Shooting 4K Log, guidance please

[Scott Berrington]

I think JVC is referring to image brightness that is 800% higher than the nominal level coming off of a 90% white source(test chart), which is about 3 stops of exposure headroom. This makes sense as with Rec709, there is about 2.5 stops from middle gray(18%) to peak white(90%), so adding 3 stops(800%) gives a total of about 5.5 stops above middle gray when shooting with J-log. The graph shows about 6.5 stops under 18% and 5.5 over...so 12 stops in total.

When shooting in J-log, I set my zebras to 70-75ire for anything in the scene that is "white" and have had good results.

[Lee Powell]

Originally Posted by Scott Berrington

I think JVC is referring to image brightness that is 800% higher than the nominal level coming off of a 90% white source(test chart), which is about 3 stops of exposure headroom. This makes sense as with Rec709, there is about 2.5 stops from middle gray(18%) to peak white(90%), so adding 3 stops(800%) gives a total of about 5.5 stops above middle gray when shooting with J-log. The graph shows about 6.5 stops under 18% and 5.5 over...so 12 stops in total.

What does "nominal level" of 90% white refer to? It appears to be an arbitrary spot on the tone curve referenced to the arbitrarily-positioned "Stop-0" mark. How can there be 3-stops of DR above 90% white when, by definition, luminance can only be increased by an additional 10%?

[Scott Berrington]

A 90% reflectance White, like that found on a DSC lab chart, would be exposed to 100ire using Rec709. Using J-log, 90% white is exposed at approx 65ire, and thus allows for approx. 3 stops(800%) of over exposure. 800% would be recorded at 109ire(level 255 in 8bit). Stop 0 refers to 18% middle gray...that's why there is about 5.5 stops above 18%, and about 6.5 stops under.

[George Odell]

Originally Posted by Scott Berrington

...Using J-log, 90% white is exposed at approx 65ire... Stop 0 refers to 18% middle gray...that's why there is about 5.5 stops above 18%, and about 6.5 stops under.

So it appears JVC is following basically the same scheme as the other camera manufacturers as to where white is at around 65 IRE and 18% middle gray at around 36 IRE.

Is this what were are saying?

[Lee Powell]

Originally Posted by Scott Berrington

A 90% reflectance White, like that found on a DSC lab chart, would be exposed to 100ire using Rec709. Using J-log, 90% white is exposed at approx 65ire, and thus allows for approx. 3 stops(800%) of over exposure. 800% would be recorded at 109ire(level 255 in 8bit). Stop 0 refers to 18% middle gray...that's why there is about 5.5 stops above 18%, and about 6.5 stops under.

"90% White" seems like a completely arbitrary point approximately 2.5 stops above "18% Gray", which is likewise defined as an arbitrary point somewhere near the middle of the gamma curve. With that scaling, 109 IRE becomes "800% Reflection", which is physically impossible (without amplification). The cherry on top is the erratic behavior of the on-screen "peak" light meter, which in J-Log1 mode can fluctuate between 350-800% while the zebra and histogram are clearly indicating levels over 95 IRE. (I suspect it's actually an area meter that averages luminance levels within its green rectangle.) As we already have an industry standard 0-109 IRE scale, what is the point of these inflated percentages?

http://pro.jvc.com/pro/attributes/4k...l_of_D-LUT.pdf

[Scott Berrington]

Originally Posted by George Odell

So it appears JVC is following basically the same scheme as the other camera manufacturers as to where white is at around 65 IRE and 18% middle gray at around 36 IRE.

Is this what were are saying?

J-log seems to be very similar to the original Sony S-log, which was designed for 8 bit recording.

90% white is not arbitrary, it's a standard for setting exposure and testing the sensitivity of cameras. 800% is used to describe an object luminance that is 3 stops above normal peak white.

http://www.dvinfo.net/forum/attachme...ings-s-log.jpg

[Lee Powell]

Yes, 90% White is standardized with respect to Rec 709, but here we're dealing with J-Log1 footage, which has a much broader dynamic range. As a consequence, the 18% Gray standard, which is based on a DR of 5.5 stops, no longer fits. That archaic assumption is what I was refering to as "arbitrary" above.

If not 18% Gray, what should you use as a standardized exposure reference point? Most often, I use ETTR to religiously protect the highlights, particularly when shooting stand-alone events that will not require intercutting with footage from other cameras. As the clips pile up, however, maintaining consistent skin tone exposure becomes a bigger issue than highlight protection. For that I use a conservative version of the Wolfcrow system described below for Sony S-Log2 footage, which is similar to J-Log1. Wolfcrow recommends pushing skin tones up 3-stops to around 60-70 IRE. I think 2-stops is more prudent, with skin tones at around 50-60 IRE. The LS300's Zebra can be set to highlight this or any other IRE range you prefer.

How to Expose, Monitor and Grade S-Log2 Footage on the Sony A7s - YouTube

[Scott Berrington]

Setting the zebras at 50ire for skin tone would probably be a safe bet. Like I mentioned in a previous post, I normally set my zebra to 70ire and expose anything that is white to that. If there is nothing white in the scene, I'll use a calibrated white balance card or piece of paper to set exposure. I find this method is great for keeping the exposure consistent, where as ETTR can be very inconsistent.

I shoot with the Sony FS7 frequently, and use the waveform for checking exposure. It will show level with or without the LUT depending on how you setup the viewfinder...very helpful, especially if there is white in the scene.