GH5 Waveform Display Calibration in VLog

Cliff Totten · June 20th, 2017, 12:23 PM

Its my understanding that the way V
log, Cineon and SLog-3 works is this:

Take whatever image sensor you have. Take a certain gray point on its output and map it to the proper coordinate gray value inside these log curves. Then, however far up and far down your sensor output acheives...the total dynamic range that it produces.....will land at whatever waveform points that it does. (High and low)

So, if have a 12 stop sensor, your sensor can only hit about 80 IRE. If you had a 10 stop sensor, you could maybe only produce a 70 IRE value. If you had a 15 stop sensor, you might be able to tickle 100 IRE on these log curves. What im saying is that your sensor is calibrated at a certain "middle-ish" grey point with the shadows and highlights falling where they naturally would on the curve/scale.

Now this is all very different in Sony SLog-2. For this log, Sony went a different direction. I am told that Sony maps SLog-2's curve based on each sensor models maximum highlight and shadow output. They map each sensor's shadows right down to almost the bottom near 0 IRE and take the sensor's maximum brightness all the way up to 109 IRE. So Sony uses the entire scale for each sensor, no matter if its a 10stop sensor or 11 or 14 stop sensor. Every sensor will be "stretched" to a max 109 value.

This is why SLog-2 is not as flat as SLog-3 and is more "8bit friendly" than all the other log curves out there. SLOG-2 uses the full 8bits for recording and does not leave any headroom unused no matter what the sensor's actual dynamic range output it has.

Technically, SLog-3, Cineon and VLog can hold over 16 stops of dynamic range even though most sensors today cant truely do 16 stops. So, as sensors get better and better, we should see highlight clipping points rising further and further up the waveform scale until we eventually hit VLog 109 IRE. But not with the chips today.

Another way to say it is, if you drop a 12 stop GH5 sensor into the middle of a 16+ stop gamma curve, you will get literally 4+ stops of unused headroom. Even more unused headroom for lower dynamic range sensors and will look even more dismal on your scopes.

This is the best I can explain with my limited knowledge on this. I have spent allot of time with Sony's guys at NAB over the years and this is what I have been told. I believe this is all basically accurate. If not, lemme know.

CT

Roger Keay · June 20th, 2017, 03:16 PM

I agree with what you are saying for a professional camera, but the GH5 is a consumer camera. I think Panasonic is pre-processing the video to achieve compatibility with consumer HDR sets. We will know more when the summer update is released and HLG becomes available. My bet is that after decoding in an edit program, HLG and V-Log L files will look the same.

Going back to the original subject of Waveform Display Calibration, I don't think it is possible to actually see the HDR component of the signal because the display is too coarse as suggested in one of the earlier posts. As it is, the WFM doesn't help with the HDR levels and reduces the value of the display for SDR levels. Configuring the WFM to display the SDR levels full scale would make more sense.

June 20th, 2017, 12:23 PM	#31
Cliff Totten Trustee Join Date: Aug 2010 Location: Florida Posts: 1,197	Re: GH5 Waveform Display Calibration in VLog Its my understanding that the way V log, Cineon and SLog-3 works is this: Take whatever image sensor you have. Take a certain gray point on its output and map it to the proper coordinate gray value inside these log curves. Then, however far up and far down your sensor output acheives...the total dynamic range that it produces.....will land at whatever waveform points that it does. (High and low) So, if have a 12 stop sensor, your sensor can only hit about 80 IRE. If you had a 10 stop sensor, you could maybe only produce a 70 IRE value. If you had a 15 stop sensor, you might be able to tickle 100 IRE on these log curves. What im saying is that your sensor is calibrated at a certain "middle-ish" grey point with the shadows and highlights falling where they naturally would on the curve/scale. Now this is all very different in Sony SLog-2. For this log, Sony went a different direction. I am told that Sony maps SLog-2's curve based on each sensor models maximum highlight and shadow output. They map each sensor's shadows right down to almost the bottom near 0 IRE and take the sensor's maximum brightness all the way up to 109 IRE. So Sony uses the entire scale for each sensor, no matter if its a 10stop sensor or 11 or 14 stop sensor. Every sensor will be "stretched" to a max 109 value. This is why SLog-2 is not as flat as SLog-3 and is more "8bit friendly" than all the other log curves out there. SLOG-2 uses the full 8bits for recording and does not leave any headroom unused no matter what the sensor's actual dynamic range output it has. Technically, SLog-3, Cineon and VLog can hold over 16 stops of dynamic range even though most sensors today cant truely do 16 stops. So, as sensors get better and better, we should see highlight clipping points rising further and further up the waveform scale until we eventually hit VLog 109 IRE. But not with the chips today. Another way to say it is, if you drop a 12 stop GH5 sensor into the middle of a 16+ stop gamma curve, you will get literally 4+ stops of unused headroom. Even more unused headroom for lower dynamic range sensors and will look even more dismal on your scopes. This is the best I can explain with my limited knowledge on this. I have spent allot of time with Sony's guys at NAB over the years and this is what I have been told. I believe this is all basically accurate. If not, lemme know. CT Last edited by Cliff Totten; June 20th, 2017 at 01:20 PM.

June 20th, 2017, 03:16 PM	#32
Roger Keay Regular Crew Join Date: Nov 2008 Location: Toronto Canada Posts: 75	Re: GH5 Waveform Display Calibration in VLog I agree with what you are saying for a professional camera, but the GH5 is a consumer camera. I think Panasonic is pre-processing the video to achieve compatibility with consumer HDR sets. We will know more when the summer update is released and HLG becomes available. My bet is that after decoding in an edit program, HLG and V-Log L files will look the same. Going back to the original subject of Waveform Display Calibration, I don't think it is possible to actually see the HDR component of the signal because the display is too coarse as suggested in one of the earlier posts. As it is, the WFM doesn't help with the HDR levels and reduces the value of the display for SDR levels. Configuring the WFM to display the SDR levels full scale would make more sense.