JVC GY-LS300 provides high-quality images from SD to 4K, low-weight, minimum bulk, reasonable price, and great versatility.
In Part 2 of Steve Mullen’s Field Report on the JVC GY-LS300 4K camera, he looks at some of the camera’s advanced features.
Because Part 2 of my GY-LS300 Field Report depends on your understanding log gamma, you may want to read the previous tutorial “Looking Deeper into Log Gamma”. And, if you haven’t already, follow this link to download 3D-LUTs file for the J-Log1 of GY-LS300CH Reference Manual.
Why Shoot Log?
The most common reason folks shoot log gamma video is to record a larger Dynamic Range (DR) than possible when shooting with BT.709 gamma correction. We’ll come back to this topic, but there are other advantages to using the LS300’s J-log1 gamma.
First, to obtain the maximum dynamic range with BT.709, one has to appropriately set the camera’s Knee Point and Slope values (Figure 1). Enabling a camera’s Knee function can create an odd look to highlights as well as unwanted color shifts. When J-log gamma is engaged, the Knee circuit is disengaged so no Knee is applied thus you have no need to worry about settings or image issues.
Figure 1: A low-knee point setting can increase dynamic range.
Second, in the Camera Function menu you can ignore FAW (Full-time Auto White), Smooth Transition, and Face Detect. In the Camera Process menu, only White Balance is relevant. With the LS300, one must make the correct White Balance setting unlike when shooting RAW.
Exposing for BT.709
The LS300’s AltaSens Super35 sensor’s 12-bit A/D supports a theoretical dynamic range of 12-stops (72dB). The AltaSens specification sheet (Figure 2) lists its maximum dynamic range as 68dB. The LS300’s DR, in f-stops, is therefore 94% (0.94 = 68 ÷ 72) of 12, or about 11.33-stops.
Figure 2: JVCKENWOOD-AltaSens Super35 CMOS sensor specifications.
The AltaSens specification confirms a graph (Figure 5) found in the 3D-LUTs file for J-Log1 of GY-LS300CH Reference Manual that shows the LS300’s dynamic range to be 11.8-stops. The diagram also shows BT.709 dynamic range to be 8.6-stops. J-log1 increases dynamic range by 3.2-stops (from 400% to 800%).
Figure 3: GY-LS300 Dynamic Range: BT.709 and J-Log1.
The Yellow arrows point to the approximate signal levels for middle-gray (18% gray card) and 90% White when BT.709 gamma correction is employed. The signal level for middle-gray is 46.2% (Figure 4) while for 90% White it is 90% (Figure 5).
Figure 4: BT.709 18% gray card (middle gray) = 46.2%.
Figure 5 BT.709 90% White card = 90%.
The BT.709 zebra settings for middle-gray are 40%:45% (Low is 40% while High is 45%). BT.709 zebra settings for 90% White are 85%:95%.
Exposing for Log
When shooting log gamma, optimum signal values are lower than when shooting with BT.709 gamma correction. Figure 6 presents, for several brands of cameras that support log, these lower exposure targets. On average middle-gray is exposed at 37% while 90% White on average is exposed at 61%.
Figure 6: Middle-gray and 90% signal levels for several log gamma cameras.
The ratio of middle-gray and 90% White estimates the “contrast-ratio” of recorded video. The average contrast-ratio for these cameras is about 24%, which explains the flat look of log footage. Captured video will also have low chroma saturation because signal levels are low.
Figure 7 shows the chart available in the 3D-LUTs file for J-Log1 of GY-LS300CH Reference Manual. Looking at the ratio between the recommended middle-gray (35.6%) and 90% White (66.5%) signal levels, the contrast-ratio is 31%.
Figure 7: JVC J-Log1 signal levels.
Thus, we can expect J-log1 footage to be flat, but not as flat as captured by some other log cameras. S-log2 has, for example, a contrast-ratio of only 27%.
When you enable J-log1, exposure must be manually set as no AE functions are available. Once shutter-angle has been set appropriately—three other parameters define exposure: the ND filter setting, ISO or Gain level, and the lens aperture. The latter two parameters require special attention.
ISO or Gain level—At low illumination levels, you may find the LS300’s histogram’s distribution cannot fully extend rightward. In this case, try increasing ISO. For example, use ISO500 or ISO800 rather than ISO400.
Aperture—As a lens is closed down, the probability of Diffraction Interference increases. The probability is not independent of sensor size. The smaller the sensor, the less a lens can be stopped down without diffraction. Because VSM alters effective sensor size, when you set it to match a lens, you need to be aware that the lower the VSM value, the less you can stop down. This caveat also applies to the LS300’s Prime Lens Zoom feature.
How to Shoot Log
Prior to shooting log, if you haven’t done so, you’ll want to appropriately define your set of User buttons. While the GY-LS300 mostly has great ergonomics, the designers totally failed with two critical buttons: AF/MF Focus and AF Lock On/Off. These identical buttons are impossibly tiny, placed directly beside each other, and located behind the LCD hinge.
I would want the AF/MF Focus button to be a horizontal version of the vertical Iris A/M button, and be placed forward of it. (Now both A/M controls would be near each other.) A physically much wider Push AE/AF Lock would replace the tiny AF/MF Focus On/Off and AF Lock On/Off buttons.
The solution I worked-out was to assign the oddly named Push AF/AF Lock to button 6, which is located rearward of the Iris A/M button. The Gain and WB toggles are located between these two buttons. All shooting controls from Push AWB/Iris to Push AF/AF Lock are conveniently located next to each other.
Figure 8: My LS300 User button settings.
Using Exposure Cards—When shooting log there are four ways to set exposure: use an 18% gray card or 90% White card, use a light meter, or use the LCD/VF. Figure 9 provides an overview of correct exposure levels and the appropriate zebra settings.
Figure 9: J-log signal levels with zebra settings.
I recommend setting exposure by employing the LS300’s ability to set Upper and Lower zebra values. (Figure 10).
Figure 10: Recommended LS300 Middle-gray and 90% White zebra settings.
Setting exposure using an 18% gray card should enable the capture of over 5-stops of light above and below middle-gray. However, the AltaSens specified SNR is 56dB suggesting the sensor’s effective dynamic range is about 9 stops. (9.3 = 56÷68 x 11.3).
Using a Light meter—Once ISO and shutter-speed have been entered into a light meter, and either a reflected or an incident reading taken, the light meter will define the correct aperture setting. For example, f/5.6. The same f/5.6 opening should result from defining zebra to be 35%:40% and then adjusting the lens aperture until zebra fills an 18% gray card. When the ND filter is used, calculate light meter ISO by dividing the camera’s ISO by the value shown in Figure 11.
Figure 11: Lightmeter ISO compensation for ND filter settings.
Using the LCD/VF—Within the limitations of LCD technology, you can use the LS300’s LCD/VF to set exposure. Some camera’s make this difficult because only a log—flat—image is displayed. No such problem exists with the LS300. Simply use the LCD/VF > LCD/VF LUT menu item to enable a 1D-LUT (Figure 12). The LUT will modify the log image so it becomes a BT.709 image on the LCD/VF. (An uncorrected image is output via the LS300’s SDI and HDMI ports).
Figure 12: Set LCD/VF to 1D-LUT.
What I’ve Learned—I used several methods to test different ways of setting the camera aperture. FPS was set to 23.98 with a shutter-angle of 180-degrees. ISO was set to 400 with ND filter at 1/16. White Balance likely was 6000-degrees.
1. The simplest way to set the aperture for a Las Vegas city view was to use a lightmeter. (ISO was set at 25 to account for the ND filter.) Figure 13 shows a moderate dynamic range that was nicely centered in the camera’s histogram. Figure 14 shows grading corrections to Lift (lower), Gamma (slight Red > Green shift), and Gain (higher). Lowering Lift reduces noise. Saturation was increased.
Figure 13: Flat video from camera.
Figure 14: Graded video with color correction from the Color Finale FCP X plug-in.
2. Another way to set aperture was to assume portions of the city view could be treated as though they had a 90% reflectance. I set zebra to 65%:70% and adjusted the camera aperture so the frame-left white buildings were filled with zebra. Figure 15 shows a wide dynamic range waveform that was nicely centered in the camera’s histogram. Figure 16 shows corrections to Lift (lower), Gamma (slight Red > Green shift), and Gain (higher). Saturation was also increased.
Figure 15: Flat picture from treating white buildings as equivalent to a 90% White card.
Figure 16: Graded image with color correction.
3. To test a very wide dynamic range scene, I first set zebra at 35%:40%. After adjusting the aperture until an 18% gray card on my balcony filled with zebra, I panned toward the city.The upper portion of Figure 17 shows the ungraded signal from the 18% gray card. The lower portion of Figure 17 shows the graded gray card. In grading, I slightly lowered Lift which lowered the darkest parts of the scene. (Lower orange arrows.) Middle-gray was minimally adjusted as indicated by the lower orange arrows.
To increase image contrast, I raised Gain until the brightest area moved to 108% (Figure 18). Saturation, as usual, was increased. Figure 19 shows ungraded and graded city views.
Figure 17: Upper—ungraded video / Lower—graded video.
Figure 18: Grading adjustments to Lift, Gamma, Gain controls.
Figure 19: Upper—ungraded video / Lower—graded video.
When setting exposure in these different ways I was not trying to shoot great-looking video as I would when shooting BT.709. My goal was, as when shooting negative film stock, to achieve the maximum transfer of information from sensor/negative to post-production software. Log gamma supports my goal. By “grading to scope” (RGB Parade), rough grading doesn’t impose a burden as it takes less than a minute to grade a clip.
The primary advantage of log is not simply its greater dynamic range. It is the ability when grading to manipulate this dynamic range to create BT.709 images you want viewers to see.
The only critical task when shooting—avoid sensor clipping. This is why I rarely use the ETTR (Expose To The Right) approach that is required by some log gamma cameras. Figure 20 shows moderately severe sensor clipping. Figure 21 shows that an image can be graded to look better, but once details have burned-out, there is no way to restore them.
Figure 20: Ungraded clip with clipping.
Figure 21: Graded clip—using the Curves tool—better contrast but still with clipping.
The LS300 offers two additional tools you can employ when setting exposure: the Histogram and Spot meter displays.
Histogram—After you have set a scene’s exposure using zebras or a lightmeter, a histogram (Figure 22) will display the distribution of light intensities within the frame. The width of the distribution is proportional to the image’s contrast-ratio. If the ratio is low, the distribution will be narrow. If the ratio is high, the distribution will be wide. Keep in mind a dark scene, with one very bright light, can have high contrast while a bright scene, with no dark shadows, can have low contrast. For these reasons, histograms are not typically used to set exposure—which is why one should use zebras or a lightmeter to set exposure.
When weak signals appear in a histogram in Red, details can become noisy. Check exposure as it may be too low. Otherwise, add illumination. When strong signals appear in a histogram in Red, highlights are likely clipped. Check exposure as it may be too high. Otherwise, reduce illumination.
When Red data appear on both sides of a histogram, the scene’s contrast-ratio are likely greater than the camera’s dynamic range. Modify illumination until the Red data are gone.
Unfortunately, enabling the histogram eliminates the audio VU meters.
Figure 22: LS300 Histogram function.
Spot Meter—I recommend enabling the LS300’s Max-Min Spot function. (Figure 23) The Green cursor displays the approximate location of the brightest area within a frame while its level is displayed in Green text. (Yellow shows the darkest point and value.) When shooting BT.709, the maximum level is 400%. It is 800% when shooting J-log1. By checking the Brightness Indication, I can instantly check for clipping.
Figure 23: LS300 Spot-meter function.
If you haven’t yet downloaded the J-log1 LUTs, this is the correct URL: http://pro.jvc.com/pro/attributes/4k/soft/J-log_cube_files.zip.
To obtain information on using the JVC J-log LUTs with Final Cut Pro X and Premiere CC: pro.jvc.com/prof/attributes/inst_man.jsp?model_id=MDL102318&feature_id=11.
For Avid Media Composer: http://www.service.jvcpro.eu/webtemp/151216_Media_Composer_D-LUT.pdf?CFID=19967609&CFTOKEN=15197971.
Grading accomplishes the task of mapping log gamma data to BT.709 gamma corrected data. You do the remapping when, using a calibrated BT.709 monitor, you make eye-based adjustments to an image.
When importing clips using FCP X, set the Import Preferences as shown in Figure 24. Insert an SD card into the card reader and issue the FCP X Import media command. Because you can play each clip from the SD card, if you are tight on storage space, you can decide which clips to not import.
Figure 24: Set FCP X Import Preferences.
During import, the clips on the SD card will be copied into a Library for the Project. Each clip will also be transcoded to HD ProRes Proxy clips. Proxy clips will be used for editing. Exports will make use of the original H.264 files. (There is no reason to request clips be Optimized).
When editing with FCP X, to eliminate the flat look (Figure 25) simply drop a JVC Tungsten or a JVC Daylight LUT on clips you place in a Timeline. I use the motionVFX mLut utility. (JVC LUTs do not work with Color Finale’s LUT function).
LUTs perform a purely mathematical correction to log footage to force it to comply with BT.709 (Figure 26). The JVC LUTs work well enough for editing.
Figure 25: Flat look on source clip.
Figure 26: JVC Daylight 3D LUT corrected footage.
If your edit is a rough-cut, when you decide to grade it, you have three options. If you have applied LUTs, you must first disable/remove the LUTs. Now you can round-trip your Timeline to DaVinci Resolve. (See Round-tripping FCP X and DaVinci Resolve.) Or, you can employ the FCP X Color Board to perform the grade.
A far better option, especially for those who are used to industry standard grading software, is the Color Finale plug-in for FCP X: try.colorgradingcentral.com/colorfinale. (Color Finale was used for grading the test clips).
No matter which method you use to grade J-log1 footage, you also have the option to apply a specialty LUT to create a “creative” look or to emulate a film stock. Figure 27 presents an example grade created by the FilmConvert (filmconvert.com).
Figure 27: FilmConvert—GY-LS300 to Kodak Tri-X (400ASA) Super 16.
Among the dozens of topics not covered in Parts 1 and 2, there is space only for three exceptions. The LS300 supports capture of 2K (2048x2160) and 4K (4096x2160) video at 24.0fps. These formats should be employed for source capture only for productions that will be shown in digital theaters using 2K or 4K projectors. (When a UHD version is required, crop rather than downscale).
The near absence of rolling shutter artifacts, moiré, and aliasing sets GY-LS300 footage apart from that shot with DSLRs and other digital cinema cameras. The LS300’s freedom from intrusive artifacts as well as its excellent image clarity and natural color make the LS300 an ideal choice for those of us who are not looking for the soft, unsaturated look that currently is in vogue.
Although my early test footage was shot at 23.98fps, I soon began to shoot at 29.97fps because of its compatibility with 30p streaming, an HDMI 1.4 computer connection to UHD televisions, as well as the 2160p60 and 1080i60 video formats—the latter without any pulldown puzzles.
The JVC GY-LS300’s high-quality images from SD to 4K, low-weight, minimum bulk, reasonable price, and vast versatility—demand a try-out by anyone considering a new camcorder.
You might also like...
Information theory can also be applied to loudspeakers, which are among the most difficult of transducers to design. Measuring the information capacity of loudspeakers is a useful tool.
In the previous article in this series, we looked at layer-2 switching and layer-3 routing. In this article, we look at Software Defined Networks and why they are so appealing to broadcasters.
Capturing the essence of a location in a single shot or series of shots can present a range of challenges for the itinerant DOP.
Here we look at some practical results of transform theory that show up in a large number of audio and visual applications.
Much of the attention enjoyed by virtual production currently goes to the spectacular stages with LED displays the size of half a dozen cinema screens. The material we put on those displays, though, can come from a number of places,…