Sony_A6300

Using S-Log2 and S-Log3 with the Sony A6300 (with LUT’s to download).

advertise-here-275 Using S-Log2 and S-Log3 with the Sony A6300 (with LUT's to download).

Sony_A6300 Using S-Log2 and S-Log3 with the Sony A6300 (with LUT's to download).The new Sony A6300 is making quite a stir. This compact interchangeable lens camera has an amazing feature set that is very similar to the features found on it’s bigger brothers the A7sII and A7RII.

As a video camera it’s also capable recording using the XAVC-S codec in  both HD and UHD(4K). It can even shoot at 120fps in HD.  It also has picture profiles so you can tailor the look of the pictures or to suit different shooting conditions. The Cinegammas are very useful for challenging lighting conditions as they offer an improved highlight roll-off. As well as the Cinegammas the camera also has the S-Log2 and S-Log3 log gamma curves as well as S-Gamut, S-Gamut3 and S-Gamut3.cine.

Log gamma curves are very different to conventional gamma curves. Conventional gamma curves (like rec-709 or Cinegamma) are designed to produce a pleasing on screen image without any post production work (although the cinegammas do typically benefit with some tweaking in post). To do this conventional gammas keep the mid-range contrast compatible with conventional TV’s and monitors, so the picture has natural contrast when viewed on a TV. Then to help deal with bright highlights conventional gammas use some kind of highlight roll off or knee to increase the brightness range the camera can capture without effecting the mid range.  Unfortunately this means that the highlights are somewhat compromised, looking very flat, lacking contrast and this is often what gives video the “video look”. In addition it also means that if you are over exposed when you shoot, the picture will look bad and no amount of post production correction will ever make it look good. Most TV camera operators will be very familiar with the phrase “protect your highlights”, meaning don’t over expose, if anything under expose a tiny bit to keep the highlights looking good.

Log gamma curves such as S-Log2 and S-Log3 are very different.  They extended the dynamic range that the camera can capture. To do this they no longer try to be directly compatible with conventional TV’s and monitors and just focus on capturing the biggest possible range. The pictures will be made compatible with the TV or monitor via adjustments made during editing or in post production. So working with S-log2 and S-Log3 is a two step process – shooting and grading (grading is the term used for adjusting the look of an image in post production).

Because log gamma no longer needs to have a contrast range that matches the display range, more dynamic range can be squeezed into a conventional recording codec. It also means that there is no longer a need to use any highlight roll off or knee, so there is a lot more picture information in the highlights and brighter parts of the image. As a result exposing log gamma extra brightly is not normally a problem and in many cases brings lot of advantages. Log gamma curves do have a shadow roll off that mimics the real world shadow roll off. As a result log gammas really don’t like being under exposed, instead they prefer to be over exposed. So unlike conventional gamma where we “protect the highlights” with log gamma you need to “protect the shadows”. Under exposed log looks bad. It will lack color and it will be noisy and grainy.

For most camera operators it’s quite disconcerting to start shooting very slightly over exposed as it goes against everything you’ve learnt about shooting with a conventional video camera. But trust me, shooting 1 to 2 stops brighter than the recommended levels given by Sony on the A6300 (and any other Sony Log camera) will normally bring the best results.

I wrote a guide on how to do this with the Sony A7s here: http://www.xdcam-user.com/2014/08/exposing-and-using-slog2-on-the-sony-a7s-part-one-gamma-and-exposure/

The very same techniques can be used with the A6300, A7SII and A7RII. The A6300 etc also have S-Log3, but as these are 8 bit cameras (even when using an external recorder) I would still recommend that you use S-Log2. Besides, viewing and monitoring S-Log3 is tough as the pictures look even flatter than S-Log3.

If you follow the link below you will find a set of LUT’s (Look Up Tables) that can be applied to A6300 footage in post production to give different looks. There are some film-look LUT’s and 709 (normal TV look) LUT’s and in each case there are LUT’s for normal exposure as well as footage exposed 1 stop and 2 stops brighter. If a LUT name includes “1OVER” this means that the LUT is designed to work with footage that has been exposed 1 stop brighter than the levels given by Sony. My recommendation is to shoot at between 1 and 2 stops bright. For both S-Log2 and S-Log3 this would mean setting zebras to 70% and exposing a piece of white paper so that zebras are just starting to appear on the white paper.

If you find these LUT’s useful please consider buying me a coffee or a beer. I’m not paid to write these articles. For the LUT set I would appreciate a Cocktail, but am happy with whatever you feel is appropriate or affordable.


Type



pixel Using S-Log2 and S-Log3 with the Sony A6300 (with LUT's to download).

To download the LUT set please click the link: Alisters A6300 Mega LUT Kit

 

36 thoughts on “Using S-Log2 and S-Log3 with the Sony A6300 (with LUT’s to download).”

  1. What are your thoughts on the A6300? The 6k over sampling just creates the most eye popping 4k I have ever seen. The amount of detail is pretty striking.

    What do you think of it.

    CT

    Btw..it was nice to meet you at NAB. Nice job on your presentations there!

    1. I bought one and it’s simply the best you can get for the price! I know the rolling shutter is probably the worst measured to date (in 24P S35 mode), but don’t forget it’s 1 to 1 read from the sensor (1.5x) which is 2.4x up sampled (8.3MP x 2.4x) for a 6K read… The detail is a bit better than the a7s ii for the most part when you learn how to use this camera and make the rolling shutter work for you. You can always shoot 30P UHD which is like a 1.9x or 2.3x crop with no binning or skipping… just crops part of the sensor and reads 1 to 1 from that crop. Def. much less rolling shutter issues… u can right away see that in the viewfinder. The Auto-focus is fun to work with… pretty much does what the a7s ii can do (except in low light situations…) … it does start to overheat in 24P S35 mode… but the camera was still shooting with no artifacts. I just turn it off when I’m in between shots… works fine… but hey, you’re getting 6K to 4K clips! Very compact and ergonomic in my hands, yet feels healthy. I was going to get the Nikon D500… but it doesn’t shoot log. Sure 10 bit would of been a plus to remap the values into, but again, you have to wield it into your comps! Oh, it does have a view assist Rec709(800) for s-log2 and s-log3… so don’t worry about using s-log3… it looks great in the viewer! Keeping your whites around 70% range seems to be the sweet spot to bring back the skies (plus or minus depending on the comp)… but the whites will display at ~90% on the waveform with the rec709(800) kicking in (using an external monitor/recorder like the Assassin…) Never underexpose when shooting log like Alister mentioned… ~1.5x over middle grey is a happy medium as well for s-log3… so good luck! I like s-log3 because the curve is very close to Cineon in which you can apply a film post workflow (creating the film scan from he negative LUT which mimics your Cineon, then apply the film of choice LUT for print and viewing while you’re tweaking in the middle) no problem using impulzLUTS and the method they outline. Fun and creative!

  2. Thanks so much for making these LUTs and giving the explanation re: log gamma vs Rec 709 gammas. I noticed that some of the LUTs are labeled as “1UNDER” can you explain when / where you would shoot 1 under with log? Thanks so much.

    1. I would never under expose log, but I included the LUT’s just in case anyone is trying to deal with underexposed or possibly low light footage.

  3. There is a lot of bad information shared here. Slog3 is made to avoid the issue of the slog2, it is the best log profile for this camera. Don’t overexpose slog3 as it is made to be well exposed, this is why a lot of so called filmmaker avoid it, because they don’t know that you shouldn’t overexpose it and then have desaponting results. For best results use it only in 4k. For easy exposure use the viewer assist function that show slog in rec709.

    1. Please explain in detail why you don’t think S-log should be over exposed? As the writer of the official Sony guides to S-log about to be publish on the Sony global website, commissioned and approved by the engineers that developed S-log, I’m curious as to why you think the information here is incorrect?

      Why only use it in 4K?

      What exactly is the “issue of S-log2” that S-log3 is made to avoid.

      Both S-log2 and S-log3 are log gamma curves. Above middle grey both perform in a broadly similar manner with no highlight roll off and more or less the same amount of data per stop. Below middle grey S-log3 allocates a bit more data to the darker stops than S-Log2 which has the side effect of making S-log3 appear noisier and lower contrast (but it isn’t actually noisier) and S-Log3 is more linear above middle grey, but these are minor details compared to S-Log3’s biggest problem with an 8 bit camera and that is it’s restricted recording range and as a result lack of data above middle grey.

      S-log2 uses almost all of cameras recording range from 3IRE to 104IRE. S-Log3 however uses a much smaller range from 3.5IRE to 92IRE. Also S-log3 allocates more data to the deep shadows which further restricts the amount of data available for the all important mid range. The parts of the image where all the important stuff in most scenes resides. With a 10 bit camera this would not be a problem but with an 8 bit camera it becomes important as with S-Log2 you have around 25 code values per stop above middle grey but only 18 code values with S-Log3. In both cases below middle grey there are even fewer code values per stop for both.

      18 code values is very, very little data to describe a full stop of exposure. Consider a face which will typically have around 2.5 stop range. Do you think it is better to have 45 code values or shades (S-Log3) to describe the subtle textures that make a face look good or better to have 63 shades (S-Log2). Neither is ideal, but I know which is going to look better. 18 shades per stop is woefully low.

      Under expose either and the situation becomes much worse.

      Next consider noise and grain. With both of these curves and 8 bit recording the bottom 3 stops (where all the noise and grain is most noticeable) are recorded in 8 bit with less than 10 code values for all 3 stops. So any noise and grain or anything in the deep shadows will look bad. If you expose 1 stop brighter you will DOUBLE the amount of code values used for the same things in the darker parts of the scene, so it will look considerably better. Expose a further stop brighter and you quadruple the amount of code values and that’s going to give you a significant improvement in the quality of the textures in the darker parts of the scene plus a 6db per stop improvement in the signal to noise ratio, and this is EXACTLY the same for both curves and in fact almost any log gamma curve.

      The only penalty for over exposing by 2 stops is the loss of 2 stops of over exposure headroom and again this applies equally to both curves. In addition the data is reduced by a factor of 2 per stop compared to a linear recording (again this is the same for any log curve). But as we don’t currently have display technologies that can show this (even current HDR displays cannot show this), from a perceptual point of view this is rarely a concern at up to +2 stops unless you are shooting very bright extremely high contrast scenes.

      With a 14 stop range and no highlight roll-off the loss of 1 or 2 stops of headroom by deliberately over exposing is rarely an issue. In part because again we have no way to display a 14 stop image and even if we did it would be uncomfortable to view these super bright highlights. It must also be remembered that unlike a traditional gamma curve there is no highlight roll off, so the top stop of a log recording has lots of perfectly useable picture information that can be pulled out of the data when grading. Normal gammas with knee etc do not, there is almost no data in the brightest stop (which is why over exposed standard gammas look bad).

      Log is the opposite of standard gamma. Log gamma rolls off the shadows, not the highlights. Standard gammas roll off the highlights through the use of a knee or similar. With standard gammas we know not to over expose because the highlight roll-off looks bad. With log gamma it is the opposite, we should never under expose because the shadow roll off will make it look bad.

      Because there is no highlight roll off, because the amount of data in the shadows doubles (for the same scene information) for every stop you expose brighter, because the signal to noise ratio improves by 6db for each stop brighter you expose, log will almost always look considerably better when exposed brighter by between 1 and 2 stops. This isn’t just my opinion, it is the opinion of some of the very best DP’s and colorists in the world.

      The view assist function is only of very limited use because it only works at the base exposure level. If you try to over expose by a stop or so to improve the quality of your log recordings the viewfinder image will look very over exposed, while if you just monitor the S-Log image it won’t look over exposed, just flat.

      1. If post color grading is the intention, than the concept of “exposing to the right” applies to any log curve as well as rec709 too! Any time you can open your iris or slow your shutter and let in more light and expose as bright as you can before clipping your highlights, the better you are. (And no,…folks, ISO is NOT “exposure” or “sensitively” at all. Its “circuit amplificatiin” of exposed light you already captured)

        Bringing waveforms DOWN in post can often hide noise. Bringing a waveform UP in post will ALWAYS increase noise. So if post grading is your goal, expose to the right no matter what gamma you are usung. Alister is easily 101% right on this and I have no clue what that other gentleman is trying to say.

        Cave man rule of color grading: Bring wave down is good….bring wave up is bad. Lol

  4. Alister – your comments are spot on as usual – only thing is that Slog-3 in the case of the a6300 is not that hard to see on the display – it has the display LUT for it in the menu – when turned on looks pretty much like other profiles as far as visually in the display. For people that don’;t have the camera reading in your article – “viewing and monitoring S-Log3 is tough as the pictures look even flatter than S-Log3.” for me has not been a problem with the Gamma Display Assist.

    1. Sorry just read yur comment above this one – I understand the view assist limitations – so no need to comment on my comment. Thanks

  5. I had never read anything more useful and clear of this comment about the slog profiles Alister. Thank you very much!
    I’m a “strong” CINE4 user. Can you please explane pro&cons to use S-LOG instead of CINE4? Specially for mid light and skin tones.
    My big limit is that I hate ND filters around, for my job (wedding) I can’t find nothing more easy and usefull than CINE4 with low iso. Thanks.

    1. S-Log needs to be graded to look right. So it’s a very different workflow. S-log exposure on an 8 bit camera must be just right and as the picture you see in the viewfinder is not the finished image, judging exposure can be tricky. S-Logs big advantage is it’s ability to handle a greater dynamic range than conventional gammas, however you pay a penalty for this with less data in the mid range. So Cine4 will actually give you better mid range than S-log, but may struggle with highlights in brighter scenes. In difficult, bright lighting S-log may be a bit more forgiving, but it’s not going to eliminate the need for ND filters.

  6. dear alister,

    many thanks for sharing all of your technical expertise with us. just to get this right: my a6300 has a native ISO and changing it to 100 or 3.200 just subtracts or adds electronic amplification? so this means it would be best to leave it at the native ISO and do the rest with nd filters or with gain in post? not only when shooting slog-videos but also when I do raw photography? do you know by any chance what the native ISO of the a6300 is?

    all the best
    mark

    1. Yes, an ISO change is an electronic amplification and noise reduction processing change. Not an actual sensitivity change. The native ISO for S-log is 800 for other gammas it will be between 200 and 400 ISO depending on the curve (gamma curves are gain curves so effect the ISO rating).

      1. It’s funny Mark. Get 100 photographers into a room an ask them this question:

        “Camera 1 is at 1/60 shutter, f2.8 and 200 ISO…..Camera 2 is at 1/60, f2.8 and 800 ISO….which camera is exposed to more light?” (same model and lens)

        I’d guess that prolly 90% of them would say “camera 2” when the answer is neither! Both camera’s sensors are getting the exact same amount of light hitting them.

        Back in the film days, ISO 200 stock was “chemically” different than ISO 1600. That truly does represent a real “sensitivity” change. Since digital cameras don’t allow you to change out our sensor, you are stuck with a sensor that always has ONE fixed sensitivity. (Unless you can get a soldering iron and glue a different one in)

        This is important: When you spin your camera’s ISO dial, you are amplifying the wave form that your sensor is already FINISHED with. Your sensor has captured it’s light and converted it to digital “video” and sent it down the pipe for you to turn the “volume” up if you wish. ISO today is not “light” and the only two things your sensor is concerned about is IRIS and shutter speed. Your sensor doesn’t know or even care about what ISO you are using. I’m not trying to imply that image sensors have feelings. Although, I wouldn’t doubt that Sony has that in their future development plans. 😉

        “ISO” is so ridiculously misunderstood these days in digital imaging. (I’m guilty too) Cant we all just use “GAIN” instead? 😉

        1. This is actually very interesting. So both of you prefer to apply noise reduction in post using neat video for example? Well, I guess the only reason to use “gain” before the shot is to be able to focus properly. I admit that I quite often use the autofocus function of the a6300 as it works pretty well. What do you think?

          1. Well, you have to use ISO in camera to deliver your video to somebody in a normal manner. And certainly as long as you don’t clip something in camera everything is OK. When you shoot a raw photograph, there is no actual ISO (aka “gain”) in that raw data. It’s applied in your raw file reader. It’s only a metadata flag.

            Your camera also has fine tuned image processing that is customized for each ISO setting. It’s got noise reduction that is made for that sensor’s noise characteristics when the camera adds a certain amount of ISO/GAIN to the sensors image. These are conveniences that you get in camera when you ramp up the ISO.

            Shoot normally but just remember that ISO is not “adding light” to a sensor. Only shutter speed and IRIS adds actual light.

            Also, when shooting in LOG, avoid ISO/gain like it’s the plague. Shoot as low as possible….as low as the camera allows you to. Over expose a stop or two only by opening your IRIS more or use a longer shutter, get a faster lens if you need to. Adding ISO is never how you want to over expose your image for LOG. (That might confuse some people too)

            In lower light conditions, LOG needs actual “light”….not “ISO/GAIN” If your light is just too low for you to over expose properly with the lens you have, you just really cant use LOG at all for that shot.

  7. Alister, I have been meaning to ask you:

    SLOG2 uses much of the IRE range on the scopes. Whites can clip over 100 IRE and black are down near 0 IRE. However SLOG3 uses less range. Those same whites clip around or close to 94 IRE and the deepest blacks are well above 0 IRE.

    Why did Sony design SLOG3 this way? I know that it was certainly intended for 10 bit encoding but even still, why leave so much unused headroom? Was anything gained by that? Couldn’t the same SLOG3 curve been plotted into 0-255 or 0 -1024? or maybe even 0 IRE to 109 IRE?

    I was always curious about that.

    CT

    1. S-Log3 is based on the Cineon log curve and can record up to 16 stops. At the moment the cameras can only do 14 stops, so the luma cuts off where the camera cuts off. If the camera could deliver more it would go all the way to 104 IRE.

      S-Log2 was specifically matched to a 14 stop electronic camera, but colorists complained as it was unfamiliar. So Sony went back to Cineon. Log-C on an ALexa is the same, it only reaches 94%.

  8. In the past couple of weeks I experimented with the various picture profiles of the a6300. Within these tests I also varied with the color profiles. Now I have some material that I recorded with CINE1 and the S-GAMUT color profile.

    Is there a LUT that translates only the S-Gamut space to REC709?

    1. That’s a very strange combination, not one I would suggest as the camera can’t actually see the S-Gamut range, so a lot of the recording data is wasted space. Plus S-Gamut introduces a green shift during grading. There are no LUT’s for this, but you could design your own using the LUTcalc app. https://cameramanben.github.io/LUTCalc/

      1. OK, I thought that it could be used separately. The idea was to have the same color space to work with when I have a situation when the cinegamma does a better job. So would you recommend to shoot a project with only one picture profile in order to get the same look for each scene?

        Actually, I think I saw this combination in a Philip Bloom tutorial.

        1. The main thing to understand is that gamma and color space are linked. A LUT cannot separate one from the other, so a LUT is designed to work with a combination of the two.

          In addition the A6300 like most cameras cannot see the full S-Gamut or S-Gamut3 range. So choosing S-Gamut as the recorded colorspace is a waste of data and brings no benefit unless you are attempting to match an F55/F65 and using the same LUT’s (not that they will match anyway). Only the F55/F65 can see and thus record the full S-Gamut or S-Gamut3 color range. S-Gamut3.cine is a better match to the A6300 sensor as is the Pro color matrix.

          I will often use different profiles for different scenes to maximise what I am recording. This is particularly important with an 8 bit camera.

          1. OK, I guess I was confused because your A6300 LUT-Kit is called “SLOG2-SGAMUT” and “SLOG3-SGAMUT3.cine”. I reckoned I should use the S-GAMUT color space with SLOG2.

            So if I understand correctly (from what I learned so far), with the A6300 I should use:

            – cine1 for bright light situations
            – cine4 for darker situations
            – slog2 for situations with a high dynamic range

            … and for all picture profiles use either the PRO or S-GAMUT3.cine range.

          2. Alister, I think your LUTs are very impressive, and I really like using them. But there are two things I am still struggling with:

            1. It is very difficult to shoot exactly at 2 stops over. When I take the 2 stops over, the image turns out a little bit too dark, when I choose 1 stop over, the image looks too bright. I use davinci for editing. To adjust the image, I select a node at the end for the 2 stops over LUT. In the previous nodes, I correct contrast and saturation. For exposure I use the offset knob. Is that the right way to do? If yes, could I also use the normal LUT and just pull down the offset until the exposure looks right?

            2. I find it very difficult to get the same look when I recorded some scenes with CINE1 and some with SLOG2. Especially when I want to use our film-LUTs like the Fujiish1 for example. Maybe you could generate the film-LUTs for the CINE1 and the PRO-matrix as well? That would just be great and make things so much easier. Or is there another way to get to the same result?

          3. Ah this is exactly the info I was looking for. I’ve read you A7 articles on recording with s-log over and over trying to get my technique dialed and found slog2-3 to be way too noisy with the a6300. So recently I started playing with sgamut3.cine and Cine4. Happy with the results but still trying to wrap my brain around what is actually being recorded and how to even start color grading. As always awesome article and appreciate the LUTs!

          4. S-Log isn’t noisy, it just needs to be exposed brightly and then properly handled in post. You need to use a decent color grading tool like DaVinci Resolve.

      2. Today I had the chance to look at LUTcalc. Thank you very much for the tip. Which gamma would you suggest to set for Rec Gamma with my strange combination? I can’t find the CINE1 gamma there. As you always refer as “standard gamma” I guess the best choice would be “Rec709 (800%)”, right? The same goes for the Out Gamma. Do I also put “Rec709 (800)” because I don’t want the gamma to be changed?

        As for Rec Gamut I will put “Sony S-Gamut”, and for Out Gamut I choose “sRGB” as my monitor can only display sRGB…

        What do you think about these settings?

  9. This camera auto focus in video is really amazing. I do not envy 4k, but the auto focus feature alone would lust me to buy it. but unfortunately the 1080p mode is not full pixel read out, so softer than a7s 1080p.

  10. Alister,

    These LUTs are very useful as the color management tools in many applications (like Resolve) don’t accommodate for the overexposure which most informed Sony users seem to be incorporating per your education.

    Would you consider adding an S-log2/SGamut3.cine to 709 LUT gamma/gamut conversion to your pack? I’ll send an additional PayPal donation 🙂

    Resolve now allows the user to specify the gamma and gamut of the input footage and will auto-calculate the transform, just, again, it does this without the intended overexposure and just changing “offset” values post-LUT doesn’t do the same conversion I’d assume.

      1. Thanks Alister. Since the a6300 isn’t available as an option, which camera might I select? I know the A7sII is available but the base ISO is different.

  11. It doesn’t matter which camera you chose or what ISO. Just use the stop correction and offset the exposure by the desired number of stops, typically between -1 and -2.

    1. Good to know. That’s easy enough then. And I’m assuming that to replicate your LUTs one would choose the “Linear/y” option for gamma output with Rec709 exp2.22 correction?

      Thanks again. LUTCalc is far more intuitive than other, similar, resources I’ve experimented with.

  12. No, set the output gamma option to ALL or Sony and choose Rec709(800) and Rec709 Gamut to match the 709 LUT’s I created.

Leave a Reply

Your email address will not be published. Required fields are marked *