Music video intro graded using v1.4 PP SLOG to SXS

[Dennis Hingsberg]

This music video I shot and directed is still in editing, but I've posted the intro for those who want to see heavy graded PP SLOG (picture profile SLOG using v1.4 firmware) shot to SXS.

This first version is graded excessively, but have a look. I will try to get a version up on vimeo since YouTube has lots of compression. Shortly I will also post the original along with a tamed down version of the grading...

PREVIEW - Tim Toishi - Only One - Directed by Dennis Hingsberg - YouTube

Details about the shoot:
- 1920x1080 @ 24p, 1/48th shutter, 0db gain.
- Picture profile used: SLOG via v1.4 firmware.
- Recorded to internal SXS (8-bit 420)
- Lighting: f5.6 at 800ISO on subject. (approx 84 fc)
- Skin at approx 45IRE.
- Shadows lit for 10-18IRE.

Despite shooting SLOG straight to SXS in only 8-bits, I feel the SLOG definitely helped keep the wide range of contrast I had in the scene between the artists very white suit and his skin color, and the dark piano and dark background. Looking at the scene in Rec 709 and cine gamma modes 1, 3 and 4 the white suit would always clip over 100 IRE with the skin (or other midtones) under exposed. SLOG was a great way to preserve the tones in this type of setup and condition.

I also find only a gentle S-curve needs to be applied to bring the image back to "normal". I plan to continue to shoot using the SLOG picture profile going forward, and like I said it is especially useful in high contrast situations. Even more so outdoors. The extra latitude of the dynamic range is simply amazing.

[Dennis Hingsberg]

Now showing original SLOG, lightly graded, and heavy grade supreme.

Original:

PREVIEW - untreated - Directed by Dennis Hingsberg - YouTube

Graded light:

PREVIEW - subtle grade - Directed by Dennis Hingsberg - YouTube

Graded heavy:

PREVIEW - Directed by Dennis Hingsberg - YouTube

[Ron Aerts]

I'm sorry, but grading in this way is not really my taste. It makes faces look ugly and you loose detail (and with that emotion). with so little DR its hard to preserve skintone.
In my opinion there is nothing to gain in using S-log on 8-bit SxS recording over cinegammas. I think you could accomplish the same, or even better with 1 stop underexposure during recording. Gives you more room in the top end for the suit.
I like the song though

[Dennis Hingsberg]

Ron, this was somewhat of a test to see SLOG 8-bit pushed around and what is possible. To many people advise against doing it (to clarify, too many people advise against shooting SLOG to 8-bit) but I wanted to show the results when it is done.

Also your comment about using cine gammas instead of SLOG, I don't tend to think about SLOG vs cine gammas. Instead I like to chose the gamma curve that will get my overall scene tones closest to where I want them baked into the image so later some subtle grading can be applied without massively shifting tones around. Remember, it is 8-bit.

[Ron Aerts]

clear. Just keep posting. it's always interesting to hear.

[Dennis Hingsberg]

Also, even though I am a big fan of using lighting for all my work, and using light meters to determine my ratios, and using waveform or built-in spot meter to see IRE's in my scene... for me the gammas are a final way to "cheat" a little in-camera before post... and I love that.

Note also that cine gammas 1, 2, and 4 will record above 100 IRE to about 105/110 IRE which can be useful in many situations with highlights or bright spots.

[Dennis Hingsberg]

Originally Posted by Dennis Hingsberg

...Note also that cine gammas 1, 2, and 4 will record above 100 IRE to about 105/110 IRE which can be useful in many situations with highlights or bright spots.

This is a typo and I can't edit the post. Clarification: cine gammas 1,3 & 4 record above 100 IRE. It is cine gamma 2 that records a legal signal clipping anything above 100 IRE.

Sorry about that.

[Alister Chapman]

While cinegammas 1,3 and 4 do record to a higher voltage/data bit level, there is no increase in dynamic range. The dynamic range of cine 1 and cine 2 is the same and highlight handling is the same. The difference is simply in the levels that the same range is mapped to. The only differences that you gain by using cine 1 over cine 2 is 9% more data bits are used, which may give a marginal improvement in grading as you have a few more bits of data per stop. In addition on a computer white is bit 255/109% so recording to broadcast safe bit 235 (cinegamma 2) can result in pictures that don't look white on a computer screen.

If you are grading or working on projects for computer delivery, then it is normally advantageous to use the extra data bits offered by cinegammas 1, 3 and 4. But there is no gain in latitude, it's just a levels shift. The same dynamic range is just mapped to different recording levels. Cinegamma 2 is very useful for anyone struggling with edit software that clips anything beyond 104% (which is more common than I think many people realise and why S-log only goes up to 104).

[Dennis Hingsberg]

How can highlight handling of cine gamma 1 + 2 be the same when 1 can record 9% higher over 100IRE? 9% is still 9% and can help contribute to an overall smoother roll-off of the highlights if one shoots properly exposed to do so?

By the way here is a PDF I put together some time ago illustrating the different cine gamma curves which can be quite useful if anyone is interested: http://www.starcentral.ca/Sony%20PMW.pdf

As for the project I made reference to in my original post, here is the finished product and grading was quite subtle in the end compared to these earlier examples, enjoy!

http://www.dvinfo.net/forum/sony-xdc...im-toishi.html

[Alister Chapman]

You can take any sensor output level and map it to any recorded signal value. The dynamic range is determined by the sensor, not by the values chosen to record it's output. It's just like the amplifier in a sound system. The quality of the microphone will determine what is captured, the amplifier simply makes it louder or quieter, but the captured range remains the same. It's the same with the DSP in the camera. It takes the sensor output and amplifies it (following the chosen gamma curve) to the recorded signal level. Cinegamma 2 is the same as Cinegamma 1, except that the signal gain and thus recording level is lower, like turning down the volume on an amplifier a little bit, the same sounds are there, just not quite a loud. It is not the clipping or limiting that is different between Cine 1 and Cine 2 but the gain (amplification) applied.

You can see this clearly in your PDF liked above. Both Cine 1 and Cine 2 have 4 stops above middle grey and 7 below. With Cine 1 the +4 stop is at 109, with Cine 2 it's at 100, the only difference is the signal amplitude (level), Both show exactly the same +4/-7 stops of dynamic range, just recorded at slightly different levels. With Cine 2 not only is the 4th stop above middle grey a little lower in level but all of the brighter stops are slightly reduced in level. You could take any of those waveforms and amplify them further, which would make the levels even greater taking them to whatever amplitude you want, but you would still only have 4 stops over middle grey. This is an area that is widely misunderstood, you are not alone in thinking that the dynamic range or latitude is decreased with the broadcast safe cinegammas or hypergammas, but this is not the case. It's a problem exacerbated by the fact that the Cinegamma 2 image won't look as bright as Cinegamma 1 due to the lower recording level and many confuse this reduced display brightness as a reduction in dynamic range.

[Dennis Hingsberg]

I understand there is no increase in dynamic range given your explanation, and it does make sense. However as discussed on this and many other forums the tendency is to over-expose resulting in poor maximization of sensor dynamic range no matter what your dynamic range might be - therefore by using cine gamma 4, or SLOG it actually can give you the most "apparent" dynamic range and increased highlight protection.

ie. using SLOG: If you expose your image for cinegamma 1 or 2, SLOG has approximately the same middle grey point but with the addition of more range at the high end. (6 stops over grey vs 4 stops). Also since cinegammas 1, 2 and SLOG will appear "darker" on monitor most shooters will tend to open up further reducing the apparent "dynamic range" - particularly in the highlights.

ie. using cinegamma 4: If you expose using cine gamma 4 which has a high grey point and bring it down 1-2 stops where middle grey will look normal or measure good again, you will effectively increase your higher end.

In either of these examples the sensor dynamic range itself has not changed, but what you end up with will surely lend a more "apparent" dynamic range and this was the point I wished to share about shooting with these powerful and amazing gammas.