Q & A Session with Dan Hudgins
August 30, 2014 at 7:55 pm #699
[Dan. I am the new owner of a KineMINI 4K Camera. I understand you need to apply the LUT included in the shot folder to get the Log curve and colors to a good place to begin grading. I have a few questions if you would be gracious enough to help me. ]
I am not in control of the user interface in the camera or anything else really although I have tried making some suggestions, so maybe someday the user interface will make sense to someone outside China, I have tried my best for them to address some issues, but don’t hear from them much any longer, Cheng is busy he says when I try to talk about things, and I have not heard from Jihua since NAB 2014.
[1. Right now I am filming Cinema DNG files, converting them to Cineform RAW with the KineStation software they provide, adding the LOG LUT in each shot folder in post, and then begin to grade. Am I doing this correctly? Is there a better way? I am a Mac user.]
I don’t have a MAC, but if you use a PC you may be able to use KineSTATION to make a Prores version with the shot’s 3D-LUT burned in, that may make a simpler work-flow, I don’t have KineSTATION as there is a license fee for that and it would not be of any use to me as I just use the DNG.
You can use the Cineform “raw” version, made in the camera (2K,HD) when shooting, or made in KineSTATION (4K, 2K, HD) during download, but you have to manually move the 3D-LUT into the Cineform codec so that each shot can find the right 3D-LUT on playback, the Cineform has a “signature” that matches each shot Cineform version to the mating 3D-LUT file, so Cineform can load the right 3D-LUT “on-the-fly” during playback and transcoding.
On a PC you drag the 3D-LUT’s into a program that comes with the Cineform codec called something with lut in its name, its been some time since I did that as I don’t have a Cineform licence. Once you load the 3D-LUT into Cineform codec, you use the codec settings to set high or low quality, low for editing in real time and high for trans-coding to DPX for making the DCP (JPG2000).
I don’t know how that works on the MAC, you will have to read the procedure on the go-pro/Cineform web site or ask David Newman on DVINFO, he had a forum about his Cineform programs there.
Is there a better way, yes use the DNG without the header information, but that takes some skill, in my programs, which I am not suggesting that you use, I have a probe to find something near white to calibrate the white balance. If you want to work with the raw data, its best to set the right K and light type when shooting as it is burned in by adjusting the analog gains for red and blue, then shoot a white paper, a stack of white copy machine paper is close to neutral, or the white side of a gray Kodak gray card, so you have an accurate reference when grading.
I do not have a mac version yet, my programs can run it seems under Ubuntu Linux. I may make a mac version later as its similar to Linux.
[2. As a DP I was confused by their ISO mode and EI mode in the menu. I am used to the Alexa with about 7 stops over middle grey and about 7 stops below. I am hoping to get close to this type of range with my Kine MINI 4K. But, I do not know what the best settings should be in ISO mode. They are recommending setting the ISO mode to 3.6? I get you can shift the sensor’s sensitivity in ISO mode, but I don;t know what is optimal. Any insight would be so helpful. ]
As I mentioned their ISO/EI settings are confusing, I have argued with them for years about that, and they refuse to fix it so that there is a super “manual override” menu with two prompts, ISO curve (digital gain) ISO 80 to 2560 and Analog gain 1x to 64x.
First you set the ISO curve, 80 to 2560, at one point the marked that as 80! to 2560! on the display so you could know what you set, but I think they changed that so there is no way to be sure what you set now.
The problem comes in that when you make changes it may change the analog gain without telling you to hold the ISO setting, so if you pick the ISO 320 curve to get less noise, it may increase the analog gain to hold the ISO at 800, so you see less highlight range and more noise but don’t know why.
I’m not sure what I can say that will resolve this confusion, I can figure out what its doing some of the time, I do that by knowing that the lowest ISO is the ISO curve in each set, in other words in the ISO menu, if you see the lowest speed as 320 and pick that one, you probably have ISO curve 320 and analog gain 1x set, but if you set it higher its not always clear if you are picking a higher ISO curve, or just applying analog gain to the 320 curve.
The headroom for the various ISO curves works this way (it goes in 1/3 stops but I will just show the even stops here),
80! = 0 stops above 90% white, ~2.3 stops above 18% gray (EI 80 to 5120 with 1x to 64x analog gain on green channel)
160! = 1 stop above 90% white, ~3.3 stops above 18% gray (EI 160 to 10240 with 1x to 64x analog gain on green channel)
320! = 2 stops above 90% white, ~4.3 stops above 18% gray (EI 320 to 20480 with 1x to 64x analog gain on green channel)
640! = 3 stops above 90% white, ~5.3 stops above 18% gray (EI 640 to 40960 with 1x to 64x analog gain on green channel)
1280! = 4 stops above 90% white, ~6.3 stops above 18% gray (EI 1280 to 81920 with 1x to 64x analog gain on green channel)
2560! = 5 stops above 90% white, ~7.3 stops above 18% gray (EI 2560 to 163840 with 1x to 64x analog gain on green channel)
Speeds above 5120 should only be used after testing to see if you find them useful, speeds above 20480 should not be used for feature films but might be usable for SD documentary with noise reduction and pixel averaging re-size applied.
ISO values are approximate, there is a light meter correction factor in one of the setup menus that lets you correct the difference between the 18% gray tick mark on the histogram and your light meter so the camera and your light meter will read the same ISO/EI number on the camera.
Analog gain override menu upsets the ISO readout, so don’t adjust the green analog gain just the red and blue to manually correct the analog gain so the raw data is white balanced (which can give better quality using all 12 bits for all three colors and maybe avoiding pink highlights and color fringes etc. by having the signals equal through the ADC and into the work-flow, just as film scans are (should have been) white balanced for the film stocks in the scanner).
Cheng did not implement the 18% gray and 90% white tick marks on the waveform monitor, in “post” mode, for Cineon, the reference level for 18% gray is 470/1023 or 45.9% waveform.
For Arri LOG-C (which uses different values for each stop than Cineon) the 18% gray waveform would be 400/1023 or 39.1%, but I calibrated the 18% gray to the same code level for all look groups, so you should use the tick mark on the histogram NOT the waveform, as the reference.
On the histogram you will see TWO vertical lines, one white and the other gray, the gray line is the correct calibration signal for 18% gray card, the white line or “tick mark” is for 90% white card, you should use those calibration points and ignore the ISO setting if you want accurate exposure because light meters are not calibrated to 18% gray card and the T/ stop errors in the lens (which may be as much as 1/2 stop off) upset any ability to correlate ISO to actual sensor code values.
The 18% gray tick mark on the histogram should read correctly no matter if the histogram and waveform are in Post (showing what the monitoring table is doing) or in raw mode. In raw mode the tick mark will move left on the histogram as you increase the ISO curve, so that is one way to tell if you have changed the ISO curve, the higher the ISO curve the more left the raw display of the tick marks moves. In Post mode the tick marks more or less stand still because the monitoring table is “lifting” the shadows in ratio to “normalize” the 18% gray signal levels, obviously.
[ 1. Is there a place which I can purchase or download the KineLOGC and the KineLOGF look groups from you? I have a USB port on the back of the KineMINI 4K camera and can load LUT’s into the camera it seems.]
My tables don’t cost anything, and as I mentioned you can make your own in my free programs, although not on a MAC, and there is no documentation about doing it so I would have to talk you though the process at least once, or maybe make it for you if I had the time.
[2. Is there an advantage or disadvantage from recording Cineform RAW in the camera to start with and not worry about shooting Cinema
DNGs? Or is it better to start off with the highest possible quality (Cinema DNG) and then convert to something you can edit with (Cineform RAW) in Premiere aPro CC (which is our edit software)? ]
If the camera crashes while shooting the Cineform MOV may get corrupted, that can happen if the battery goes dead, like doing long interview shots. I have suggested they make a recovery program to “fix” corrupted MOV, I don’t know if they did that yet.
If you are shooting DNG, you only lose one or a few frames at the tail of the shot when the camera dies during shooting, probably, so that’s an advantage.
Cineform raw is a good quality format, but its lossy, if you can or cannot see any difference is something you would have to work out for yourself.
A bigger difference is the de-Bayer method used, the Cineform codec has its own high speed de-Bayer for real-time processing, with DNG you can use any method you want since the DNG are “open” format and you can write your own de-code methods.
It may be possible to reverse convert Cineform raw into DNG which would blur the distinctions between the options maybe, best to test the workflows you like for speed vs. cost vs. quality.
You must be logged in to reply to this topic.