Post Prod - workflow - Slog into NLE
- Thread starter Shooter
- Start date
slipstrea808
Well-known member
nyvz
Veteran
Only if you use a monitor that supports LUTs. Whether in avchd or 7Q prores, I expose SLOG2 to the right as much as possible, aiming for around 1-3 stops overexposed. Usually about +1.7EV looks 'correctly' exposed for SLOG2 which has an unusually low manufacturer specified middle grey point. For my NLE I usually use premiere pro and use Vision Color's ImpulZ LUTs, usually the Kodak Vision3 50D DP FC LUT. I'll usually just try a couple of my favorite LUTs on a good representative shot and find a look using a Colorista 2 effect before the LUT effect and then copying that to all the clips on the timeline.
austinmartifilms
Active member
According to this guy, that approach is all wrong.
He says:
You can and should expose according to the LUT. If your footage looks over exposed when you activate the LUT, then this means your SLog2 is also over exposed. The LUT performs a very accurate and precise conversion from S-Log range to 709 range. As the monitor is a 709 device you really need to be viewing in 709 range to make any kind of accurate exposure judgement, unless using a light meter or grey card and waveform.
It must be remembered that the exposure levels for S-log2 are very different to the exposure levels for standard gammas. Middle grey for S-Log2 is only 32% compared to the 45% of normal gamma. White for S-Log2 should only be 59% compared to the 85-90% of standard gammas. Faces and skin tones for S-Log2 should be approx 45-50% compared to the 65-70% of standard gammas.
The entire S-Log2 range is recorded at a much lower level to make room for the extended 14 stop range. If you over expose SLog2 then you will find:
1: LUT's designed for SLog2 will make your footage look over exposed (because it is).
2: Grading of skin tones and highlights will be difficult as they are in the wrong part of the log curve.
3: Skin tones may look washed out or lack contrast after grading (unless you apply an exposure correction LUT in post). Or may never look good if you are more than +1 stop over exposed.
4: Highlights will lack contrast, even after grading.
5: After grading the image may look plasticky and not as good as if shot with a conventional gamma.
It is MUCH better to expose SLog2 correctly, your finished results will look much better even though things may look dark when shooting. Invest in a grey card and white card and learn how to use them correctly. You can over expose SLog2 by a small amount (+1 stop max with 8 bit), but even at +1 stop skin tones will still be darker in SLog2 than conventional 709. I believe that the majority of people tend to over expose SLog2 and as a result do not get the best results that they could. Shooting with Slog require a new way of thinking about exposure, especially if you try to judge your exposure by looking at the log image on a conventional monitor. Much better to use a LUT to convert from log space to normal 709 space as then you can simply expose by eye using conventional levels.
-Alister Chapman
Alisters explanation is good - particularly the bit about faces and skin tones. Works out perfectly if you look at the WFM or false colour on the O7Q and switch the LUT on and off while monitoring.
Would be great if CD or someone else could write an in depth guide to using SLog2 and RAW recording modes, talking about exposure etc. would really help out those new to Log recording.
Would be great if CD or someone else could write an in depth guide to using SLog2 and RAW recording modes, talking about exposure etc. would really help out those new to Log recording.
nyvz
Veteran
Yes I'm aware of Alister's preferred methods, and perhaps they are important if you are using Sony-designed LUTs that are designed around their specification that SLOG2 middle grey falls at 32%. In my experiences such LUTs risk bringing up a lot of noise, which is not surprising since exposing SLOG2 "normally" is equivalent to underexposing CINE4 by -1.7 stops as they both run the sensor at the same hardware gain. However, the LUTs I mentioned are not designed to bring 32% up to 45%, and when applied to SLOG2 footage that has supposedly overexposed by about +1.5 stops, the resulting image looks correctly exposed, and to get "correctly" exposed SLOG2 would require a pre-LUT LOG exposure adjustment of over +1 stop.
The whole point of SLOG and any LOG is that it has a consistent response through most of it's range so it is easy to adjust exposure in post, so exposing to the right and using the full DR available is much easier to do than with a "video/709" gamma which tend to have a big knee and a less consistent slope throughout it's range where you risk putting certain details in an area of the curve that reacts very differently than the middle of the curve.
The whole point of SLOG and any LOG is that it has a consistent response through most of it's range so it is easy to adjust exposure in post, so exposing to the right and using the full DR available is much easier to do than with a "video/709" gamma which tend to have a big knee and a less consistent slope throughout it's range where you risk putting certain details in an area of the curve that reacts very differently than the middle of the curve.
Last edited:
cuervo
Veteran
I'm using the same process as Noah. I would agree that ETTR is giving me much better 4K/Slog2 results than any other approach. It's scary how much noise I get in post if i don't ETTR. Curiously, white faces are coming out pretty much at 50% IRE. Argue with me, as you wish, but, this is what's working, here. ETTR means, simply, that the right hand edge of the histogram is between 100-109% IRE. Under no circumstances would I recommend exposing beyoyd 109%. Blown hilites are blown hilites, under anyones definition of uglishe.
austinmartifilms
Active member
I have to say I feel that my footage has been fairly noisy at times. Lately I have been exposing a little higher than what looks right in the LUT.
Can't think of anything to argue about
So 50% IRE on the image with NO LUT applied on the O7Q?
cuervo
Veteran
DOH!!But 50% is what Alister was recommending - so you're doing what he suggests!... "Faces and skin tones for S-Log2 should be approx 45-50% compared to the 65-70% of standard gammas" ...and if you apply the LUT, you magically get around 70%
nyvz
Veteran
They would fall wherever they are. I'd keep them well above middle grey, and it will likely depend a lot oh the way you light and shoot and the type of content. The idea that skin tones should alwayd be exposed to the same value seems odd to me and perhaps only relevant if you only light and shoot talking heads for the news, otherwise there are so many kinds of lighting situations that don't call for the same exposure.
Basically I usually expose slog2 so that it is as bright as possible without anything in the scene clipping except maybe speculars or sources. This always gives you as much DR and as little noise as possible If your slog2 looks underexposed on your monitor/lcd, it will probably be noisier than you want. If you are using a particular LUT, you may want to expose specifically for that, but if you expose as I've described to maximize DR and minimize noise, you can always adjust the exposure in post before the LUT pretty easily. That is the beauty of LOG.
The way Alister outlined means always exposing for middle grey with 7stops of DR over middle grey. That means most of the time you'll be exposing in a way that doesn't use several of the top, cleanest stops of the camera's DR and instead putting most of your image in a range with a lot of noise. Generally 7stops of DR over middle grey is useful for keeping window detail in day interiors but is overkill for even most day exteriors and night exteriors, and massive overkill for overcast day exteriors and could leave you with a lot of noise in night int/ext's.
slipstrea808
Well-known member
I like the false color tool a lot actually! So my table for anything FS700 SLOG2 Raw WITHOUT a LUT is outlined here. I'll talk based on what I'd see in front of the camera, what the O7Q's screen looks like and what I'm doing with it in post. Depending on what I need to do there are a few options for workflows but (especially until Adobe fixes the Magenta highlight issue) this is all based on a workflow with 3 things. 1) The footage is brought in and colored in Davinci Resolve Lite. 2) No odd plugins. Just coloring. I'm doing NOTHING to remove noise other than to avoid it or bury it in the blacks. 3) I'm not exporting footage until the first light and main coloring is done. Soooo...
Purple <15% - This is stuff I don't care about being black. As I'm looking at it in person it usually IS black if not shadewed into the black and you would have to walk up to it to make out details . But anything here is in that NOISEY black range on the 7Q (more on this later). Closeup you can make out some detail but barely any. I'm expecting no data out of it useable. Truth be told I'm actually never even leaving this stuff as is in coloring in post. 90% of the time pushing it LOWER to get rid of that noise or to just add the right amount of contrast to the image.
Blue <30% - This stuff is either very dark or actually (depending on lighting conditions) black. But in person you can see details of course. The monitor will show detail. You can see details in these very dark areas. In post this stuff might or might not get pushed down. But there is a choice at this point. Also there is data for coloring changes. But be careful. You cant really stretch much of this into the mid tones otherwise it will get noisey.
Dark grey <50% - Useful information here. Looks just darker than normal caucasian skin tones. Perfectly viewable in person and in the monitor. I find 18% grey falls at the very top of this range. In post you can do nearly whatever you want with this range til pushing it above 80IRE. Then I'd say expect some issues. But still that is a lot of leeway.
Green 50-60% - Ahhh the sweet spot! I aim skin tones here so in person... this is the middle. On the 7Q this is the most desaturated compared to what thigns actually look like. If the lighting is flat I will try and adjust the iris or lighting so that most all of a person's face is in this range. If lighting is strong to one side I'll try and draw a stripe across the middle of the dark to light area. As in yes there will be a green false coloring stripe on a person's face. Then I'll adjust accordingly here or there to make it right to my needs. In post you can do whatever. The data is there and you can crush it or bring it to the top. It wont get noisy or anything.
Light grey >60% - Still normal looking in person this is just the lighter stuff. Exposing for skin this will be the lighter skin, light taupe and off white walls inside (with flourescent office lighting). It will look washed out in the monitor but it will be fine in post of course. You can really bring this stuff down dark if need be and all the info will be there ready for saturation. Also lighter skin tones will fall into here depending on lighting and iris settings.
Yellow >90% - This is where stuff is getting white actually as you look at it. Or brighter for sure. The monitor will make this stuff look white or very desaturated and light. In post, though, since we are using linear RAW there is a good amount of info here! You can pull this stuff down and saturate it. You can't go too far but this range can go down to 50IRE in Resolve and look fine! Coloring isn't limiting either. Good amounts of data. But I would really keep this area to what you KNOW are brighter areas. And light/iris the shot accordingly.
Red >105% - So at this point you are looking at white or the sky on a bright day or you are indoors and this is the outside. The monitor really won't show much of anything. Outlines and a hint of a color if there is one present, but you won't see what your eyes see. But in post, you'd be suprised... That linear RAW... There is data up there. This is where instead of going to white you can recover it a bit an roll off in the highlights slightly. Good coloring and experimentation work here. Most people watching wont notice but we the filmers/DP's/colorists will go "oh wow". Just that bit of something extra. Choices are important here. This is where I usually say OK I'm not worried about losing these highlights, or it is what I want, or lets adjust the iris/lighting to get it right. I don't recommend feeling safe about this range for anything. Don't think "oh I can pull it down in post..." Sure you can a bit but it might not be as much as you want. Like I said, its mostly just for nice highlight "roll offs".
OK wow lotsa typing! That is just a general guideline I've found that works and keeps me "out of trouble". But experimenting and tweaking is always a good thing. This seems to be a safe starting point. I'm actually thinking of making the green zone 55-65% to be honest. I'd love to hear what others are doing. I've just wanted to write that up for a while.
Purple <15% - This is stuff I don't care about being black. As I'm looking at it in person it usually IS black if not shadewed into the black and you would have to walk up to it to make out details . But anything here is in that NOISEY black range on the 7Q (more on this later). Closeup you can make out some detail but barely any. I'm expecting no data out of it useable. Truth be told I'm actually never even leaving this stuff as is in coloring in post. 90% of the time pushing it LOWER to get rid of that noise or to just add the right amount of contrast to the image.
Blue <30% - This stuff is either very dark or actually (depending on lighting conditions) black. But in person you can see details of course. The monitor will show detail. You can see details in these very dark areas. In post this stuff might or might not get pushed down. But there is a choice at this point. Also there is data for coloring changes. But be careful. You cant really stretch much of this into the mid tones otherwise it will get noisey.
Dark grey <50% - Useful information here. Looks just darker than normal caucasian skin tones. Perfectly viewable in person and in the monitor. I find 18% grey falls at the very top of this range. In post you can do nearly whatever you want with this range til pushing it above 80IRE. Then I'd say expect some issues. But still that is a lot of leeway.
Green 50-60% - Ahhh the sweet spot! I aim skin tones here so in person... this is the middle. On the 7Q this is the most desaturated compared to what thigns actually look like. If the lighting is flat I will try and adjust the iris or lighting so that most all of a person's face is in this range. If lighting is strong to one side I'll try and draw a stripe across the middle of the dark to light area. As in yes there will be a green false coloring stripe on a person's face. Then I'll adjust accordingly here or there to make it right to my needs. In post you can do whatever. The data is there and you can crush it or bring it to the top. It wont get noisy or anything.
Light grey >60% - Still normal looking in person this is just the lighter stuff. Exposing for skin this will be the lighter skin, light taupe and off white walls inside (with flourescent office lighting). It will look washed out in the monitor but it will be fine in post of course. You can really bring this stuff down dark if need be and all the info will be there ready for saturation. Also lighter skin tones will fall into here depending on lighting and iris settings.
Yellow >90% - This is where stuff is getting white actually as you look at it. Or brighter for sure. The monitor will make this stuff look white or very desaturated and light. In post, though, since we are using linear RAW there is a good amount of info here! You can pull this stuff down and saturate it. You can't go too far but this range can go down to 50IRE in Resolve and look fine! Coloring isn't limiting either. Good amounts of data. But I would really keep this area to what you KNOW are brighter areas. And light/iris the shot accordingly.
Red >105% - So at this point you are looking at white or the sky on a bright day or you are indoors and this is the outside. The monitor really won't show much of anything. Outlines and a hint of a color if there is one present, but you won't see what your eyes see. But in post, you'd be suprised... That linear RAW... There is data up there. This is where instead of going to white you can recover it a bit an roll off in the highlights slightly. Good coloring and experimentation work here. Most people watching wont notice but we the filmers/DP's/colorists will go "oh wow". Just that bit of something extra. Choices are important here. This is where I usually say OK I'm not worried about losing these highlights, or it is what I want, or lets adjust the iris/lighting to get it right. I don't recommend feeling safe about this range for anything. Don't think "oh I can pull it down in post..." Sure you can a bit but it might not be as much as you want. Like I said, its mostly just for nice highlight "roll offs".
OK wow lotsa typing! That is just a general guideline I've found that works and keeps me "out of trouble". But experimenting and tweaking is always a good thing. This seems to be a safe starting point. I'm actually thinking of making the green zone 55-65% to be honest. I'd love to hear what others are doing. I've just wanted to write that up for a while.