creative insults and, and discussions about good old filters?

In that case. Try to think out of the box, can I suggest the 'chainsaw method': cutting trees until the light is good enough?

So far my hijacking of this thread, keep on the good work!
 
Last edited:
Hi LennyLevy,

I can see from your original post mentioned that dancers are in shade. I think you said there were some background elements that were being hit by direct sun. Is that correct? Also, I can imagine there might be dapples of light coming trough the canopy of trees, which might hit the dancers and the foreground. But even if there isn't...

If you were thinking of ultra-bounce, then maybe you have the crew and time to set up an 8x8 or 12x12 bounce, with a 1.2k hmi or a 1.8k him bouncing into it from either the side or 3/4 overhead. This might be enough to boost some levels and add contrast to your dancers. Alternatively, you can use two Hive 1K fixtures and use a smaller generator.

If you have it in the budget, go with a really fast lens (T1.3 or 1.5ish). This way you can open up all the way, knocking a lot of stuff out of focus. The more telephoto the lens, the more out of focus too. Doing so will allow you to hide stands and scrims more easily in the background and such, since the background will be so blurry you will not be able to distinguish it from the background. Depending on your location, you can hang a scrim over where the background is seen by the camera, knocking the background levels to suit. If done right, the scrim fabric (20x20?) is hidden out of focus, and you don't clip. But not all locations would allow this, since depending on how big the "window" to the background is, the harder it is to scrim. Even if you can't scrim the background, if the clipping happens out of focus, it rolls into the highlight clip more creamy. This will at least help you deal with the excess DR by allowing your background to clip, but because it's all blurry, you roll more smoothly into the whites. However, this doesn't help you with foreground I -focus clipping.

All the ideas of mirrors into bounces are smart too, depending on your location. If the mirrors and bounces have to be too far away from their source and/or final destination, inverse square law will minimize the transfer of light onto your dancers.

As far as camera goes, are you able to shoot RAW? If so, that would allow you to have a bit cleaner shadows, with more color info. The 10bit log, whether 2 or 3, will give you slightly less quality across the latitude...since the highlights and the shadows are less compressed. Actually, the whole thing is less compressed.

So yeah:
-Rent one or two or three Master Primes, Cooke S5s, or even the Sigma primes.
-Rent a 8x8 or 12x12 with ultrabounce, stands and everything you need to secure it.
-Rent bright fixture (M18 or two Hive 1k)
-Rent generator.
-One gaffer, two grips.
-Shoot RAW if possible.

But ultimately, no matter what you decide to do, as someone already said, the trick is to test. Hopefully you can share them here.

In any case, best of luck!
 
Last edited:
Thanks, Rafael, though as mentioned earlier - my background is as a gaffer so I don't need advice on how to light - over 30 years in the biz. No access to large HMI's - 2 joker 400's w 1 K generators, an 8x8 griff., 4 super soft Mattboards ( which seem the most promising ) and a few LED's that actually have a bit of punch under the canopy. No way in hell could I scrim the background and I have fast lenses. No Raw with FS7. Crew of one experienced Grip and one entry level guy. Test test Test - always been my philosophy. Need to be limited, smart and fast. I'm going to be OK.

I do have a question though - With Slog3 and Slog 2 I always overexpose by a stop or more. So far I'm planning to shoot at 1250 for highlight latitude, but am wondering how safe it is to go higher EI. I don't want to tax my Post person with the need for noise reduction and gather its not a good idea in Log to do it in camera (can't remember why though). I've never tested for noise much myself.

Lenny
 
LennyLevy said:
I do have a question though - With Slog3 and Slog 2 I always overexpose by a stop or more. So far I'm planning to shoot at 1250 for highlight latitude, but am wondering how safe it is to go higher EI. I don't want to tax my Post person with the need for noise reduction and gather its not a good idea in Log to do it in camera (can't remember why though). I've never tested for noise much myself.
Click to expand...

You said "test, test, test". Isn't this exactly what you would test?
 
You must log in or register to reply here.
Back
Top