tag:blogger.com,1999:blog-19092890.comments2024-03-19T01:49:59.848+02:00Image Sensors WorldVladimir Koifmanhttp://www.blogger.com/profile/01800020176563544699noreply@blogger.comBlogger21491125tag:blogger.com,1999:blog-19092890.post-36429432938440773732024-03-18T20:23:01.650+02:002024-03-18T20:23:01.650+02:00They are understandably coy about saying anything ...They are understandably coy about saying anything about the pixel size or pixel count. Based on the vendor product page, this was likely a 4.86um, 1280x720 imager ( https://www.prophesee.ai/event-based-sensors/ ), which is hardly a winning proposition for mobile photography. Prophesee should be respected for making the effort, getting this architecture to 1280x720 is a meaningful improvement that may open new doors. It is an interesting technology, which has been looking for a killer use case for decades. Mobile photography probably isn't it, and betting too much on it may get them to the same place where other elegant and promising technologies ended up. Anonymousnoreply@blogger.comtag:blogger.com,1999:blog-19092890.post-76967516175872797192024-03-16T12:15:16.480+02:002024-03-16T12:15:16.480+02:00wow, another Philips spinoff. here Photonis, but t...wow, another Philips spinoff. here Photonis, but there are big ones like NXP, ASML, ASM and many more, see https://innovationorigins.com/en/laio/how-philips-spin-offs-outshone-the-mother-company/Anonymousnoreply@blogger.comtag:blogger.com,1999:blog-19092890.post-27689530850382985542024-03-16T04:17:31.441+02:002024-03-16T04:17:31.441+02:00The ADNV G14P is better than this one. The ADNV G14P is better than this one. Anonymousnoreply@blogger.comtag:blogger.com,1999:blog-19092890.post-66987398096936644222024-03-15T23:17:20.728+02:002024-03-15T23:17:20.728+02:00Gen0 used in such as Bushnell comes from Ukrain, s...Gen0 used in such as Bushnell comes from Ukrain, so no longer available now.Anonymousnoreply@blogger.comtag:blogger.com,1999:blog-19092890.post-33649754289253179542024-03-15T20:41:49.937+02:002024-03-15T20:41:49.937+02:00Definitely recommend a lint roller if you still ha...Definitely recommend a lint roller if you still have Gen 0's lying around the house.Aaron Hendricksonhttps://www.blogger.com/profile/04558774822320792376noreply@blogger.comtag:blogger.com,1999:blog-19092890.post-61180421421751332332024-03-15T17:55:10.758+02:002024-03-15T17:55:10.758+02:00It will be interesting to see whether the concept ...It will be interesting to see whether the concept of event based deblurring gets traction. Any OEM will need to benchmark this against restoring images and videos using emerging generative ai techniques, which are becoming practical on mobile SoCs.Anonymousnoreply@blogger.comtag:blogger.com,1999:blog-19092890.post-55536945879790075302024-03-15T10:33:01.628+02:002024-03-15T10:33:01.628+02:00Check out as well Boyd Fowler's paper presente...Check out as well Boyd Fowler's paper presented at IISW2021 (paper R08)(www.imagesensors.org)Anonymousnoreply@blogger.comtag:blogger.com,1999:blog-19092890.post-6232187980168025402024-03-15T10:19:20.849+02:002024-03-15T10:19:20.849+02:00Pixel of the event camera and Smartphone's sen...Pixel of the event camera and Smartphone's sensor are different. The event camera continuosly monitor variations of pixels above or below a predefined analog threshold. This trigger a new event with known polarity that are "binary" information. In order to do this few photons are needed and this is why the few microseconds are enough for event detection and event generation event. In the meantime the traditional sensor integrates photons for a longer exposure time providing good quality images but affected by blur. Event information captured all the motion details (tracking of edges) and are used to recover the blur from images coming from the traditional sensor.Anonymousnoreply@blogger.comtag:blogger.com,1999:blog-19092890.post-86621606776575114602024-03-15T08:27:36.098+02:002024-03-15T08:27:36.098+02:00usefuless stuffsusefuless stuffsAnonymousnoreply@blogger.comtag:blogger.com,1999:blog-19092890.post-79092548585510633252024-03-14T22:25:42.147+02:002024-03-14T22:25:42.147+02:00Thank you so much Eric !!Thank you so much Eric !!Anonymousnoreply@blogger.comtag:blogger.com,1999:blog-19092890.post-15570557095430126802024-03-14T15:13:39.533+02:002024-03-14T15:13:39.533+02:00Alpsentek has accounced something similar? Alpsentek has accounced something similar? Anonymousnoreply@blogger.comtag:blogger.com,1999:blog-19092890.post-58773553062852295322024-03-14T14:19:56.446+02:002024-03-14T14:19:56.446+02:00The authors mention that with a stacked architectu...The authors mention that with a stacked architecture, multiple sample non-destructive readout can take place in a pixel-parallel way, thus reducing readout time and perhaps widening the application space of this non-destructive readout approach beyond long integration time and long readout time.Eric R Fossumhttps://www.blogger.com/profile/09413387887974939684noreply@blogger.comtag:blogger.com,1999:blog-19092890.post-8415715697979280342024-03-14T14:16:24.979+02:002024-03-14T14:16:24.979+02:00"Skipper" CCD was a name given to a CCD ..."Skipper" CCD was a name given to a CCD with non-destructive readout capability that was demonstrated by Janesick et al at NASA JPL for reducing read noise circa 1990. The basic idea is the the charge is sensed multiple times by a floating gate and the multiple reads are averaged together to improve SNR by averaging read noise. The noise decreases like the square root of N where N is the number of reads. Ultimately the noise then becomes limited by either dark current buildup or the SF 1/f noise. The drawback is that often N>1000 and this slows the readout rate. In fact, Janesick et al proposed that some uninteresting pixels or regions could be skipped in the multiple readout process shortening readout time. I think this is where the name "skipper" came from. Skipper CCDS have been used more recently for use in physics experiments such as the search for dark matter. Integration times are very long so the slow readout is not a big issue, relatively speaking and deep sub-electron readout noise has been demonstrated. The devices are operated at cold temperatures to reduce dark current. The reduction of read noise using non-destructive CCD readout goes back to the early days of CCDs. Try google scholar search for "skipper CCD"Eric R Fossumhttps://www.blogger.com/profile/09413387887974939684noreply@blogger.comtag:blogger.com,1999:blog-19092890.post-33021464862260385962024-03-14T12:23:47.505+02:002024-03-14T12:23:47.505+02:00what is the light level needed for "... with ...what is the light level needed for "... with microsecond events to algorithmically extract ..." ? If the light level is so strong, why the image from main camera is blurred? Anonymousnoreply@blogger.comtag:blogger.com,1999:blog-19092890.post-67470024924768289432024-03-13T18:18:18.604+02:002024-03-13T18:18:18.604+02:00Interesting what's costInteresting what's costAnonymousnoreply@blogger.comtag:blogger.com,1999:blog-19092890.post-71760340952448587812024-03-12T23:33:16.581+02:002024-03-12T23:33:16.581+02:00Running a one-sample PCH-EM characterization on th...Running a one-sample PCH-EM characterization on the data in fig. 10 (bottom) yields a read noise of 0.185e-, which is in line with Eric's estimate of 0.2e- and the comments about peak overlap. A more accurate number can be obtained with the raw sensor data (I simply digitized the plot).Aaron Hendricksonhttps://www.blogger.com/profile/04558774822320792376noreply@blogger.comtag:blogger.com,1999:blog-19092890.post-17438838968950914922024-03-12T10:41:46.229+02:002024-03-12T10:41:46.229+02:00What are the major improvements in this realizatio...What are the major improvements in this realization please, Eric? Thanks!Anonymousnoreply@blogger.comtag:blogger.com,1999:blog-19092890.post-73598352135326593102024-03-12T07:54:38.682+02:002024-03-12T07:54:38.682+02:00The full well and the CG is not mentioned. The CTE...The full well and the CG is not mentioned. The CTE is 99.9999% or even higher?Anonymousnoreply@blogger.comtag:blogger.com,1999:blog-19092890.post-17775806188005986182024-03-12T00:21:23.511+02:002024-03-12T00:21:23.511+02:00what is the recent skipper-CCD please ? What are t...what is the recent skipper-CCD please ? What are the improvements compared to ? Thanks!Anonymousnoreply@blogger.comtag:blogger.com,1999:blog-19092890.post-89389296022914313572024-03-11T16:13:46.497+02:002024-03-11T16:13:46.497+02:00I think this is a fine non-peer reviewed paper/rep...I think this is a fine non-peer reviewed paper/report and the approach is a good improvement compared to recent skipper-CCD. The noise, after 3000+ non-destructive reads at -100C is probably closer to 0.20e- rms based on the PCH valley-peak ratio. At 0.15e- rms read noise the valley minimum should go to zero. Hopefully in a peer-reviewed journal version more critical details will be provided. Eric R Fossumhttps://www.blogger.com/profile/09413387887974939684noreply@blogger.comtag:blogger.com,1999:blog-19092890.post-14680987626335954092024-03-10T20:07:00.973+02:002024-03-10T20:07:00.973+02:00Venice 2 is only 6KVenice 2 is only 6KAnonymousnoreply@blogger.comtag:blogger.com,1999:blog-19092890.post-91597220858612317832024-03-10T15:18:00.698+02:002024-03-10T15:18:00.698+02:00"There is no real picture, full stop" I..."There is no real picture, full stop" I think this is a statement that is misleading. I think we can safely call the real picture the raw signal collected by the pixels, at least in most instances. After that, the picture is often transformed into something that agrees with our perception of the scene, and then sometimes "improved" beyond that. It is a slippery slope from the real picture to art, and there is nothing wrong with art. Film photographers have done this for a century or longer, and even our own brains perform improvements or hallucinations on images we thing we can recall with clarity. I think we can all agree that authentication of the "real picture" and a record of how it was embellished is important in some special circumstances. It is good that Samsung has made a stab at that here and reducing the unintended consequences of emerging AI technology for image manipulation is an altruistic aim. That genie is certainly already out of the bottle, for better and worse.Eric R Fossumhttps://www.blogger.com/profile/09413387887974939684noreply@blogger.comtag:blogger.com,1999:blog-19092890.post-41109449360905360212024-03-09T21:17:48.470+02:002024-03-09T21:17:48.470+02:00Gpixel needs to buy at least 2 dozen Sony Venice 2...Gpixel needs to buy at least 2 dozen Sony Venice 2 cameras (to get a good statistical sample size) and then do extensive testing on the dark noise and dynamic range to fill out the missing data on their GCINE comparison chart.Anonymousnoreply@blogger.comtag:blogger.com,1999:blog-19092890.post-71818471285220951762024-03-09T19:54:51.616+02:002024-03-09T19:54:51.616+02:00Modern day smartphone cameras are as much software...Modern day smartphone cameras are as much software as they are cameras. To fret over AI enhancements seems silly to me. Thanks for the article Atul.Aaron Hendricksonhttps://www.blogger.com/profile/04558774822320792376noreply@blogger.comtag:blogger.com,1999:blog-19092890.post-36037918100091332232024-03-08T04:13:39.969+02:002024-03-08T04:13:39.969+02:00The information I anonymously commented on last ti...The information I anonymously commented on last time, about "military" and "aerospace", has no formal information source. However, due to the problem of anonymous reviews, I cannot manage the comments. Please help the blogger delete it or not approve it. I regret posting this kind of information without verification. Thank you blogger.Anonymousnoreply@blogger.com