Bob Hart
August 12th, 2008, 02:20 AM
As this is essentially a double post, I'll limit it to a link to the other post in the Letus threads.
http://www.dvinfo.net/conf/showthread.php?t=127879
http://www.dvinfo.net/conf/showthread.php?t=127879
View Full Version : Letus Extreme and SI2K Bob Hart August 12th, 2008, 02:20 AM As this is essentially a double post, I'll limit it to a link to the other post in the Letus threads. http://www.dvinfo.net/conf/showthread.php?t=127879 Bob Hart December 20th, 2008, 09:23 AM SI2K - LETUS EXTREME TEST. By Bob Hart On ExposureRoom (http://exposureroom.com/members/DARANGULAFILM.aspx/assets/e69f5c04e6054217a341e7016e0ffb59/) Jason Rodriguez December 21st, 2008, 10:40 PM Hi Bob, Nice clips you've uploaded :) BTW, the "magenta highlights" issue is because the actual RAW data, the green channel has clipped before the red and blue channels, so there is still color information in those other two chanels, but green on the other-hand has hit the full 1.0-white-point. This typically happens most often with daylight balanced scenes, since the green-channel is the most senstitive channel on the camera, and daylight actually has quite a bit of green-light in it (much more so than tungsten). The way to get around this is to add a little bit of gain to the red and blue channels in the highlights in order to get them to clip. You can also use a histogram control as well, but this often gains the entire channel rather than simply the highlights. We ship the standard color-correction .Looks that come with the camera with the compensation adjustments to get rid of the magenta highlights. But if you are color-correcting from the "RAW" source, without the .Look files, then keep these things in mind to avoid magenta clipped highlights. Thanks, Jason Bob Hart December 22nd, 2008, 01:53 AM Jason. Thanks for the tips. I am on a steep curve of learning. Steve went to the P+S course in Munich. Basically I'm blundering around trying to go the trial and error route. As David Newman says, "Welcome to the power". One should apply one'self more to formal learning discipline. To quote from Disney's Pocohontas, "you'll learn things you never knew you never knew". At the Sonic Sessions shoot, I had a crisis of confidence with the histogram display when I managed to get it to come out of its cave. I was on the spot in that I had not had time to read the good book but went straight in cold on the SI2K and had to supervise three other cameras as well, all not mine. Once we were rolling there was no time to rifle through the book searching for how-to's. One of the RED guys was shooting with the histogram on the cold side where mine was so I thought all was well with the world. - Wrrronggg. I have been doing the go outside, shoot test, come inside, look, go outside shoot some more, approach. Time and the access window to the camera is a bit short. Being able to hook out test clips on a USB thumb drive and test them in PP2/Prospect4K is a real bonus, no frigging around with captures and conversions. The turnkey workflow is the real winner with your system. Here's a few grabs and screen grabs of the correction settings I had to apply. Being red-green colourblind does not help. The vignetted frame indicates a lens view from a Peleng Fisheye which just fits the standard 24mm wide 35mm academy motion picture frame as seen by a P+S Technik Mini35-400. The Letus Extreme, with a few hacks, enables a much wider view and the sharper "apparent resolution" to be had. Not all 35mm stills lenses cover this larger view through a groundglass screen without brightness falloff. The Peleng is not sharp as there is a glass panel in the Letus Extreme behind the lens which affects the backfocus. (The Peleng lens is very sensitive to this and has in the instructions not to remove a clear glass filter from the back of the lens itself lest backfocus be adversely affected. Removing the filter would probably restore backfocus on groundglass adaptors.) The groundglass motor was not on. You will observe the groundglass texture in the image. You will observe a little bit of blue CA in the right margin. I had adjusted the centering a bit too far left and began to pick up an edge. You will also observe from light levels across the Lemac chart there remains a edge falloff or centre hot spot. This can be seen with a white card in front. I am demanding of the Extreme far more than Quyen and Hien Le, originally intended it to yield. The visible groundglass is normally restricted to a considerably smaller area With 1/3" cameras on my own AGUS35, I foud that with the smaller prism I used, a longer setback and less powerful dioptre on front of the relay lens actually gave me a cleaner path, the opposite to what I expected. I imagine the same may apply here. The 35mm and 40mm stills lenses I used as relays for the test are going through the 5+ approx dioptre for Sony EX1. A less powerful dioptre, say ballpark of 3+, a 50mm or 58mm stills lens, and longer setback may yield a better result. In the earlier days of home-building of 35mm adaptors, there were a few here who described a method of applying in post, a layered effect to correct edge and corner brightness fall-offs. It required making a calibration image for each lens used and hoping the adaptor did not bend or go off-centre in the meantime. Now all we need is for David and his crew to invent some sort of software lens mapping facility for the SI2K like some broadcast cameras apparently have. Bob Hart December 22nd, 2008, 09:27 PM A very short clip which is vision from which the frame grabs came can be found here :- FURTHUR TEST OF SI2K - LETUS EXTREME AFTER RE-CENTERING By Bob Hart On ExposureRoom (http://exposureroom.com/members/DARANGULAFILM.aspx/assets/519a11dfd638414ea8703892218d491a/) Bob Hart December 23rd, 2008, 07:17 AM Here are some grabs from furthur tests, this time with a little more finesse in getting light levels right. The Voigtlander f2 40mm was retested and the Nikon f2.8 45mm, a lens I had thought would be a non-contender. So little one knows. The 45mm appears to be the best of them and possibly sharper. The field-of-view is narrower and a little closer to the normal field-of-view of the unmodified Letus Extreme. The edge falloff is gone. Unfortunately, this lens, although it is a F mount will not fit the new positive lock P+S Technik Nikon F-mount because of a narrow collar on back of the lens. I tested it by crudely holding it in position by hand. Genuine Nikon mounts for this lens have a shoulder cut in the outer 1mm of the flangeface to clear the collar. The details of this individual lens are :- GN Auto NIKKOR 1:2.8 f=45mm Nippon Kogaku Japan No. 718777. It is a specialty lens for flash photography and not all that common. This one is an older version. Newer versions were made by Nikon. There are a few turn up on eBay. I guess there will be a run on them now. To use this lens for relay of the Letus Extreme to the SI2K, the newer style P+S positive locking version of the Nikon mount may have to be modified by having a clearance turned from the outer diameter of the P+S Technik Nikon Mount flange face to accommodate the collar. Newer versions of the same lens may be missing this collar and may fit up okay. To mount this lens to the Letus Extreme, a 52mm lens filter thread diameter to rear of Letus Extreme body adaptor ring is needed. This may be a part already made by Letus, otherwise it will have to be custom made. I can only vouch for the dimensions of my own adaptor ring and cannot assure that another ring adapted by using stepdown rings will work. I have also not cranked the contrast up on the images relayed from the 45mm lens. There may still be unacceptable brightness falloff in there. I have also not tested this lens against the Lemac chart for its resolution. There may be a reduction in apparant sharpness due to the smaller area of the groundglass aquired by the sensor through the 45mm relay lens. This is because the relative size of each "grain" of the groundglass texture is now scaled larger. This lens is also f2.8 wide-open so low-light performance may not be all that you might want it to be. The posted images here were all set up with a spot level reading of 77 from the limestone wall. The iris of the prime on front was in the ballpark of f2 - f2.8. The relay was set to f4 - f5.6. Frames are numbered :- (1) (2) (3) (4) (1) is f2 40mm Voigtlander-for-Nikon relay, f1.4 35mm Nikon prime on front of Letus Extreme. (2) is f2 40mm Voigtlander-for_Nikon relay, f1.4 85mm Nikon prime on front of Letus Extreme. (3) is f2.8 45mm Nikon relay, f1.4 35mm Nikon prime on front of Letus Extreme. (4) is f2.8 45mm Nikon relay, f1.4 85mm Nikon prime on front of Letus Extreme. Before anyone gets all excited about putting the Extreme on front of 2/3" 3xCCD cameras with this lens, be aware this will likely not work due to the added optics between the lens mount on the camera and the sensors. There will liekly be a CA problem. The SI2K is more forgiving because of its single CMOS sensor. That will be my lot for experimenting for a while. Bob Hart December 23rd, 2008, 07:46 AM Here is an image of the adaptor ring, in this instance fitted up to a CP16R film camera. This adaptor was a roughie. The piece of flowcast I had was a bit too small. The wider diameter fitment to the shoulder on back of the Extreme should be deeper and end almost against the rear of the flip enclosure, not set back like this one. The screws should also be furthur forward. Jason Rodriguez December 24th, 2008, 03:30 AM Hi Bob, By using the "Linear" profile, you basically guarantee exposing the color-channel clipping issue that creates the "magenta" highlights. It basically again comes from a combination of color-channel clipping in the green channel as well as the matrix itself adding gains to the color channels and linearly saturating the image, which then causes any slight color shifts in the highlights to become more saturated, and there-by more evident. The linear-light LUT itself further compounds the issue because you are removing any nice natural "knee" natural roll-off into the highlights, and instead allowing the color-channel clipping to consume most of the visually displayable dynamic range of the monitor viewing system. "Knee" systems provide a type of visually pleasing highlight compression that avoids this issue, placing visual emphasis on the areas of the image below the clip point that are evenly balanced. For instance, with a linear signal, there is only 1 f-stop of overhead from the perceptually linear middle-grey point to the white highlight clip when viewed on a monitoring system. With a 12-bit image that means 11-bits are being assigned to everything below middle grey, and there is only 1-bit (the most signficant upper bit which takes up 2048 values in a 12-bit image) assigned to everything above middle grey. But basically by doing that you are placing extreme emphasis on the upper bit of the image, and that area is the "last f-stop" highlight exposure area where these types of channel clipping issues occur, and you are mapping that area to fill the entire range above middle grey on a normal monitor viewing system. If there is any color imbalance in the extreme highlights, it will become very evident very quickly (and is what you see with that magenta color caste). The first part of the solution is to expose and work in a non-linear gamma-corrected color-space like REC709, etc. That way you at least have some highlight headroom to work with, and get some highlight compression to avoid the magenta highlights being a very prominent portion of the image. You can get REC709 using our "default.look" 3D LUT that ships with the camera. The second part of the solution is not to turn off the matrix, but rather you have to gain up the blue and red channels in order to get them to clip with the green channel. This though can change the color-balance of the image, so another method is to set a new white-point that has all the color-channels clipped rather than just one or two color-channels. You can do either operation easily enough with a histogram, or if you want to use the metadata controls from CineForm, increase the "gain" setting in the matrix, which will uniformly increase all the color-channels so that you also clip the red and blue channels at the same time the green-channel is clipped. Naturally this clipping does reduce a bit of the head-room in the image, but this is headroom that you would need to clip anyways if you want to avoid those color-shifts on a global scale. Otherwise you can use secondaries to get rid of the magenta shift in the highlights (by desaturating the highlights), or you can use some custom contrast/gamma curves that compress the highlights and provide a nice smooth transition into those clipping areas, thereby making the highlights turn white again since they are smoothly clipping the red, green, and blue channels all at the same time. This for instance is what we've done with our .Look files that we've provided as presets in the camera so that you don't have to deal with the magenta highlights issue. Thanks, Jason Bob Hart December 24th, 2008, 04:13 AM Jason. Thanks again for your response and advice. Since I got the camera from Steve, I have been roaming around in its various offerings. Linear just happend to be already there so I thought it was some sort of default setting. I have to do other things over Christmas and a day's shoot on the 27th so it will be a while before I get time to play with it again. Outdoors in the only near-pure light I could get, I created my own preset of sorts against a Dumonde chart while peering through a cardboard hood on the screen. Hardly a professional method but it enabled me to use the screen for a rough WYSIWYG impression in the same outdoors operating conditions. There is bound to be a way to set up the touch screen display like I do with the Sony Z1 to the internal colour bars with pluge bar in the actual shooting conditions and outdoors glare etc., falling on the screen. It is a false image but helps keep me intuitively in the correct manual exposure range when things are in a hurry. I also used the spot light metering function, again guessing at a threshold of sorts until I can read up more on it. I called my preset the "honest" look, good for the display itself outdoors but quite ugly back on the computer. Think custard yellow highlights and you get the picture. Back on the computer, I applied the film print look which I downloaded from SI and that is the look applied to the most recent four grabs on this thread made under that regime. No other adjustments were made. I am getting the impression one needs to regard this camera as a "film" camera in the way it is managed. Another aspect is that I have been running the camera in daylight condition only with the Letus Extreme on front, not without. The only time I have run it pure was at the "Sonic Sessions" night shoot. Regards to all for Christmas and the New Year. Bob Hart December 24th, 2008, 06:39 AM SETUP DIMENSIONS FOR SI2K ON LETUS EXTREME CINE BUNDLE - FIRST VERSION. READERS ARE FORWARNED THAT THE PROCESS DESCRIBED BELOW IS NOT ENDORSED BY EITHER THE COMPANY WHICH MAKES THE SILICON IMAGING SI2K CAMERA SYSTEM OR THE COMPANY WHICH BUILDS THE LETUS EXTREME GROUNDGLASS BASED IMAGE RELAY ADAPTOR. FURTHURMORE ATTEMPTS TO JOIN AND OPERATE THESE APPLIANCES MAY DAMAGE THEM AND RENDER WARRANTIES OR GUARANTEES VOID. OWNERS OR OPERATORS WHO ATTEMPT TO REPLICATE MY EXPERIMENTS DO SO AT THEIR OWN RISK. THIS INSTRUCTION SET ASSUMES YOU INTEND TO USE THE NIKON F2.8 45mm LENS. IT IS ALSO ASSUMED YOU HAVE AQUIRED OR MADE AN ADAPTOR RING TO MOUNT THE 45mm NIKON LENS TO THE REAR OF THE LETUS EXTREME BY THE 52mm DIAMETER 0.7mm PITCH FILTER THREAD. IT IS ALSO ASSUMED THAT THE LETUS EXTREME YOU ARE USING HAS BEEN MODIFIED FOR CORRECT CENTRICITY. FORTUNATELY, COMPARED TO THE SONY PMW-EX1, THE UNMODIFIED EXTREME'S IMAGE THROW, WHICH IS BIASED TO THE RIGHT IN SOME INSTANCES, MAY FALL CLOSE TO CENTRALLY ON THE SI2K SENSOR WITHOUT NEED OF MODIFICATION TO THE LETUS EXTREME. IT IS ALSO ASSUMED THAT THE CONDENSOR LENS ORIFICE ON THE FRONT WALL OF THE FLIP ENCLOSURE HAS BEEN WIDENED BY ABOUT 1mm EITHER SIDE. HOWEVER, THE ORIFICE MAY BE ALREADY WIDE ENOUGH FOR THE 45mm LENS. IF AN EDGE COMES INTO THE IMAGE, IT MAY BE NECESSARY TO PERFORM THE CENTRICITY MODIFICATION OR WIDENING OF THE CONDENSER ORIFICE OR BOTH. THESE DIMENSIONS WILL BE OFF IF THE LETUS EXTREME'S CINE BUNDLE'S SLIDING WEDGEPLATE IS NOT STRAIGHT ON THE BOTTOM OF THE SI2K. IN ABSENCE OF THE NIKON F2.8 45mm LENS, THE VOIGTLANDER-FOR-NIKON F2 40mm IS A LESSER SUBSTITUTE. THE EDGES OF THE PRISM PATH WILL BEGIN TO CAUSE BRIDGHTNESS FALLOFF AND SOME STILLS LENSES USED ON FRONT OF THE LETUS MAY ALSO THROW A VIGNETTED IMAGE. IT MAY BE HELPFUL TO LEAVE THE FASTENING SCREW SNUG BUT NOT TIGHT SO THAT THE WEDGEPLATE CAN TURN A LITTLE ON THE BASE AND WILL COMPLY AS YOU MOUNT THE CAMERA TO THE LOWER WEDGEPLATE SUPPORT AND CLAMP IT TIGHT. AFTER EVERYTHING IS ALL LINED UP AND SET, MAKE AN OUTLINE MARK ON THE BASE WITH A SHARPIE, REMOVE THE CAMERA AND THEN TIGHTEN THE WEDGEPLATE FIRMLY. *********************************************************************** DISTANCE FROM LOWER SURFACE OF REAR BRIDGE PIECE DOWNWARDS TO TOE OF REAR SUPPORT PILLARS - 44mm. *********************************************************************** NOTE: YOU CANNOT GO LOWER WITH THE PILLARS BECAUSE THE WING OF THE CLAMPING SCREW ON THE RIGHT SIDE WILL FOUL AGAINST THE UNDERSIDE OF THE SI2K CAMERA BODY. *********************************************************************** DISTANCE FROM LOWER SURFACE FRONT-OF-LETUS EXTREME BRIDGE PIECE UPWARDS IN VERTICAL CHANNEL TO TOE OF SUPPORT PILLAR - 9mm. *********************************************************************** NOTE: THE NIKON F2.8 45mm LENS WHICH IS ONLY ONLY ONE I HAVE SO FAR PROVEN AS ADEQUATE TO RELAY THE EXTREME TO THE SI2K, IS NOT AN INTERNALLY FOCUSSING LENS. IT GROWS LONGER OR SHORTER WITH FOCUS MOVEMENTS. THEREFORE, THE FRONT LOWER BRIDGEPIECE CLAMPING SCREW WHICH LOCKS THE BRIDGEPIECE TO THE RODS MUST BEFT BACKED OFF SO THAT THE BRIDGEPIECE IS ALLOWED TO SLIDE ON THE RODS BECAUSE THE LENS FOCUS MOVES THE ENTIRE LETUS BODY. YOU WILL NEED TO ASSIST THE BRIDGEPIECE TO SLIDE FORWARD OR REARWARDS WITH THE FOCUS MOVEMENTS AS THE LENS ITSELF WILL JAM. DO NOT FORCE THE LENS TO MOVE. ************************************************************************ HORIZONTAL OFFSET OF LEFT EDGE OF WEDGEPLATE PLATFORM (BUTTON SIDE ON LEFT) TO LEFT EDGE OF REAR SUPPORT PILLARS - 21.5mm. ************************************************************************ THAT SHOULD PUT YOU IN THE BALLPARK FOR YOU TO BEGIN FINER ADJUSTMENTS. THAT FINER ADJUSTMENT WILL BE TO CHECK THAT THE FRONT SUPPORT PILLAR BRIDGEPIECE IS LOADED ONLY WITH THE WEIGHT OF THE LETUS EXTREME'S BODY. NO ADDITIONAL LOADING SHOULD BE IMPOSED BY BY PRESSURE EXERTED THOUGH THE NIKON LENS BODY BY MISALIGNMENT. ANY ATTEMPT TO ADJUST RELAY FOCUS IN THIS CONDITION WILL DAMAGE THE RELAY LENS. THE LETUS BODY SHOULD BE LIGHTLY SUPPORTED BY THE BRIDGEPIECE ON THE RODS SO THAT THE LENS FOCUS MOVEMENT WILL SLIDE THE BRIDGEPIECE. YOU MAY NEED TO LOOSEN THE SINGLE FRONT SUPPORT PILLAR SCREW AND SUPPORT THE WEIGHT OF THE LETUS BODY JUST A LITTLE AND REFASTEN THE SCREW. WITH A LITTLE TRIAL AND ERROR YOU WILL GET THERE BUT IT IS FAR FROM A SATISFACTORY ARRANGEMENT. TO SET RELAY FOCUS, YOU NEED TO USE THE "LOUPE" FUNCTION OR THE ZOOM FUNCTION OF THE SI2K AND USE FOR A TARGET THE RESOLUTION BLOCKS OF A FOCUS CHART OR A HAIR COMB STICKYTAPED TO A PIECE OF WHITE PAPER AS A SUBSTITUTE. Bob Hart December 28th, 2008, 11:02 AM CORRECTION: The header above should read f2.8, not f2.9. This lens is also known as a "pancake lens" because of its shallow shape. This is the one which seemed to achieve best relay image quality to the SI2K from the Letus Extreme. I can't vouch for how well it will work on unmodified Letus Extremes. There is an attached image of this lens. There were a few variations over the years. Apparently the basic design principle of this lens has seen little need for variation since its invention about 100 years ago, except for improved coatings and finish. The photographic vendors may be visiting dvinfo.net for there is now a larger number of these being offered on that famous unmentionable auction website where there were fewer a few days back. ( Then again, I may have serious delusions as to the significance or lack thereof relating to my posts.) |