steppenw0lf Posted April 22, 2016 Share #1 Posted April 22, 2016 (edited) Advertisement (gone after registration) Since the discussions about FW 2.0 I heard that there are two possible shutters: Is it electronic and mechanic or electro-mechanic ? Can anybody try to explain which shutters the SL offers, the difference between them and finally why the electronic shutter only appears now (and when exactly is it used by the camera). If the manual tells all this I won't mind a link to it. (And close the thread asap.) Stephan Edited April 22, 2016 by steppenw0lf Quote Link to post Share on other sites More sharing options...
Advertisement Posted April 22, 2016 Posted April 22, 2016 Hi steppenw0lf, Take a look here Electronic Shutter vs. "normal" Shutter. I'm sure you'll find what you were looking for!
LocalHero1953 Posted April 22, 2016 Share #2 Posted April 22, 2016 I hope others will offer a more technical explanation, but to start it off.... The sensor needs to measure the amount of light collected by each sensel in a given interval. A mechanical shutter (electronically driven) does this by physically closing, emptying the sensels, opening to collect photons, then closing again to allow the photons to be counted. In my understanding, a purely electronic shutter is not really a shutter, but just a means of measuring the photons falling on the sensels in an interval without physically blocking the light to start with or shutting it off afterwards. The SL now uses both mechanisms, but the electronic shutter, for reasons I don't understand, is only available at very fast speeds. I can't answer the other parts of your question. 3 Quote Link to post Share on other sites More sharing options...
scott kirkpatrick Posted April 22, 2016 Share #3 Posted April 22, 2016 I think the electronic shutter simply gates the pixels on and then off a very short time later. Since it takes a significant amount of time to expose and then read all of the pixels out (something like the 1/60 sec refresh rate that LiveView presents. The 1/18000 of a second while equal in duration at all pixels is not happening at the same time on all pixels. so the exposure might be vulnerable to artifacts that result from the subject moving within the time that it takes for the image all to be collected. This is just a guess on my part, based on hints picked up in discussions of the electronic shutter issues on other brands of camera. Has anyone taken a picture at 1/18000 second of a fast moving car, to see if the image slides across from top to bottom, as in early shots with a slow moving focal plane shutter (Lartigue's is best known). scott 1 Quote Link to post Share on other sites More sharing options...
steppenw0lf Posted April 22, 2016 Author Share #4 Posted April 22, 2016 (edited) As I have been working a long time in unix support and know the Power chips in depth I would like to make a comparison with the IBM Power8 and the popular Intel Haswell chips. (comparing them to an unknown/anonymous sensor chip in the SL). The Power8 has a memory bandwidth of 384 GB/s The Haswell of 68 GB/s For a transfer of 72 MB (24 MB times 3) they take around: (let's assume 100 MB per image to make it simpler) Power8 1/3840 s = 0.00026 s Haswell 1/680 s = 0.00147 s So they would offer a faster refresh rate (than 1/60 s), but they are using much more energy than a camera can provide. (Just some sort of "back of the envelope" plausibility check.) GPUs are also in this range (beginning at 14.4 GB/s up to 336 GB/s), but even bigger in size and energy consumption. Any hints where these discussions about sensor chips can be found ? A special forum ? Thanks. Stephan Edited April 22, 2016 by steppenw0lf 1 Quote Link to post Share on other sites More sharing options...
FlashGordonPhotography Posted April 22, 2016 Share #5 Posted April 22, 2016 The SL has, like most CMOS sensors that use it, a rolling electronic shutter. When used (only between 1/8000 and 1/16000 in the case of the SL) the camera reads one line of pixels at a time in sequence from the bottom of the sensor to the top (which is seen in reverse because the image is inverted). These then *build* an exposure. It's like an electronic version of a swing shutter like those found in Noblex panoramic cameras. Each line reads for the correct amount of time but the overall shutter time is the time it takes for the camera to read and write each line and turn that into a single file. You could potentially see rolling shutter effects but these are most visible as the shutter speed (read speed) lowers so in the case of the SL it's unlikely you'll ever "see" a rolling shutter effect unless Leica decide to make it available at lower shutter speeds as well. Gordon 2 Quote Link to post Share on other sites More sharing options...
scott kirkpatrick Posted April 23, 2016 Share #6 Posted April 23, 2016 (edited) But if the electronic shutter exposes one line at a time, even if we count only the exposure time, overlapping readout and control delays, 4000 lines at 1/8000 sec per line requires 1/2 sec! I suppose the lines are exposed in moderate-sized groups. A live view refresh rate of 60 to 120 frames per sec (as needed for good autofocus response) would require groups of 30 to 60 lines. These same issues come up with cameras like the latest Olympus M5 and Hasselblad, which shoot multiple images to be combined in a single frame with sub-pixel resolution, moving the chip after each image is taken. It is interesting to see the gaps between still imaging, video, light field and full-blown computational imaging fill in. scott Edited April 23, 2016 by scott kirkpatrick 1 Quote Link to post Share on other sites More sharing options...
steppenw0lf Posted April 23, 2016 Author Share #7 Posted April 23, 2016 Advertisement (gone after registration) In order to detect the rolling shutter effect, I could imagine that it is a good idea to go to a small airfield with older propeller planes. The effect should be visible with rotating propeller blades. (easier than following a F1 car, and probably also easier to find on the countryside) Stephan Quote Link to post Share on other sites More sharing options...
Jared Posted April 23, 2016 Share #8 Posted April 23, 2016 As I have been working a long time in unix support and know the Power chips in depth I would like to make a comparison with the IBM Power8 and the popular Intel Haswell chips. (comparing them to an unknown/anonymous sensor chip in the SL). The Power8 has a memory bandwidth of 384 GB/s The Haswell of 68 GB/s For a transfer of 72 MB (24 MB times 3) they take around: (let's assume 100 MB per image to make it simpler) Power8 1/3840 s = 0.00026 s Haswell 1/680 s = 0.00147 s So they would offer a faster refresh rate (than 1/60 s), but they are using much more energy than a camera can provide. (Just some sort of "back of the envelope" plausibility check.) GPUs are also in this range (beginning at 14.4 GB/s up to 336 GB/s), but even bigger in size and energy consumption. Any hints where these discussions about sensor chips can be found ? A special forum ? Thanks. Stephan Both CCD's and CMOS sensors have much less throughput, but it's not necessarily because of limitations in processing speed. Generally, the faster you read out a sensor the higher the read noise. A compromise needs to be reached, and that is usually around the 1/60s that allows good Live View performance. Some astronomy cameras in order to minimize read nose take up to 30s to read out the contents of the sensor. As to the original question... A m chain all shutter has the advantage of no rolling shutter since the exposure takes place during the same interval for all sensels. In the cas of an electronic shutter, the exposure takes place at different times for each row, so you can get what are called "rolling shutter" effects where vertical lines appear to slant on moving objects (or on stationary objects if you are panning). Leica uses a mechanical shutter whenever it can--up to a shutter speed of 1/4000s--and an electronic shutter for faster speeds where the mechanical just can't keep up. - Jared 1 Quote Link to post Share on other sites More sharing options...
pico Posted April 23, 2016 Share #9 Posted April 23, 2016 In order to detect the rolling shutter effect, I could imagine that it is a good idea to go to a small airfield with older propeller planes. The effect should be visible with rotating propeller blades. (easier than following a F1 car, and probably also easier to find on the countryside) Stephan Tun a bicycle upside-down and spin the rear wheel in top gear. Quote Link to post Share on other sites More sharing options...
jrp Posted April 23, 2016 Share #10 Posted April 23, 2016 http://blog.kasson.com/?p=13858 has a number of instrumented tests of which this is the most recent. What at is slightly odd is the EFCS causes problems with fast shutter speeds on the Sonys. Quote Link to post Share on other sites More sharing options...
steppenw0lf Posted April 23, 2016 Author Share #11 Posted April 23, 2016 (edited) Maybe a useful link for general understanding: http://www.cse.wustl.edu/~jain/cse567-11/ftp/imgsens/index.html Especially chapter two: "Performance Metrics for Image Sensors". Stephan Or this (proprietary but nice graphics about reading out data): Mainly about Exmoor sensor https://pro.sony.com/bbsccms/assets/files/cat/camsec/solutions/E_CMOS_Sensor_WP_110427.pdf Edited April 23, 2016 by steppenw0lf Quote Link to post Share on other sites More sharing options...
FlashGordonPhotography Posted April 23, 2016 Share #12 Posted April 23, 2016 But if the electronic shutter exposes one line at a time, even if we count only the exposure time, overlapping readout and control delays, 4000 lines at 1/8000 sec per line requires 1/2 sec! I suppose the lines are exposed in moderate-sized groups. A live view refresh rate of 60 to 120 frames per sec (as needed for good autofocus response) would require groups of 30 to 60 lines. These same issues come up with cameras like the latest Olympus M5 and Hasselblad, which shoot multiple images to be combined in a single frame with sub-pixel resolution, moving the chip after each image is taken. It is interesting to see the gaps between still imaging, video, light field and full-blown computational imaging fill in. scott It's more that the read of line 2 begins before the read of line 1 finishes. It's still one line starting at a time. How fast that is, I suppose that would depend on processor speed and circuitry efficiencies. It'd be something like: L1 on> L2 on> L3 on> L4 on> L1 off> L5 on> L2 off> L6 on, etc. Gordon Quote Link to post Share on other sites More sharing options...
steppenw0lf Posted April 23, 2016 Author Share #13 Posted April 23, 2016 This is also proprietary, but gives some insight into shutter operations with image sensors. http://www.onsemi.com/pub_link/Collateral/AND9195-D.PDF Stephan Quote Link to post Share on other sites More sharing options...
mjk Posted April 25, 2016 Share #14 Posted April 25, 2016 I tried the 1/16,000 sec speed with electronic shutter after upgrading to FW 2.0. See the fan in the centre of the picture. It was running at full speed. The shutter managed to freeze two out of three blades. It seems as if it somewhat shortened the blade in the bottom left hand corner, and lengthened the one in the bottom right hand corner. The fan was rotating clockwise. ISO 50,000, f/3.2. Welcome, dear visitor! As registered member you'd see an image here… Simply register for free here – We are always happy to welcome new members! 1 Quote Link to post Share on other sites Simply register for free here – We are always happy to welcome new members! ' data-webShareUrl='https://www.l-camera-forum.com/topic/259592-electronic-shutter-vs-normal-shutter/?do=findComment&comment=3033225'>More sharing options...
FlashGordonPhotography Posted April 25, 2016 Share #15 Posted April 25, 2016 Assuming the banding is from the TV or light source that image quality is rather surprising at ISO 50,000. And here's me still dubious about ISO 1600. Gordon Quote Link to post Share on other sites More sharing options...
scott kirkpatrick Posted April 26, 2016 Share #16 Posted April 26, 2016 (edited) It's more that the read of line 2 begins before the read of line 1 finishes. It's still one line starting at a time. How fast that is, I suppose that would depend on processor speed and circuitry efficiencies. It'd be something like: L1 on> L2 on> L3 on> L4 on> L1 off> L5 on> L2 off> L6 on, etc. Gordon I agree. Staggering the readouts rather than clumping them would give smoother results. The bandwidth required for signalling and coordinating all of this could be minimized by putting the coordinating logic mostly on the imaging chip. But the tough part of the job remains buffering and reordering the output lines into a 4000x6000 image that the rest of the camera can handle, and exporting this image 60 to 120 times a second. That's 14.4 to 28.8 GB/sec, or as much as 150 to 300 Gbits per second, with error correction, which is pretty much state of the art for medium power components (not for your average smart phone). scott I missed several notes above. See Jared's in particular, for the tradeoff between read speed and read noise. Edited April 26, 2016 by scott kirkpatrick Quote Link to post Share on other sites More sharing options...
LocalHero1953 Posted April 26, 2016 Share #17 Posted April 26, 2016 I may be misunderstanding you, Scott, but that is what your average smartphone actually does. It only has an electronic shutter, and (checking one typical shot I took recently) took an 8Mp shot at 1/400s. I used my Olympus OMD EM5ii to take 16Mp shots at 1/60s. The puzzle is not how advanced the technology is (though I'm interested in the technology as a geek) but why Leica has only implemented it at a very fast speed. Quote Link to post Share on other sites More sharing options...
FlashGordonPhotography Posted April 26, 2016 Share #18 Posted April 26, 2016 Because either: 1. They're concerned about rolling shutter that really shows more at lower speeds. Shouldn't that be a user decision? 2. They didn't realise how useful a silent shutter can be for many photographers. Like weddings, movie sets etc. etc. 3. As usual Leica don't quite get it right first time around. Gordon Quote Link to post Share on other sites More sharing options...
steppenw0lf Posted April 26, 2016 Author Share #19 Posted April 26, 2016 (edited) 1. They're concerned about rolling shutter that really shows more at lower speeds. Shouldn't that be a user decision? Gordon I am just guessing, but unless the sensor chip is specifically designed for global shutter operations, a "electro-mechanical" shutter provides better quality (no rolling shutter). At very high speeds the end is reached for the electro-mechanical shutter (or a disproportionate effort is necessary to improve it), while the rolling shutter problem effectively disappears. Unfortunately we still know almost nothing about the sensor chip (producer, design, specs etc.) Stephan Edited April 26, 2016 by steppenw0lf Quote Link to post Share on other sites More sharing options...
airfang Posted February 22, 2017 Share #20 Posted February 22, 2017 To follow up on this topic, do people leave electronic shutter on or off? Quote Link to post Share on other sites More sharing options...
Recommended Posts
Join the conversation
You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.