• Cinematography

  confused on log, exposure and final image

  Posted by Clyde Villegas on June 30, 2017 at 10:34 am

  I’m just starting to learn about exposure and log.

  I’ve read that while filming in log, you should already be decided on the final image. Either you over expose or under expose to protect either the shadows or the highlight.

  I am confused. If you are already decided on the final image, why not just get rid of log, use standard picture profiles, and expose properly on the highlights or the shadows, depending on which you want to appear properly on the final image? Does using a log somehow save some of the highlights on final image when your priority is the shadows?

  ut in omnibus glorificetur Deus

  Chris Wright replied 8 years, 9 months ago 3 Members · 2 Replies
• 2 Replies

• 

  Blaise Douros

  June 30, 2017 at 6:54 pm

  In a word, yes. Log footage has almost double the dynamic range of Rec709.

  For example, in a dark situation, with regular Rec709 gamma, you might see some noise in your shadows if you exposed the image right down the middle. So what you do is overexpose by one stop on set, to increase the amount of data you capture in the shadows. Then, in post, you drop the exposure by a stop, thus, crushing the shadows and reducing the amount of shadow noise. Because you have so much additional dynamic range, you also lose very little data from the highlights; with normal gamma, the highlights would have been too blown out to recover, but with the additional dynamic range of log, you still retain information there.

  You can do the opposite when you have a very bright situation, and need to retain extra highlight detail.
• 

  Chris Wright

  August 14, 2017 at 11:23 pm

  its not so simple. dynamic range is from the luma logarithmic curve that steals bits from
  blacks or whites to increase dynamic range. that is why 8 bit log doesn’t work well because it creates
  banding from low precision transfer functions. The human eye can see steps in brightness in this way if
  not enough information is available via precision.

  for example, rec. 2020 uses the same sample non-linear transfer function as rec. 709 but at 10-12 bits instead of 8 bits.

  Standard range video with a 2.4 gamma curve and a bit depth of 8-bits per sample has a dynamic range of about 6 stops.

  The thing that creates HLG TV is the logarithmic curve for the upper half of the signal values. When HLG is displayed on a 2,000 cd/m2 display with a bit depth of 10-bits per sample it has a dynamic range of 200,000:1 or 17.6 stops.

  as you probably can guess, this changes how you properly expose shooting slog2 as this moves the grey point around from 42% to 32%. shooting a7s at the camera’s native 3200 iso(sensor noise floor) will also maximise dynamic range.

  a slog2 MM +2 70% zebra will match rec. 709 skin tones and help a lot with 8 bit cameras as they won’t have to interpolate much during post grading. true also for white balance if not using RAW. ETTR iso changes only moves the noise range around which crushes dynamic range.

  interestingly, a 8 bit camera will look better with a more conservative gamma curve at the expense of dynamic range. that is why 8 bit green screen keying works best with neutral profiles over log. there’s more precise bit gamma data per green pixel value.

  https://www.xdcam-user.com/2014/08/exposing-and-using-slog2-on-the-sony-a7s-part-one-gamma-and-exposure
  https://www.xdcam-user.com/2014/05/what-is-a-gamut-or-color-space-and-why-do-i-need-to-know-about-it/

Reply to this Discussion! Login or Sign Up