Last year we saw the first Chinese cinema camera, the Kinefinity KineRAW S35. It recently started shipping and at NAB they presented their new Kineraw MINI: A portable camera that delivers a very brilliant 2K RAW image from a 4K sensor. It costs 3800€ (inc. tax) and it should start shipping very soon.
cinema5D took the KineRAW mini for a test shoot and here’s the detailed review:

201311914444186You know the times when all things just “click” together allowing one to achieve great results?
That’s how I felt when Michel and Ludwig from “hd video shop”, currently the sole distributor of Kinefinity cameras were kind enough to lend us the new KineRAW mini for a few days. At the same time Ram Leetal who is a great DP, friend and mentor asked me to help and shoot/edit a small story about 2 young sailors who are trying to make it to the 2016 Olympics on their own.

These sportsmen are very determined to make it despite the many obstacles on their way. (the main one is finding a sponsor who can support them fulfilling their dream). What else can I ask for, a nice little camera to test and a great opportunity to try and help the 2 sportsmen achieve their goal….

DSC04709

The camera is available here for 3800€ (inc. tax): hdvideoshop.com
(shipping to US is 89€)

DSC04667The Chinese company Kinefinity currently has 2 camera models: The new KineRAW mini shares the same sensor with the bigger KineRAW S35. The mini is shipping now and available for pre-order worldwide. (3-4 weeks delivery time)

 

Physically the camera is small, very small. I have yet to find a photo that makes justice to its size…

Here is a short list of camera specification from the manufacturer:
• Super35mm Sized 4K CMOS Sensor: Recording format 2K, 1080p, 720p and 576p
• You can record 16 seconds of 100 fps in 720p mode or 48fps in 2K mode
• There’s a S16 mode
• Uncompressed Digital Negative RAW as 12bits CinemaDNG
• Two Mount Options: Native EF Mount (with Electronic Iris Tuning) and KineMOUNT
• Three Color Options: Black/Gray/Champagne
• 3D LUT: In-camera LUT files as KineLOG/Kine709/KineColor
• Light-weight and Compact
• Offline Output Cineform RAW Codec Option: Fast Post-Workflow
• RAW Port Option for external Cineform recorder: Turn MINI into Super-Camera Head

It’s worth mentioning that I had the Canon EF mount version and more importantly the company is about to introduce an external RAW recorder made by Cineform which will allow the camera to get full 4K resolution out of it.

The camera is priced very well for what it can do and deliver. It is €4402 for the camera body and basic kit which includes a side handgrip with battery, KineMAG (64GB SSD), Canon EF or Nikon F adapter and AC adapter.

I love the form factor and the picture quality coming out of this camera. Its ergonomics and usability are much better in my opinion than other RAW cameras in its price range. It is true that it takes some time to get used to the menu and some functions have yet to be fixed via firmware updates but when and if those are done I am sure we will look at a solid and powerful product.

While spending 3 days with the camera I did notice a few issues that deserved Kinefinity’s response. I’m happy to say that the company’s headquarters in China and the European distributor hd video shop were extremely fast to respond. I see it as a good sign for future servicing requirements, which had been a concern when the first Kinefinity camera was announced.

So here is what I noticed about the KineRAW mini and Kinefinity’s responses to it:
• The REC button is not flawless. Sometimes I was struggling to push it to start recording.
Kinefinty: “Yes, the REC button is not good. Our handgrip will have new rec button and more control ways to MINI, so that you will get operation experience like DC.”

• Recording can not start when the menu is active.
Kinefinty: “Yes, we will change the function to make the REC operation be top priority later.”

• Booting time is much to long.
Kinefinty: “Yes, we may provide SLEEP mode to make the camera sleep and wake instantly. But the boot time is long like iPhone.”

• Sometime when booting the camera there is no image displayed in the Live View. The only way to solve this issue is to reboot by disconnecting the power, then booting works normally.
Kinefinty: “Yes, booting failure happens some times.”

• Image Stabilization lenses do not work. Please activate stabilising in lenses if possible.
Kinefinty: “Will try”

• Kine709 has a green tint to it
Kinefinty: “You may use the AWB to get balance in field. The AWB operation can be found in this way… CONFIG–>LUT–>Gain Override–>Manual–>Auto WB, use 18% Grey or 90% white card to set the WB.
Yes, the LUT of Kine709 is a little bit green tint.”

• Timecode is a must! (currently there is only “free run”) I don’t want to think about all the audio files I have to sync manually which otherwise could have been done in DaVinci Resolve.
Kinefinty: “There is time code embedded in every DNG image indeed, so that the KineStation can extract the timecode and construct a timecode. Yes, the time code mode is free-run now. It will take some time to add record-run option.”
Kinefinty: “Actually, you can use KineStation from our website to transcode the CinemaDNG footage into Cineform RAW mov. It can integrate the DNG sequence, timecode track and audio track automatically. Then you can use Cineform workflow for post work. This is our recommended workflow. KineStation is very fast. It takes only 10 mins to transcode 64GB SSD with DNG into Cineform RAW mov on i3 notebook, in win7.”

• The in camera audio is producing 2 audio channels. They are in mono. Is there a way to make them Left/Right? (stereo)
Kinefinty: “Yes, it is mono now.”

• It happened to me twice, camera will not stop recording after pressing the REC button. The only solution was to disconnect the battery. After rebooting I had to rebuilt the SSD.
Kinefinty: “It may happen when you have some settings pop-up.”

• It would be be great if the fan would turn off automatically as soon as you hit REC. After finishing recoding it should turn on again – automatically.
Kinefinty: “Yes, it will have the function as smart fan control.”

• The volume when using headphones is much to low (even when volume is turned on to a maximum)
Kinefinty: “Yes, the headphone volume is low…”

• The active EF mount sometimes stops giving aperture reading to lenses which it gave before. After re-attaching, it was fine.
Kinefinty: “OK, we will check the Iris issue, it happens seldom.”

• Biggest issue: Slow frame rate of the EVF: When using an EVF or monitor via hdmi the framerate on it is slower which is disturbing and appear lagging.
Kinefinty: “Actually, it is around 20fps, not 25fps. It seems maybe the choppy issue makes the delay severe.”

• There is strong rolling shutter like on DSLRs.
Kinefinty: “For 2K in S35 mode, yes the rolling is bad, you may try S16 mode, the rolling is much better if you upgrade the camera to new firmware.”

• Internal audio quality is not good.
Kinefinty: “Yes, the audio quality is not very good, though.”

DSC04664

All in all as you can see from Kinefinity’s response, some of the issues are fixable via a firmware update and I’m sure the company will do its best to achieve that soon.

I must emphasize that despite those issues the camera is fun to use and the picture quality is very nice!

One thing I find conflictive is the fact that this mini camera could have been great for shooting documentaries if it had some sort of compressed file recording option or alternatively allow a clean full HD output from its HDMI connection. You could go straight into an Atomos Ninja for example. I guess the hardware does not have the power to do so.

Slow motion:

kineraw_720aliasing In 720p mode it is possible to record 16 seconds at 100fps. As always, slow motion can be very impressive and 100fps is special for an affordable camera like this. Unfortunately aliasing and moire are noticeable in the resulting image yet the footage is still usable as you can see in the video.

 

Workflow:
I’ve imported all cinema DNG footage to DaVinci Resolve 9 lite (download free here) and exported to ProRes 422 1080/25p.
I then edited in Premiere Pro and graded in Filmconvert (10% off with code “cinema5D”. LINK)
A small amount of sharpening was added in post.

HDvideoshop gave us a special deal for our readers:
Free 128GB Samsung 840 Pro SSD with your KineRAW mini order. CODE: cinema5d
3800€ (inc. tax) available here: hdvideoshop.com
(shipping to US is 89€)

My fellow cinema5D man Nino Leitner shot a low light and latitude test comparing the KineRAW Mini and the 5D Mark III RAW, which will be posted in the next days on cinema5D too. Stay tuned!

The music was kindly provided by The MusicBed: http://c5d.at/1rg

DSC04665

Many thanks to Bar and Omer who contributed from their time for this video.
You can follow them here: https://www.facebook.com/BrandHalevySailingTeam

Reply
Dan Hudgins September 6, 2013

The color balance can be changed by adding new look groups, if the firmware has not been altered to prevent it, my free de-Bayer program can make new LK5 files which set the base ISO and Look Group, and the LT5 file which sets the analog gain values for white balance can be edited in any text editor such as notepad.exe. You can email me if you want to add or develop new look groups aside from the ones you get from Kinefinity.com (sm)

Reply
Dan Hudgins September 6, 2013

With regard to the audio quality, as far as I know the on screen VU meters are calibrated to WAV file clip, so the audio should NEVER be set to the red area except for recording a pure sine wave, for normal uncompressed audio the VU meter should read -20 and the audio gain in the camera be set to near minimum. The external preamp should not be outputting “professional” 0db VU of 6volts or more, as you may overdrive the input to the camera, if you want the external mixer VU meter to match the camera you can insert an adjustable pad between the mixer/preamp output and the camera input to avoid over-driving the camera’s input. Its a good idea to insert a 600 to 600 ohm line transformer between the camera’s 600 ohm 3 pin XLR and the balanced line from the pre-amp/mixer if the cable is long or there is a chance of EMI/RFI on the cables or input to the camera. Its a good idea to insert a peak limiting compressor between the pre-amp/mixer and the camera’s input (or line transformer) to avoid clipping the WAV files. Its a good idea to set the wind filter in the preamp active, that may be marked “voice” or “high pass” or “low cut” on your pre-amp or mixer, as you want to avoid low Hz from getting into the camera’s audio input and causing clipping of its circuits or the WAV recording from wind noise or mic handling. You can also use external Dolby A or DBX 2:1 compander to reduce digital crossstalk and increase the dynamic range of the camera’s internal recording, that would be inserted after the peak limiting compressor and before the line matching transformer. As far as I know there is no phantom power or low noise preamp in the camera, so you cannot plug in a microphone level signal and expect to get good results as its intended for line level signals that are pre-leveled to the target signal level, which with normal voice uncompressed would be about -20db VU on the internal meters when the audio gain is set near minimum, so you need an external pre-amp with adjustable gain of high quality to get the mic level signals up to the line level required, you should check the quality of your line level signal before thinking the problems are just with the camera’s recordings, which are 16bit 48000 s/s and should be better than CD quality except for the digital crosstalk. Avoid having audio cables near the monitor or camera as they can pick up EMI/RFI, likewise for a 600 to 600 ohm line balancing transformer.

Reply
Peter September 6, 2013

Waiting for a KineRAW Pocket for under 1000 euro :)

Reply
Dan Hudgins September 7, 2013

Email Jihua and suggest that, the S8 was meant to be under $1000 and out 5 years ago…

Reply
Dan Hudgins September 7, 2013

I asked if BlackMagic could support 2048×858 with uncompressed DNG recording in the Pocket, but I did not see a reply on their forum. 2048×858 uses less bandwidth than 1920×1080 so they should have enough bandwidth to support that in their compressed DNG mode if that is working at some point, but then what do you do about the lack of OLPF filter (?) when you have things in focus with a sharp lens. From my experience with the KineRAW-S35 ™ its OLPF filter is effective with sharp Cinema-35mm movie camera lenses, and give 35mm angle of view using those lenses including the Angenieux 14.5mm in Eyemo mount. Really, if your target is a DCP in a theatre, and you like 2.39:1 format, using the Mini in 2048×1080 mode helps avoid any blow-up so you can fill the correct with on the DCI projector. With the Pocket and cinema lenses how much chroma moire are you going to get when some things are in focus? Chroma moire is not removable through software in a way that can equal using a good OLPF filter which is why real professional cameras like RED ™ and ARRI ™ use an OLPF filter in their cameras, and why KineRAW ™ cameras use an OLPF. The issue in the Mini currently for modes that use more than 2x binning is that they did not put a clip in the camera so you can change the OLPF filter to match the binning ratio, its something that was understood and Jihua made the final say on such details, you can talk to him about any design changes you feel are important.

Reply
Dan Hudgins September 7, 2013

With regard to SMPTE time code, based on what the firmware in the S8p and S35 models was doing, the SMPTE LTC output should correspond to the TOD embedded in the DNG headers, my de-Bayer software can burn that into the image space for workprint (my Brother likes to edit 24fps converted to 29.97fps so that’s one way to keep track of the frame accurate editing for the conform in 24fps). You should be able to use the camera as a SMPTE LTC source or generator, and have that go to the external sound recorder and time code slate. I talked to them some time ago about adding a voltage controlled capacitor in the sensor clock to allow phase-locked frame sync using the SMPTE input, but rather than implement that for 3D they wanted to use a separate multi-pin connector for multi-camera sync. Again, in that case if you need to sync both a Mini or S35 and some other brand of camera you can use the KineRAW ™ as the SMPTE LTC generator and slave the other brand of cameras to it assuming their camera will phase-lock correctly to incoming time code. So the lack of the KineRAW ™ doing phase-lock to incoming SMPTE is probably not an issue because you just use one of the KineRAW ™ as the master, sync the KineRAW ™ between each other with the special cable, and slave the other brands of camera, external audio, and SMPTE slate to the master KineRAW ™. I purchased a SMPTE LTC reader with LED readouts, when I get time I was going to shoot the display and then burn in the SMPTE embedded into the DNG headers to show if there was any drift, which there should not be. I have in my de-Bayer program added pass through of the SMPTE time code as well as the date stamp for each frame to the DPX header if DPX file type is selected as the output. In the anamorphic test on my Vimeo page I think you can see SMPTE burn in shown in some of the behind the scene shots at the end of the video, those came from the DNG header which can also flag a burn in for the head and tail auto-slate beeps. For making burn in from DPX files the idea is the same as the SMPTE time code should carry over when those options are selected in the KineRAW ™ camera set-up. The auto-slate beeps can go to the internal track for making it simpler to do looping or ADR when recording to WAV recorder on playback, you put the camera sound on the right and the mic sound on the left, then when you import the looped stereo track you see the head and tail beeps so you know were to trim the looped WAV, the head slate beep is 1000Hz and the tail 2000Hz so you can tell them apart by ear or looking at the waveform in your audio editing software. There is (was) a special pin on one of the connectors to output the auto-slate beeps line level so that you can record the auto-slate on an external audio recorder, that solves all the audio issues since you get clean audio for all shots and don’t need to use a clap board since the auto-slate marks the head frame and tail so that you can trip the audio even if you don’t have the image frames in sync or loaded. You then just drop the audio from the external recorder back in on the first frame of each shot, or drop in the looped track, and its in sync without fiddling since you don’t need to look for were the slate closed, the head beep is on the first frame of that shot (the tail beep may be a few frames short of the picture tail of the shot so you can trip the tail a bit long if you want).

Reply
Dan Hudgins September 7, 2013

With regard to the low earphone level, you may be able to use a splitter cable and two high-Z crystal headphones to get more volume, here is one listed on Amazon, http://www.amazon.com/Crystal-Radio-Earphone-Impedance-3-5mm/dp/B0052EE0AU because it fits into the ear it also blocks out external sounds. You may want to add a non-polar capacitor in the earphone lead to block DC offset from the headphone output that could clamp the diaphragm and make muffled sound.

Reply
Dan Hudgins September 7, 2013

With regard to the white balance or other color problems, there is (was) a menu that lets you set the analog gain absolute values in the sensor manually, you can use that to get an exact white balance. To do that use a white card as a gray card may show IR issues, adjust the lens iris and camera shutter so that when you are zoomed in 800% on the white card the green waveform reads 45% post (monitoring not raw). Be sure to have any ND or other filter on that you want netural on the camera during these adjustments, and turn off any lights that will not be on while shooting. After you get the green waveform to 45% post (monitoring), set the histograms to readout post RGB, not luma, so you can see the color balance as three peaks red, green and blue, then go into the manual analog gain adjustment menu, and adjust the red gain to have the red peak overlap the green peak, if the red peak it to the right of the green peak you decrease the red gain value, if its to the left you increase the red gain value. You then do likewise with the blue. Do not adjust the green gain as that changes the EI ISO rating, well unless you are under about 2000K or over 7000K or have some odd light source that requires the red or blue to be lower gain than green, but most sources green is always the lowest gain. Once you get the three peaks aligned you can write down the three gain numbers, if green is not 1.0, you can find the ratio by dividing the red and blue by the green value. You can then enter those results into the LT5 file using a text editor and then copy the edited LT5 file onto a USB flash drive to load into the camera, it will then appear in the cameras menus with the label you gave it after you re-boot the camera. At NAB I set the color balance right for those lights, but everytime someone fiddled with the camera it reset to icky green, so I asked Cheng if he could have the camera add the manual analog gain adjustments to the LT5 in the camera automatically letting the user enter a label in the same way the Slate data is entered for the DNG meta-data (one letter at a time with the encoder). That way you can save user presets for the white balance, like for your set of LED lights etc. You can ask them to add that feature if they leave it out of the firmware revisions. I sent Cheng information about how to do the auto white balance, if its not working right maybe there was some problem understanding the formulas I sent, or the imposed some shortcut of their own, you can tell if the white balance is working right by looking at the three peaks on the post histogram as described above. If the histogram looks right, but the processed DNG are not netural, then its a problem with the DNG header values and the software you are using, you can override the DNG header values and maybe solve that problem, my de-Bayer program ignores the DNG header values so with full manual adjustment the results should be netural if the camera analog gains were giving netural results. You can check if the LK5 are netural by selecting RGB RAW histogram and opening the iris a bit, the peaks should align in the same way as the Post RGB histogram, the LK5 “looks” I sent them have identical curves for R,G, and B rather than compensating for any drift in the prototype S35 they sent to do the calibrations with, so the histograms should be netural in both RAW and Post modes. You can make new look groups of your own using my programs if you feel the need for that, the 3D-LUT made in the camera can be modified by the output range selection to get, full range, ITU 601 limits with midtone at 46%, ITU 601 limits with midtone at 40.7%, and Cineon ™ film LOG with softclip. The KineLOG outputs Cineon ™ film LOG without softclip (range selection in that case is full because the internal data is LOG). The camera was designed (originally) to accept user defined LK5 (“looks”) and LT5 (white balance) settings to deal with special issues such as more or less highlight range, lower mid-tone, or odd light sources. At some point their developing their own “look” maker software was a prospect, but they have not talked to me about how to go about doing that, so my free program can be used to generate LK5 and the latest LT5 can be edited in a text editor such as notepad.exe ™. You can also remove light sources from the cameras menu if you want by editing the EQUAL.LT5 file, just delete any lines you don’t like, and then add new ones if you want, you can then load that into the camera and the menus should update on re-boot.

Reply
Dan Hudgins September 7, 2013

If you want to change the color (chroma) matirx values, black level, and other settings passed to the DNG header of the frames saved, you can do those in my de-Bayer program by editing or making new LK5 files. The camera will then read the revised “look” files and if they have the same filename as the original LK5 files they will replace the current settings with the new ones, and so when that “look” is selected, the DNG header values will be changed, so if you have problems with the color balance of the camera with any given workflow you can alter it. Likewise, the camera’s monitoring and the 3D-LUT that matches it with the Cineform ™ workflow are based on the matrix and curves in the LT5 file, so you can fix issues in the Cineform ™ workflow if you adjust the LT5 files, and then use the resulting 3D-LUT in each shot folder to load and view the Cineform ™ produced by DNG to Cineform ™ conversion (at least that is the way the S35 works and so should have followed through to the converter program for the Mini version).

Reply
Nathan Corrona September 7, 2013

great info Dan. How does the super 16 mode work on this camera? is it just windowing the sensor? Are there any frame rate advantages doing this (or is it 2k 60p and 100fps 720p too) ? I am considering this as a b-cam for my A-CAM DII- mainly for the frame rates.

Reply
Dan Hudgins September 7, 2013

If you and Cheng are at IBC he can tell you what the current firmware is doing, the sensor can bin and skip like other sensors, so I asked them to add 1:1 pixel modes as its a “4K” sensor, that means you can get about half width in 1:1 pixel mode. Since the OLPF is just good enough for 2x2x binning, using 1:1 pixel should show little aliasing and still work if you increase the anti-OLPF sharpen, the noise is low enough to stand up to the extra sharpen probably if you don’t push the gain up too much in S16 mode, you should be able to get speeds higher than Acam dII ™ though because of its splits calibration issue when under exposed, although it seems Acam dII ™ got some improved firmware to help with that problem. No splits in KineRAW ™ cameras so far as I have seen, and the FPN is under control although you can mask it in my software if you need to shoot and some shows up as with the DNG and my software you can make the black frames BEFORE or AFTER you shoot, unlike the Cineform ™ (and REDCODE ™) which requires shooting the black frames BEFORE you shoot. So with DNG you can save a shot that shows FPN better maybe. Anyway, the idea was that the sensor supports windowing, so you can get 2K maybe in S16 size, as well as 1828×1556 for anamorphic (if Cheng has had the time to add that) because the record bandwidth at 24fps is enough for that many pixels (although the sensor bandwidth seems faster for binned sensor areas for some reason). They also wanted to use 3x and 4x etc. binning for lower resolutions and higher frame rates without the need for smaller lenses (the KineRAW-S8p ™ shoots up to 220fps but in a small 648×276 window so the lens crop factor is large), but the issue there is that the OLPF is optimized for 2x2x binning area, and with larger sub-patches of the sensor some aliasing and chroma moire can show up on things that are high contrast and in focus. The solution to that would be to add another OLPF in front of the one there and re-set the lens FFD, or swap the OLPF altogether to get one that is optimized for each binning ratio, Cheng suggested having the OLPF mounted in the lens adapter so you would change lens adapter for the needed binning ratio, 1x, 2x, 3x, 4x, etc. That makes for more adapters and some dust problems with dust getting between the OLPF and the sensor cover glass, but solves the aliasing, resolution, and moire issues. Jihua seems to have not responded to that technical point, and gone with one permanently mounted OLPF. You could always pry it off the sensor case and make your own mounting clip maybe to switch the OLPF as needed which is how it should have been designed, but its Jihua’s decision based on cost issues that rules? You can talk to him about this, he may listen to you since its part of your purchase decision.

Reply
Dan Hudgins September 7, 2013

The Kine709 LK5 should be netural and have equal curves for R, G, and B. If not they made changes or the files got mixed up because that is the way I made the last revision, try a different headroom setting and see if the tint changes, some of the headroom settings showed black level offset causing a color imbalance in the black level, if that happens it is not the look group that is at fault. They have something called “expert mode” or something like that that turns on the headroom adjustment, try not having green analog gain near 4x and see if that impacts the color balance, you can also go into the manual white balance menu and make a small adjustment and see if that effects the color balance more than it should, showing the gains are near one of the invalid levels. The saturation is higher for Kine709 obviously, if you mixed illumination you will see more color shifts, if there is daylight and florescent or tungsten and florescent the shadows can go green because that is were the florescent fill the shadows with greenish light, likewise if you are near plants or anything that is green to start with. If they knew there was a problem they should have asked me about it, I would have told them its not do to the look group unless there is an error some place and that should have been looked into before releasing the camera.

Reply
Johnnie Behiri September 7, 2013

Hi Dan.

Thank you very much for taking the time and sharing your amazing technical knowledge with us. Appreciated!.
Hope to see you at IBC.

Take care.

Johnnie

Reply
Dan Hudgins September 7, 2013

Hi Johnnie, I don’t think I can afford to come to IBC, but I told Jihua if he can send me the Mini with the latest firmware, OLPF, and sensor, I can take some months to re-do the color calibrations, the issue was that I told years ago that the color calibrations should only be done AFTER the cameras was locked and no changes would be made, so that the LK5 and LT5 would work over the product life, as it turns out a few weeks after I spent months doing the first production quality calibrations they made some changes to the camera that upset everything, and asked me to do it over again, I went to NAB to try to talk to them about what happened to upset the calibrations and did not really get a useful answer about that, but when Jihua emailed me after I told him it would be simpler for me to re-do things in San Francisco since we get good clear sun here for the 5600K Daylight reference exposures needed, they every everything calibrated in 1/3 stops and 50K increments, so that is a lot of reference points to get from EI ISO 80 to 163840 and 2000K to 7000K, right. No answer from Jihua about getting that done, so I am sorry about the color issues, I have not seen any data from the revised sensor so cannot compare it to the previous one, which is why I wanted to keep the S35 prototype #3 here to shoot side by side calibration frames to try to see what changed exactly and how to best compensate for it. I spoke to Cheng about making some additions to the LT5 file that might improve saturated color balance across the K range, and the LT5 should also support unbalanced and other color models, right now only EQUAL.LT5 is supported by the firmware. Originally the camera was designed to work with any color model including native sensor balance, so the data path and LK5 structure allows for that, its possible to load maybe 25 or more “look” groups at once into the camera, Jihua seemed to think that too many options would confuse buyers in Asia, which is why they crippled the headroom adjustments, and force you to enable “expert” mode to set the headroom, which is really setting the EI ISO curves by “sub look group”. Each EI ISO curve from 80 to 2560 can be gained 1x to 64x, so 80 becomes up to 5120 and 2560 becomes up to 163840 although speeds above 20480 are not useful for 2048×1080 DCP use, but could be used for SD/DVD with heavy noise reduction applied maybe. 80 x 64 = 5120 may show less noise than 2560 x 2 = 5120, but would have less highlight range over 90% white subject. Normally you would not select speeds slower than 160 x 1 = 160, which reads out as 160! on the monitor telling you that you are set for minimum noise with green gain at 1x, red and blue gain are above 1x to get white balance which can improve the wavelet compressed Cineform ™ raw conversion as well as avoid some color aritfacts in the DNG workflows, and also makes the workflow more like film scans since all the frames are white balanced and can be used as is if you set the K and light type right when shooting. In the KineRAW-S8p ™ there are two menu options for the K and light type, but for the KineRAW-S35 ™ Jihua wanted to reduce the number of settings and combined both the K and light type into one menu, the disadvantage of that is that you need to scroll through more options looking for one in particular, I suggested that they make the encoder velocity sensitive so the faster you turn it the more options fly by, I talked about that years ago but they may not have understood what I was saying, and now the hardware may need to be re-worked to have a counter on the encoder rather than doing an interrupt for each count(?) But if you ask them to add velocity control to the encoder they may listen to you. If people want their own look groups and the camera still supports my software’s output I can try to help you get that working, I think some softer look groups could be useful for DCIP3 and filmmaking use under harsh lighting so it does not matter if Kinefinity.com ™ wants them or not, if the camera still supports the LK5, they can be made and shared by the camera users which is why I designed the data path and files to be user programmable (somebody always wants something different, more shadow contrast, less green saturation etc.).

Reply
Myles September 8, 2013

The sample video looks quite muddy. Did something go wrong in the export?

Reply
Johnnie Behiri September 8, 2013

Hi Myles
Nothing was wrong with the export.
I guess what looks muddy to you looks just fine to me.

Thanks!

Johnnie

Reply
Dan Hudgins September 9, 2013

In general putting 12bit data from the camera into monitor range, without crushing the highlights too much makes the images look a bit flat. Also one tends to have midtone under 46% where it would be for computer monitors graphics, but if you look at the images in a darkened room, like a movie theatre would be, then it should look less muddy if you adjust the monitor black level (better done in your own grading software as LCD monitors don’t really have a “brightness” control that sets the black level like a CRT monitor sometimes). ARRI is working on new DCI projectors calibrations that give more data range above 18% gray to double the highlight brightness in movie theatres I have heard, in that case one would grade the DCP to have midtone lower CV than 46%, but as I mentioned if your monitor is brighter than a DCP projector looking at in a pitch black room can have the same effect in your eyes if the black level is set right on the monitor. Some monitors have a “look” setting that changes the gamma, you can try that as well.

HD monitors require ITU 601 limits, so if the person grading has a monitor that clips at 16/255 and 235/255 those images will look muddy on a monitor that displays full range 0/255 to 255/255, so you need to keep that in mind, the reverse is true if you grade for full range and then look at the images on a HD ITU 601 limited monitor you will see clipped highlights and no shadow detail.

Try downloading the video and re-grading it for your monitor. You can see other videos I have made on my Vimeo page, google KineRAW Vimeo Dan Hudgins or just KineRAW Vimeo and KineRAW YouTube to see more samples.

Reply
Dan Hudgins September 9, 2013

What I have been doing on my Vimeo videos now is to include both full range and ITU601 limited versions so that people with monitors that clip can see the full highlight and shadow range maybe, but the ITU601 limited version will look “muddy” to people that have monitors that are not clipping as much. Also some people may have video drivers that already do the ITU601 limits, so in that case the full range would be displayed limited, and the limited version would show with double limits making it double muddy. This is not a camera problem, its a monitoring issue and one of grading for a particular monitor or projector because LCD and other non-CRT monitors don’t always conform to the standards for range, gamma, saturation or gamut (sRGB etc.) If you use things like Adobe gamut rather than sRGB or have a OLED or wide gamut monitor the footage must be converted from its source XYZ color space to the monitor RGB color space before viewing.

Reply
Dan Hudgins September 9, 2013

Depending on how he camera’s output range limits are set, when you import the data using the 3D-LUT on the source frames, you need to check if the program is importing as full range or ITU601 (HD) range limited. If you shoot KineCOLOR with the default range limits set to full range, then it needs to be imported full range if the 3D-LUT that the camera makes is involved in the workflow. Likewise if you use Kine709 and import the footage (and the 3D-LUT is made with default HD ITU601 limits active), with the import range limits at full range (as would be used with default KineCOLOR) then the images would look muddy since black would be high and white would be low. The reverse is true as I pointed out, that if import default KineCOLOR as Rec.709 the highlights and shadows will be clipped. That is why there is an output range override menu, to match the 3D-LUT range to your workflow, the defaults are set to what seemed to make sense, KineLOG is full range for making 10bpp DPX files range 0 to 1023 where the data out is 95 to 685 plus the super white that varies depending on the EI ISO base curve selected, KineCOLOR is full range but can be reduced as needed with the output range override, and Kine709 defaults to HD ITU601 limits because that is the normal range for Rec.709 type footage so the DPX files made would range 64/1024 to 940/1024 (4×16=64, 4×235=940).

If you are using the DNG frames without the 3D-LUT the camera makes for each shot, you need to manually tell your software what target range you want to make on export of the RGB, YUV, or other format or color space that is your target workflow.

Reply
Dan Hudgins September 9, 2013

I should note here that if you convert the DNG to Cineform ™ and don’t use the 3D-LUT the camera makes, then the color matrix will not be applied since that is part of the 3D-LUT, in that case if you just apply saturation increase in grading the ratio of saturated red, to saturated green, to saturated blue may come out out of balance, meaning you may get too much red or blue or green saturation because the matrix was not applied to balance those as part of the 3D-LUT compensating the Cineform ™ LOG90 native sensor ADC output (levelled to equal with analog gain) to equal saturation for the primary colors on saturated subject matter. This is very different from white balance, you can have perfect white balance but still have unbalanced saturated colors do to not using the 3D-LUT in the workflow (or because of the LT5 not being revised for the current sensor plus OLPF balance etc. and light type (some light types may not be full spectrum so cause less exposure for some colors of the subject)

Reply
Dan Hudgins September 9, 2013

Related to the sensor XYZ color balance issues, going to a standard file type, is the issue of various monitors and projectors not using exactly the same primary colors and that monitor colors are not 100% saturated by definition their gamut is desaturated. sRGB is intended to be a monitor standard that is a sub-set of what monitors can show, but if an sRGB graded images is shown on any given monitor without a corrective compensation to reduce the color space to sRGB, then excess saturation will be seen, so look at the software you are using and see what steps you need to take to insure you are viewing sRGB or whatever color space you need while you are grading. So there 11 important values, the RGB ratios of the monitor primaries and the white and black levels which one assumes are the same for the primaries (but which may not be in LCD monitors that look violet in the shadows on a netural gray step test image). If you compensate for lack or netural tracking in your monitor over its brightness range, such as adding green in the shadows to compensate for violet shadows on a LCD monitor, then when the graded footage is looked at on a projector or monitor that does track you would see the inverse color error, green shadows, and so on. Don’t grade for your own monitor, you need to grade to the common standards to make the files compatible, something harder to do than say since you cannot know exactly what that would look like since every display device shows the same file a little differently.

Reply
Dan Hudgins September 9, 2013

With regard to the monitor ITU 601 limits use on the camera, the firmware should have monitor limits menu, in that menu you can setup the monitor limits to 16 for black and 235 for white, or whatever your small camera monitor works best with. The highlight extension is to help avoid clipped highlights on the camera monitor, you can increase it from 0 to the value that works best, maybe around 8 or so, too large or small a value will reduce the highlight detail visible on some LCD monitors. The camera monitoring adjustments also allow you to adjust the monitoring gamma which can compensate for the ambient lighting your eyes are used to if you are shooting in bright sun or at night, or if your small monitor is not gamma calibrated. You can also activate RGB mode for the monitor adjustments to compensate for small monitors that have a purple black rather than are neutral on all gray steps. To help with the monitor adjustments there are test patterns in the guide line menu, the color bars also has a gray step and small “plunge” type patches for checking the ext ream highlight and shadow detail range on the small LCD monitor you are using. Its important to check the monitor adjustments if you change monitors as no two monitors will look the same. The camera monitor adjustments do NOT change the recorded signals, trim adjustments have not been implemented for the 3D-LUT yet as far as I know, but you do have output range limit adjustments for the making of the 3D-LUT, full range, ITU601 (HD) with midtone at 46%, ITU601 (HD) with midtone at 40.7%, and Cineon ™ film LOG with softclip, these works with KineCOLOR and Kine709 but not KineLOG as KineLOG should always be set to full range output since it is not soft clip but uses super white over 685/1023 levels. Kine709 is full range internally but has output range limits set to ITU601 (HD) limits with midtone at 40.7% as the default (forced by a flag in the LK5 file). The output range limits do not impact what you see on the camera monitor, just looking at the footage using the 3D-LUT made for each shot. The monitoring adjustments do not impact the 3D-LUT or what you see when you look at the recorded footage.

Reply
Dan Hudgins September 9, 2013

I got this back about seeing the KineRAW ™ cameras at IBC next week,

Quote: [Hi Dan

We will not attend IBC next week. But there will be a booth shows our
cameras, its number is 11C 80 (Lanparte Booth).

Jihua]

Reply
Robert A. Ober September 13, 2013

So the price works out to 5K US$. Won’t sell any here. Camera has a lot of issues for that kind of money. It would at least need to do 4k out and even then the upcoming BlackMagic 4k cam would be the better buy.

Your praise for an unfinished camera limits your credibility. Hopefully for them Dan can get them to listen to him.

Robert A. Ober

Reply
Dan Hudgins September 15, 2013

Robert, you are free to email Jihua yourself. Saying the camera will not sell is really not something you can predict, I was at NAB and many people asked when they could purchase one after looking at the cameras on display and seeing the demo reel shown on a reasonably calibrated HD monitor, they also had the HD-SDI output going live to the monitor showing the same result you would get (or less because post de-Bayer can be higher quality then the live real-time de-Bayer) and most said they were impressed by the look of the results, in particular when I had the white balance set right for the lighting, and the liked the way it had clean breakout of the highlights, since its a CinemaDNG recording camera the live results and the recorded results are similar, although the workflow needs to be set-up right as well as the monitoring to see that.

The issue with the run button might be fixed by using a resistor in series and capacitor in parallel to the run switch contacts, I have told Jihua that but am not sure he mentioned it to Cheng, you can email them and ask Cheng to talk to me about that issue. It might be possible to wire a capacitor across the miscellaneous connector for the external run switch and plug that in as well, the pc traces may be thin enough to give the R/C filter effect.

Even if there are some issue, I have found the image quality and focus aids to be of high enough quality that the choice of another camera or theirs is not a hard choice, the lack of a working OLPF in the BM cameras makes their results almost useless for what kind of work I am interested in. The OLPF is not a low cost item, it might be as much as maybe $600 but its false economy for BM to leave it out of their designs, there reasoning seems obscure other than faulty cost cutting.

Acam dII ™ and maybe Digital Bolex use CCD with a larger fill factor, but none the less, a good OLPF filter should be employed since any light that can fall from an image point on just one pixel will produce ambiguous color information, there is no way around that with a CFA Bayer filter camera.

KineRAW ™ like ARRI ™ Alexa ™ and RED’s Epic ™ use a good OLPF filter as any professional quality Digital Cinemad camera should. So if you define a Digital Cinema Camera by the quality of its OLPF filter, then what are BM and Digital Bolex?

Reply
Dan Hudgins September 15, 2013

Talking about which is the “better buy” might make sense for “equal” products, but these cameras are not “equal” so there are many factors other than price that matter, most of all what the end results look like on the theatre screen. The buyer and filmmaker will have to consider the cost to the success of their project from not selecting a camera than can deliver the film like and in focus results needed for the audience’s pleasure, its not a paperweight, its a filmmaking tool you will use 12 to 14 hours a day, it will need to work with your workflow budget, it will need to work with your light kit, it will need to scream “Cinema” when people see your results, not gimmick cam cheepy muck with aliasing and moire. If people want messed up looking footage they can shoot with 8bit H.264 recording cameras that give no grading range without adding histogram gaps, a CinemaDNG camera is a speciality product for Cinematography so will not sell in similar volumes as pro-sumer cameras that don’t shoot 2048×1080 for DCP use, its not intended to, people who know what they want will know which product they need.

Reply
Elisha Rhea November 2, 2013

For $8grand and still rolling shutter? No thanks I will pass. Might as well stay with the 5dmarkIII

Reply
Johnnie Behiri November 2, 2013

Hi Elisha, ma nishma?

The camera is €3800 ($5131,14).
All in all it is a nice camera which caters a different knish then the 5dmIII.

If you have a chance to test it, please do. You might be surprised!.

Le it ra ot.

Johnnie

Leave a Comment

Connect with:

Your email address will not be published. Required fields are marked *


three + 5 =