c6gowin Posted January 6, 2010 Share #41 Posted January 6, 2010 Advertisement (gone after registration) It had my name in the filename - now that's what I call personal service! That is pretty cool but, I wonder if that means you will have to get personalized firmware each time there is a generic firmware update? Link to post Share on other sites More sharing options...
Advertisement Posted January 6, 2010 Posted January 6, 2010 Hi c6gowin, Take a look here Does Leica Have a Problem?. I'm sure you'll find what you were looking for!
tashley Posted January 7, 2010 Share #42 Posted January 7, 2010 That is pretty cool but, I wonder if that means you will have to get personalized firmware each time there is a generic firmware update? Nope, this was a one-off fix and future FW generic updates will work 'off the shelf'... Link to post Share on other sites More sharing options...
Bo_Lorentzen Posted January 7, 2010 Share #43 Posted January 7, 2010 So the camera specific calibration file must be stored in flash, in a way that firmware can be updated without overwriting the camera profile. It would have the be the same thing in the M8 and M9 - otherwise the cameras would suddenly display all kinds of problems after each firmware upgrade. Guess the big news - is the ability to load a updated camera calibration file - as a end user operation compared to sending the camera to solms. If I'm really lucky Leica have included the same functionality in the M9. :-) . Link to post Share on other sites More sharing options...
delander † Posted January 7, 2010 Share #44 Posted January 7, 2010 A little update... I gave Leica a copy, yesterday afternoon, of a DNG file taken at ISO1250 showing a column defect from a bad pixel that wasn't mapped properly (see higher up this thread) and this morning I received by email a small (600kb) FW update which, when installed, fixed the problem completely. Tim Leica is a strange company. They come out with these exquisite products, with, for the price, seemingly basic defects and then introduce cool ways of fixing problems. I do hope they are able to introduce this for the M9 and not just reserve it for the S2. Little surprises like this restore one's faith in the company. Jeff Link to post Share on other sites More sharing options...
nugat Posted January 8, 2010 Share #45 Posted January 8, 2010 Fixing a row failure through mapping out...? So far I heard of single pixlels mapping out , which is way easier. Row mapping out must interpolate missing data from far less available sources (less pixels around the dead ones to draw from). That must be a major progress in the area, but then I'm no expert. I am just wondering how it works when there is an object projected by the lens on the sensor's mapped out row being exactly its size (a thin pole)? It does not register?I know it's an extreme example and might never happen, nevetherless it should illustrate the challanges behind the approach. Link to post Share on other sites More sharing options...
tashley Posted January 8, 2010 Share #46 Posted January 8, 2010 I believe that row failures are caused by single pixels affecting the readout for the rest of the row so fixing the pixel sorts the whole row without having to map out the whole row. Mark Norton will know the answer for sure though! Link to post Share on other sites More sharing options...
dfarkas Posted January 8, 2010 Share #47 Posted January 8, 2010 Advertisement (gone after registration) I believe that row failures are caused by single pixels affecting the readout for the rest of the row so fixing the pixel sorts the whole row without having to map out the whole row. Mark Norton will know the answer for sure though! Tim, I'm pretty certain this is the case. David Link to post Share on other sites More sharing options...
haribo Posted January 8, 2010 Share #48 Posted January 8, 2010 It had my name in the filename - now that's what I call personal service! I call that too few customers ... Link to post Share on other sites More sharing options...
georg Posted January 8, 2010 Share #49 Posted January 8, 2010 It seems to some kind of banding caused by the single "hot" photosite. It appeared in my M8 from one frame to another after several months and thousands of shots and they had to replace the whole sensor board - nice to hear that they finally insourced this knowledge. @haribo You're right, they should have send him a general fix for a single dead photosite - the chance would be 1:37500000 that it actually works... Link to post Share on other sites More sharing options...
Jamie Roberts Posted January 8, 2010 Share #50 Posted January 8, 2010 it didn't happen with any of my nikons and hasselblads but with all my leicas.peter Wow. And yet both my D3s fail (CD error; card full error) with both my Nikkor 85 1.4s Oh, and my 1ds2 and 5d went back to Canon for precisely the same hot pixel defect that Tim is showing. The power of anecdotes Link to post Share on other sites More sharing options...
Jamie Roberts Posted January 8, 2010 Share #51 Posted January 8, 2010 A little update... I gave Leica a copy, yesterday afternoon, of a DNG file taken at ISO1250 showing a column defect from a bad pixel that wasn't mapped properly (see higher up this thread) and this morning I received by email a small (600kb) FW update which, when installed, fixed the problem completely. Tim Great news Tim. This has happened on every reasonably high resolution digital I've ever used. Nikon and Canon fix these defects by remapping them as well, though I still had to send them in to their respective service departments. When the M8 had the same thing after a couple of years of shooting it required a sensor change, which I thought was a little extreme... Still, I'm happy to hear that Leica are catching up! I hope they can do this for the digital Ms as well. Link to post Share on other sites More sharing options...
tashley Posted January 8, 2010 Share #52 Posted January 8, 2010 Great news Tim. This has happened on every reasonably high resolution digital I've ever used. Nikon and Canon fix these defects by remapping them as well, though I still had to send them in to their respective service departments. When the M8 had the same thing after a couple of years of shooting it required a sensor change, which I thought was a little extreme... Still, I'm happy to hear that Leica are catching up! I hope they can do this for the digital Ms as well. Thanks Jamie! BTW I am now happy to give my personal answer to the question the OP asked at the top of this thread. No. :D:D Link to post Share on other sites More sharing options...
markowich Posted January 8, 2010 Share #53 Posted January 8, 2010 Wow. And yet both my D3s fail (CD error; card full error) with both my Nikkor 85 1.4s Oh, and my 1ds2 and 5d went back to Canon for precisely the same hot pixel defect that Tim is showing. The power of anecdotes wow. that was very anectotical. peter Link to post Share on other sites More sharing options...
marknorton Posted January 9, 2010 Share #54 Posted January 9, 2010 I believe that row failures are caused by single pixels affecting the readout for the rest of the row so fixing the pixel sorts the whole row without having to map out the whole row. Mark Norton will know the answer for sure though! Not sure I agree with that. The pixel and column defects (not rows, think Excel) remain on the sensor, no matter what. What changes is that the camera is taught that the defects exist and takes corrective action after the data has been read out to hide them in the "published" image. There's likely to be an eye-opening test mode in the camera which allows the image to be captured without defect correction, warts and all. Without boring you all rigid, you can think of a sensor as being like an array of buckets which are filled with water when you make the exposure according to the light level. How to measure the contents of each bucket? Behind these "buckets" is a network of vertical "bucket brigades" which allow the contents of one bucket to be poured into the next position down, eventually reaching the bottom row where (using the published M8 sensor structure as an example) they are now moved horizontally to one of two readout points at the end of the row and measured. Get the sequence of bucket tipping operations right and each of the millions of pixels is measured in turn by just a handful of measuring points. This analogy allows a simplistic explanation of the common sensor defects. A stuck pixel is like a bucket which is always full or empty, whatever you do with it. A cluster fault is like a little local flood where the buckets overflow and the drains are backed up. A column defect is a problem with the bucket brigade - the levels are not maintained as the water is transferred from one bucket to the next due to leaky buckets, buckets not emptying properly or it pouring with rain. The Kodak sensor spec (M8) accepts a bucket brigade with leaky buckets but not one trying to work in a rainstorm. No amount of masking out single bad pixels will fix a column defect, so if there is one, the camera has to interpolate across the entire column. The Kodak spec says bad columns must be at least 4 pixels apart to allow meaningful interpolation without the result being visible. Clearly, there comes a point where masking out a large number of columns meaningfully affects the sensor resolution and there's an upper limit of 30 allowed. As for whether a thin vertical detail in the image will be rendered invisible, the theory says unlikely. If you take a lens at 40 lp/mm, the smallest resolvable detail is 25 microns across which, with a 6 micron pixel pitch you are sampling 4 times. Sampling theory requires you to sample at least twice, so even if you loose one column, you should be OK. Link to post Share on other sites More sharing options...
Jamie Roberts Posted January 9, 2010 Share #55 Posted January 9, 2010 wow. that was very anectotical.peter Um, indeed Link to post Share on other sites More sharing options...
Gentleman Villain Posted January 10, 2010 Share #56 Posted January 10, 2010 Mark, thanks for the post and the bucket brigade analogy Link to post Share on other sites More sharing options...
Recommended Posts
Archived
This topic is now archived and is closed to further replies.