Proper Exposure with Slog2 and the Odessey 7Q

cuervo said:
My exposure process really uses the zebra settings on the O7Q set to 108%, combined with turning on the LUT and the WFM. With the LUT on, I set exposure so the highlites are just kissing the 109% limit on the WFM. This method seems to yield the best results.

How is everyone else doing it?
Click to expand...
Hm u are exposing ETTR by looking at a 709 LUT. And then recording Slog2? U will underexpose by around 2 stops or am i missing someting?!?!
 
Mitch Gross said:
When the custom LUTs become available you will be able to alter exposure level.

In rating the FS700 at 800, I don't do anything to it. The RAW signal at ISO 2000 and SLog2 goes to the Odyssey7Q. I still have tons of room in my highlights, as evidenced by the various exposure tools on the Odyssey. But I still use a proper lightmeter to design and preset lighting, as I have done for more than two decades. With that experience, I actually don't need the tool all that much. I know when I place lamp X at a certain distance it will deliver me T-stop Y given ISO Z. I've lit entire features by eye with a consistent stop on the lens, and even when the Gaffer insists on doublechecking I'm rarely off by more than 1/10th of a stop. It's just something you get used to with a trained eye and experience.

All that said, I consistently prefer the noise level and performance of the FS700 when I rate it at 800 ISO while shooting RAW & Slog2 to the Odyssey7Q. That's just my taste, but it works for me. The shadows tighten up and I still have plenty of range in the highlights. Technically it's 14 stops but usably it's more like 11.5-12. I chop off a similar number of stops on the bottom for other cameras as well, so again this is a consistent thing for how I like to expose for RAW and/or Log images.
Click to expand...


Mitch , does the FS700 have more stops of dynamic range in slog2 than the F3 when using the Odyssey 7Q?
 
Im playing around with the odyssey and there are some questions occuring. How do u expose Slog2? U cant get the zebras under 70% on slog2 they need to be around 52% on skin tones. It seems to be impossible to show any under exposure under 14% in false colors. And u cant change yellow and red under 70%. But Slog2's white is at 59%. So the question is how do u expose slog2? Or is it only possible to expose for the rec709 lut and then manually underexpose by two stops? Would be nice if someone could explain it to me because i didnt understood the those tactics from page1. thanks
 
Baddgsx said:
Mitch , does the FS700 have more stops of dynamic range in slog2 than the F3 when using the Odyssey 7Q?
Click to expand...
Yes, I believe the F3 is just shy of 13 stops in Slog and the FS700 is 14 stops in Slog2. But those stops at the bottom have more noise so you may choose to crush them away, so then effectively they are about the same.
 
SKYLIKE said:
Im playing around with the odyssey and there are some questions occuring. How do u expose Slog2? U cant get the zebras under 70% on slog2 they need to be around 52% on skin tones. It seems to be impossible to show any under exposure under 14% in false colors. And u cant change yellow and red under 70%. But Slog2's white is at 59%. So the question is how do u expose slog2? Or is it only possible to expose for the rec709 lut and then manually underexpose by two stops? Would be nice if someone could explain it to me because i didnt understood the those tactics from page1. thanks
Click to expand...
We will be expanding the ranges for the zebras and the false color modes. The waveform reads accurately and we will add reference marks to the histogram.
 
Mitch Gross said:
Yes, I believe the F3 is just shy of 13 stops in Slog and the FS700 is 14 stops in Slog2. But those stops at the bottom have more noise so you may choose to crush them away, so then effectively they are about the same.
Click to expand...

so is there a good chance that the FS700 uses the same sensor as the F5? since the F5 is rated at 14 stops.
 
Is that, in fact, a 709 LUT? No one has really said what it is. And, since I first posted, I've noted that that process I described is not working. I'm consistently getting grain in the shadows, implying that I'm underexposing, then trying to bring the shadows up and generating too much noise.
 
cuervo said:
Is that, in fact, a 709 LUT? No one has really said what it is. And, since I first posted, I've noted that that process I described is not working. I'm consistently getting grain in the shadows, implying that I'm underexposing, then trying to bring the shadows up and generating too much noise.
Click to expand...

Yes something is way off here. My post #2 in this thread outlines how I get good results. Don't let anything below 30% on the WF on the odyssey, and don't worry about seemingly blown highlights on the WF they are fine and can be recovered. This will really over expose the internal 700 recording but make a very usable recording as 2KRaw Slog2 in the O7Q. I don't know if it is Sony or how CD interprets the Sony output, but it is not like RAW in other cameras at all, the blacks are very noisy if you expose as is generally recommended for RAW linear, basically unusable.

The unfortunate result is that the internal recording is not really good to use as proxy footage. But the O7Q footage looks great recorded this way. Strange...
 
It's not so strange if you are intimate with how certain manufacturers have historically done things.

As I stated before, I expose the FS700 as if it were 800 ISO, even though the base RAW exposure is set to 2000ISO. So 1.25 stops over. This would normally put me pretty near the clip point (that's 2.5 stops over in REC709), but not so high that my exposure in a REC709 is unrecoverable in the mid tones. The highlights will clip out, but isn't that the point of the difference between shooting REC709 & shooting RAW? But by saturating the sensor more the result is that you'll naturally pull the levels back down some in post once you begin to throw on a grade, and those noisy shadow areas will drop down into the darkness. The results can be quite nice. Proxies recorded REC709 will sufer blown highlights, but you can't get something for nothing.

Highlight recovery is where Sony sensors show their strength. And highlight recovery is what sets RAW workflow and real Digital Cinema work apart from old school clippy video. These silly stories about shooting under street lamps and not needing any lights are all just gimmicky marketing nonsense. Except for some very particular technical need I really don't care about that sort of cinematography. But rescuing detail in my highlights, with bright skies, daytime windows when shooting interiors,hard sunny days with huge contrast and dynamic range within the image -- that stuff is real world situations and highlight recovery can make 'em beautiful. Don't starve your sensor of light and then suffer a noisy image. Blast it a bit and let the DR do it's magic.
 
Baddgsx said:
so is there a good chance that the FS700 uses the same sensor as the F5? since the F5 is rated at 14 stops.
Click to expand...
Yes, but there's a LOT more to a camera than just the sensor. Note that the RAW from an F5 is 16-bit depth while from the FS700 it is 12-bit depth. This contributes to lesser performance at the very bottom of the exposure range.
 
cuervo said:
Is that, in fact, a 709 LUT? No one has really said what it is. And, since I first posted, I've noted that that process I described is not working. I'm consistently getting grain in the shadows, implying that I'm underexposing, then trying to bring the shadows up and generating too much noise.
Click to expand...
I assume you are referring to the monitoring LUT that can be applied to the deBayered RAW image on the Odyssey7Q. Yes, it is the REC709 LUT provided to Convergent Design by Sony, and it matches the one in the camera itself.
 
Hm still experimenting with exposure. What is it exactly what we are getting there in the .dng's? How is the slog2 baked in and do u need to expose for slog2 or a linear raw? I just did a test where the luminance would be about 130IRE in wfm (if this would be possible, with LUT on it would be 150 or so) and i could retain everything with highlight recovery on resolve. Its very strange because it goes against all i know. Who can u trust .... the exposing tools, the values or ur experience. If u would expose for slog2 the mids would be at 30 and the highs about 60 but what is going on when u can go till 130. Always thought 109 is that 1300% slog2. Are the highs over 60 more compressed? Linear or not. What GAMMA do we have here?!?! Please someone! Im confused.
 
Are you setting the camera to Slog2 in the Picture Profile of the camera? What you are seeing is 14 stops of dynamic range! In REC709 you have 2.5 stops over key and you clip. Here you will have 4.5 more stops to go, plus seven below. That's why I still say to overexpose some. You have huge room in the highlights so you won't clip them. When you darken the image in post you will take all the well exposed mid and dark tones and bring them down into darkness, taking the noise floor down with you as yu go. The result is retained highlights with low noise shadows and midtones.

On the Odyssey7Q you have two ways to view the image, with and without a LUT applied. Assuming you have Slog2 on and other settings such as gain switched off on the camera, the image without the LUT will be the equivalent to Slog2. Turn on the LUT and it is REC709.
 
What Alister was saying about exposing for S-Log2 is that with each f-stop going farther up into the highlight range, the amount of information doubles with each stop but that cameras can't record log that way because they would quickly run out of memory. So instead, each additional stop is allocated fewer shades of gray to record, meaning that data that are pushed into the higher stops are getting highly compressed.

The practice of exposing log gammas for middle gray rather than for the ends is in recognition that if skin tones are pushed up into the highly compressed range where there are fewer shades to represent it, it will look plastic and be hard to recover the realistic tones when bringing the levels down in post. I agree because Sony have appended the middle gray point into their nomenclature for hypergammas, ex. HG800932 means 800% at 109% with mid gray at 32%.

Under this system, white (card) is at 59% instead of 80% of standard gammas.

So if you expose to the right, you risk putting skin tones into the highly compressed regions where fewer gray shades are available to represent it. It isn't just about having and using using all the available dynamic range, but being careful about *what* gets pushed into the highly compressed regions.
 
Last edited:
Tom Roper said:
What Alister was saying about exposing for S-Log2 is that with each f-stop going farther up into the highlight range, the amount of information doubles with each stop but that cameras can't record log that way because they would quickly run out of memory. So instead, each additional stop is allocated fewer shades of gray to record, meaning that data that are pushed into the higher stops are getting highly compressed.

The practice of exposing log gammas for middle gray rather than for the ends is in recognition that if skin tones are pushed up into the highly compressed range where there are fewer shades to represent it, it will look plastic and be hard to recover the realistic tones when bringing the levels down in post. I agree because Sony have appended the middle gray point into their nomenclature for hypergammas, ex. HG800932 means 800% at 109% with mid gray at 32%.

Under this system, white (card) is at 59% instead of 80% of standard gammas.

So if you expose to the right, you risk putting skin tones into the highly compressed regions where fewer gray shades are available to represent it. It isn't just about having and using using all the available dynamic range, but being careful about *what* gets pushed into the highly compressed regions.
Click to expand...
This is true and the "compression" (actually more accurately a smaller allocation of bits) of the highlights is apparent if one was to try to put skin tone at 70% as one would in REC709. But that would mean that you'd have to overexpose Slog2 grossly, by perhaps four stops. That's not understanding how log images works, and it is also rather different than Hypergamma, which is an extension of the highlights, not a true log image.

I'm talking about overexposing perhaps 1.3 stops, and I say it's useful in order to hide the noise in the bottom of the exposure range when it is brought back. As with everything, there is subtlety in all measures. My suggestion will reduce dynamic range slightly as the very bottom of the exposure range will be crushed away, but it will also tighten up the grain. And the allocation of bits will not particularly effect the image with this modest an overexposure.
 
Mitch Gross said:
...and it is also rather different than Hypergamma, which is an extension of the highlights, not a true log image.
Click to expand...

No problem with that, only mentioned Hypergamma because middle gray is now appended to thenomenclature for illustration. S-Log2 doesn't have the middle gray specified but it is said to also be IRE 32. Agree with you, there is latitude on both sides of the middle gray point for skin tones to be accommodated by modest overexposure.
 
It's interesting to see that the new SLog3 is more of a cineon curve (more film like). It raises the mid grey up even further which allows for the potential of more detail to be recorded in the shadows and mids. Of course you can do that by exposing slog2 higher (to a certain degree anyway)

AFAIK middle grey for Slog2 is specified in one of the Sony PDF documents, i've written about it here previously but i can't remember off the top of my head where the document was.

Mitch is bang on with exposure in Slog IMHO, there is excessive noise in the shadows on the 700. We can see that in the normal 8 bit profiles too. I think this is as a result of a pre F5 generation sensor but more importantly only a 12 bit linear signal path. Both the math of linear and Alistars side by sides with the F5 indicate poorer shadow performance.

The poorer shadow performance also restricts the dynamic range too, it's mush down there so you just need to be mindful at which point you want your shadows and mid tones to be. Correcting for that afterwards should be a simple affair of a reverse Slog2 curve or LUT then an exposure adjustment. NOT exposure adjustment then reverse curve though.

There are also posts i've written around here that show the effects of unrolling the standard curves properly before grading and the kind of different they make. You could even unroll skin tones to a slight less plasticy look if you tried hard enough.

I would love to see the LUTs the O7Q can apply (for monitoring) include versions that do allow us to expose higher in SLog2, that would be very useful IMHO

cheers
Paul
 
Hey Mitch as u rate the fs700 at 800 would it be possible to give the 7Q some kind of EI gain? I wouldnt bet on Sony!
 
You must log in or register to reply here.
Back
Top