Tuesday, July 23, 2013

Sony F55 using Canon FD Lenses, Part 6, Gamma Curves

In this post, I wanted to see what each of the gamma curves would do under the exact same conditions within a scene. These are the preset gamma choices that the F55 provides. Selections include STD (standard) 1-6 and HG (hypergamma) 1-4, plus 7-8. I don't know why Sony doesn't simply call 7 and 8, 5 and 6, instead of skipping those numbers.

The setup was to shoot a simple scene of my wife wearing some black, white, and red, so that I have the extremes on the grey scale and a mid-grey value like red. The exposure is set for skin at 70%, but I also added a heavy key side kicker that equals the intensity of the key. It is meant to be rather harsh in order to see what the gammas do to or for the extreme highlights. The fill side dropped off by 2 stops from the key. The background is also about 2 to 4 stops below the key at it's darkest points.

My goal was to shoot each setting with the same stop, which I did for the most part, but found a couple settings dramatically increased the key level, well above the 70% IRE on the skin and would have caused the skin to overexpose. Since all of these shots had a "normal" color space setting, it was easy to detect when I was over 70% and therefore had to close down the aperture. Otherwise, if the gamma setting got darker, I just let it happen to see where things changed. So STD1 was where I set my key light skin for 70% as a starting point.

Since some of the changes were quite subtle, I recommend that interested people download the stills and look at each under waveform and vector scopes to really see what is happening.

 

STD1 Skin was set to 70% IRE. Harsh highlight, deep shadow, saturated color, fairly high contrast, were some of the attributes I would describe.


STD2 showed a slight bump up in the shadows without much movement in the mid tones or highlights according to scopes and eye other than a negligible amount of desaturation.
STD3 went back to more like what STD1 was only a little more control in the highlights and a slightly deeper shadow. No saturation change.
STD4 very slightly elevated the shadows, but the highlights stayed the same as STD3. No noticeable saturation change.
STD5 only slightly expanded the shadows yet again, leaving the highlights, mid tones and saturation alone.



STD6 showed the largest expansion of shadows and only slightly elevated highlights. This had the most low contrast effect of all of them.




Of the 6 STD gammas, 3-5 offered the most control of the highlights, with 4 and 5 elevating the shadows slightly each time. All had strong color saturation.

Now for hyper gammas.

HG1 Highlights, mid tones, and shadows all dropped in level as well as saturation compared to STD1.
HG2 slightly elevated the highlights, mid tones, and shadows without much saturation increase. Still well below STD1
HG3 jumped the levels up dramatically, requiring me to stop down the aperture, but after that what the scopes show is an expanded shadow, mid tone, and highlight range, yet the highlights were still in more control than STD1.
HG4 was the most level drop of all the gammas thus far. Highlights, mid tones, shadows and saturation all dropped and compressed the value scale.
HG7 had the biggest level jump overall, even more than STD6 or HG3. I had to stop down to 4-5.6 split to get zebras back to 70% on skin. Highlights, mid tones, shadows, and saturation all were raised on this setting.
HG8 is more like HG3 in that levels came up over HG1, HG2, and HG4 but highlights stayed in control. Saturation was very much like HG3.




All the HG setting were lower in color saturation compared to STD. The most control over highlights in STD was STD2, but only by a small margin. HG settings offered greater control over saturation and highlights, specifically HG3, HG7, and HG8. Overall, I liked HG8 the most for highlight control, but with a slight lift in shadows over HG3. HG7 was also good once I stopped down to compensate and found the image expanded across the grey scale better while still protecting the highlights.

Now comes some corrected S-Log-2 comparisons under the same lighting conditions.

First thing I noticed was how much more yellow this is compared to STD or HG images, but also how easy it is to bring highlights, mid tones, and shadows in line. I added only 20% more color saturation since I was in the "normal" color space setting.
This is S-log-2 with the S-gamut color space. Still yellow compared to STD or HG curves, but now I have to set saturation to 200%.
Finally, this frame is the XAVC proxy file given after going to the Cine EI mode in order to record a 4K image on the AXSM recorder. All other paint functions are greyed out, including whether or not you are shooting the proxy in S-log-2 while the 4K image is in raw. The uncorrected proxy looks like S-log-2 to me and this is the corrected result. You will also notice the little extra real estate on the sides and black bars top and bottom. The frame size is still 1920x1080, but this mode seems to give all the raster and then some.

And finally, the 4K image.
This was recorded at the same time as the proxy above but turned out very different. In RAW I expected a very flat, dim image, but instead this seems over driven. I wanted to correct it in SpeedGrade CC but every time I loaded it in, the computer crashed! Even loading a single DPX frame made the computer crash! Have to figure that out later. Might be a corrupt file. But trying to deal with 4K on a machine not up to the task is asking too much anyway.

Thinking about how I would use these gamma curves is a bit of a quandary as several seem to have very slight change to them. I could see rarely using one of the STD curves (more likely STD6) when I want a lot of saturation with minimal color correction, and I could see using HG3, HG7, HG8 when I want less color saturation and more highlight protection with minimal correction. But since there always seems to be a need to adjust something, I might just as well shoot most things in S-log-2, normal color space, for confidence in highlight, mid tone, shadow, and color saturation protection. Not sure when I would use S-gamut as I'm trying to figure out how I could make a case for that much color space in the almost all Rec.709 world of mine, but nice to know it's available for the right producer.

Stay tuned for more to come!


No comments:

Post a Comment