wparsonsgisnet Posted July 23, 2008 Share #21 Posted July 23, 2008 Advertisement (gone after registration) Jaap, I agree, Leica has a formidable challenge that results from their being a small manufacturer. They can't put the firmware in a chip because it costs too much to make a chip and their volume is too low. Not having looked at any of the code, I can only imagine how small the computer memory is in the M8, and how great the tasks required of it. I laud their accomplishments in getting w/b right by now, in the new profiles that C1 has provided, and especially in the quality of the images that come from this little gem. I just put my money where my mouth is -- I bot a second body (I'll have to rehab my own, I can't get a replacement). Regards to all, Link to post Share on other sites More sharing options...
Advertisement Posted July 23, 2008 Posted July 23, 2008 Hi wparsonsgisnet, Take a look here The reality of the 8-bit Leica. I'm sure you'll find what you were looking for!
diogenis Posted July 23, 2008 Share #22 Posted July 23, 2008 Please could you read the following: KammaGamma Articles Solving the Leica M8 DNG riddle and let us know how it compares with the LFI article. Regards Per More or less they say the same thing yes. The LFI article is more descriptive and a lot more interesting ofcourse. Link to post Share on other sites More sharing options...
t024484 Posted July 23, 2008 Share #23 Posted July 23, 2008 If you PM your e-mail I can do so, but it will only work if you use Leica's compressing algorithm Jaap, I do not know Leica´s compression algorithm, but most likely it will be a logaritmic algorithm. When using this assumption, I come to the following algorithm. Compressed 8 bit value=Integer [(43.3*Ln(14 bits value+43.3)) - 163.16] This gives me the following translation: 14bits----Calcul-----Integer 0 ______ 0 ______ 0 1 ______ 0.99 ____ 1 2 ______ 1.96 ____ 2 3 ______ 2.90 ____ 3 4 ______ 3.83 ____ 4 5 ______ 4.73 ____ 5 6 ______ 5.62 ____ 6 7 ______ 6.49 ____ 6 8 ______ 7.34 ____ 7 9 ______ 8.18 ____ 8 10 _____ 9.00 ____ 9 100 ___ 51.82 ___ 52 1000 -- 137.78 --- 138 16000 - 256.11 --- 256 My email address is: hapolak@telfort.nl Hans Link to post Share on other sites More sharing options...
Shootist Posted July 23, 2008 Share #24 Posted July 23, 2008 I just put my money where my mouth is -- I bot a second body (I'll have to rehab my own, I can't get a replacement). Regards to all, Bill what do you mean "(I'll have to rehab my own, I can't get a replacement)"??? Replacement M8 from Leica NJ for your original? Now that you have a second body you need to send in your orignal to have the On/Off switch fixed and any other glitches it may have. Link to post Share on other sites More sharing options...
marknorton Posted July 23, 2008 Share #25 Posted July 23, 2008 Jaap,I do not know Leica´s compression algorithm, but most likely it will be a logaritmic algorithm. When using this assumption, I come to the following algorithm. Compressed 8 bit value=Integer [(43.3*Ln(14 bits value+43.3)) - 163.16] This gives me the following translation: 14bits----Calcul-----Integer 0 ______ 0 ______ 0 1 ______ 0.99 ____ 1 2 ______ 1.96 ____ 2 3 ______ 2.90 ____ 3 4 ______ 3.83 ____ 4 5 ______ 4.73 ____ 5 6 ______ 5.62 ____ 6 7 ______ 6.49 ____ 6 8 ______ 7.34 ____ 7 9 ______ 8.18 ____ 8 10 _____ 9.00 ____ 9 100 ___ 51.82 ___ 52 1000 -- 137.78 --- 138 16000 - 256.11 --- 256 My email address is: hapolak@telfort.nl Hans We don't know whether the A/D converter on the M8 is 12 or 14 bit but whichever it is, they will simply translate the value output to values in the range 0 - 255 using a lookup table which they can tweak endlessly to provide the characteristic they are looking for. It may be the table has changed over time with successive firmware releases. The table is specified in the DNG of course which is how the DNG processor gets the values back to 12/14/16 bit. One simple approach is to multiply values in the range 0 - 4095 by 16 to give values in the range 0 - 65520, then take the square root to give values in the range 0 - 255. I assume you have obtained your formula using some sort of curve fitting tool; the m8 doesn't have the horsepower to do much on the fly, so a lookup table is the most likely scenario. Link to post Share on other sites More sharing options...
sandymc Posted July 23, 2008 Share #26 Posted July 23, 2008 Mark is correct on the multiply and square root..... Sandy Link to post Share on other sites More sharing options...
psss Posted July 23, 2008 Share #27 Posted July 23, 2008 Advertisement (gone after registration) i would still loooooove to see a 16bit raw option....no matter how long it owld take to write and how large the files would get....just give me the OPTION and the option to switch it back tho the DNG i am getting now.... there are so many discussions about pixel depth...not only with the m8...DMF back makers call their backs 16 bit and there are many who dispute those claims....for me the biggest diffence between files has always been the pixeldepth and i have always felt that the higher bit capture will provide better results and better files to uprez....and work with in post production.... and imo the m8 provides files that are very close to 16bit (or what is considered, what i have seen form what is labeled 16bit), and definitely better then 12 bit canons/nikons... i am also a bit disappointed with the 14bit files the new canons and nikons produce....i just don't see it.... Link to post Share on other sites More sharing options...
wparsonsgisnet Posted July 23, 2008 Share #28 Posted July 23, 2008 Bill what do you mean "(I'll have to rehab my own, I can't get a replacement)"???Replacement M8 from Leica NJ for your original? Now that you have a second body you need to send in your orignal to have the On/Off switch fixed and any other glitches it may have. Steve, joking about my own body, the one with the legs and arms. Link to post Share on other sites More sharing options...
jaapv Posted July 23, 2008 Share #29 Posted July 23, 2008 Now I'm wondering about that on/off switch.... Link to post Share on other sites More sharing options...
jaapv Posted July 23, 2008 Share #30 Posted July 23, 2008 Hans, I'll do the following: Take two images, M8 and DMR side by side, matching as closely as possible. Obviously not the same lens, but I'll take the Elmarit-R 2.8-60 and the Elmar-M 2.8-50 The fingerprint is quite similar. I'll mail both DNG's to you to play with. It will be Friday, tomorrow is desperately busy. Jaap,I do not know Leica´s compression algorithm, but most likely it will be a logaritmic algorithm. When using this assumption, I come to the following algorithm. Compressed 8 bit value=Integer [(43.3*Ln(14 bits value+43.3)) - 163.16] This gives me the following translation: 14bits----Calcul-----Integer 0 ______ 0 ______ 0 1 ______ 0.99 ____ 1 2 ______ 1.96 ____ 2 3 ______ 2.90 ____ 3 4 ______ 3.83 ____ 4 5 ______ 4.73 ____ 5 6 ______ 5.62 ____ 6 7 ______ 6.49 ____ 6 8 ______ 7.34 ____ 7 9 ______ 8.18 ____ 8 10 _____ 9.00 ____ 9 100 ___ 51.82 ___ 52 1000 -- 137.78 --- 138 16000 - 256.11 --- 256 My email address is: hapolak@telfort.nl Hans Link to post Share on other sites More sharing options...
wparsonsgisnet Posted July 23, 2008 Share #31 Posted July 23, 2008 P-l-e-a-s-e don't switch me off. Link to post Share on other sites More sharing options...
Mauribix Posted July 23, 2008 Share #32 Posted July 23, 2008 Now I'm wondering about that on/off switch.... i don't have the faintest idea... Link to post Share on other sites More sharing options...
Mauribix Posted July 23, 2008 Share #33 Posted July 23, 2008 Hans,I'll do the following: Take two images, M8 and DMR side by side, matching as closely as possible. Obviously not the same lens, but I'll take the Elmarit-R 2.8-60 and the Elmar-M 2.8-50 The fingerprint is quite similar. I'll mail both DNG's to you to play with. It will be Friday, tomorrow is desperately busy. please, Jaap, may I have them too? thanks! Link to post Share on other sites More sharing options...
jaapv Posted July 23, 2008 Share #34 Posted July 23, 2008 Please remind me of your mail by PM. Link to post Share on other sites More sharing options...
Artichoke Posted July 24, 2008 Share #35 Posted July 24, 2008 where bit depth is most critical is at the capture stage where the analog data coming off the sensor gets digitized I use a Fujifilm S3 which has a 14 bit AD converter the jpgs from this camera are outstanding, despite being 8 bit files & I believe this is because the files presented to the jpg engine (a superior one, perhaps the best available) have more data with which to work Leica could take a lesson from Fujifilm when it comes to jpgs the RAW files from the S3 are 25 MB in size and slowed down this camera to the point where this became a liability for many users ...I shoot the S3 almost exclusively using jpgs Leica could give Fujifilm a lesson in how to design an efficient compression of its capture files ...using the DNG standard was a brilliant move on their part the DNGs from the M8 are very rich & can take a great deal of post processing flogging before breaking up whether having all the original 14 bit data would make much of a difference is an unprovable point ...almost certainly these would not suffer in quality, but equally certainly they would be significantly bloated I think Leica should have given users an option to have the "full" file capture available as a true RAW file, though I suspect few users would take the hit in speed and disk space for what likely is a marginal improvement in file quality Link to post Share on other sites More sharing options...
diogenis Posted July 24, 2008 Share #36 Posted July 24, 2008 Ok this is how I understand the procedure inside the m8: You have the 16bit Kodak sensor that can generate files with pixel depth of upto 65536 brightness or tonal values using electricity charge (2^16). Assume the higher value 65536, multiply by 4 and square root it: So now you get 256 max, which can be coded using 8bits. 16b x 4 square root=8b. To get back the original value you simply square and divide by 4 again. Adobe's DNG protocol, gives the option to store these tonal values using pointers that point in a table. It seems, that Leica is the first that incorporates that table within every DNG file you create while shooting, inside the m8 that is. So, with every photo, you get that table which contains tonal values @ 14bits and then just indexes that point to this table for every pixel => just 1 bit per pixel! What the article dont say is, if that table is the same for all your photos, or it dynamically changes to accommodate for every lighting situation you encounter, in which case that would simply mean that even 256 indexes would be sufficient, as they would indeed point into a narrower tonal spectrum. In short, if your scene is "balanced" leica's conversions will resolve as if it was a lossless 14b raw file. Link to post Share on other sites More sharing options...
luigi bertolotti Posted July 24, 2008 Share #37 Posted July 24, 2008 We don't know whether the A/D converter on the M8 is 12 or 14 bit but whichever it is, they will simply translate the value output to values in the range 0 - 255 using a lookup table which they can tweak endlessly to provide the characteristic they are looking for. It may be the table has changed over time with successive firmware releases. The table is specified in the DNG of course which is how the DNG processor gets the values back to 12/14/16 bit. One simple approach is to multiply values in the range 0 - 4095 by 16 to give values in the range 0 - 65520, then take the square root to give values in the range 0 - 255. I assume you have obtained your formula using some sort of curve fitting tool; the m8 doesn't have the horsepower to do much on the fly, so a lookup table is the most likely scenario. Mark... just as an idea... do you think that, to try to understand the lookup table that probably is "written" in M8 firmware, could be useful to shot a grayscale target (many are easily available) and then analizying the DNG values through a tool like PhotoME ? I ask this for I appreciate PhotoME, and am guessing how to make some "scientific" use of it... Link to post Share on other sites More sharing options...
sandymc Posted July 24, 2008 Share #38 Posted July 24, 2008 The lookup table isn't dynamic - it remains the same for all images. Although there is no reason why Leica couln't change that if they wanted. It is easily extractable from the DNG. In fact, If you set CornerFix to "verbose" mode, you will see the first part of the table displayed. Sandy Link to post Share on other sites More sharing options...
diogenis Posted July 24, 2008 Share #39 Posted July 24, 2008 Do you know of any tool available to translate dng files, since DNG is open source, no? edit: this is how Adobe refers to those tables in its recently released (May 08) dng spec sheet: "LinearizationTable Description LinearizationTable describes a lookup table that maps stored values into linear values. This tag is typically used to increase compression ratios by storing the raw data in a non-linear, more visually uniform space with fewer total encoding levels. If SamplesPerPixel is not equal to one, this single table applies to all the samples for each pixel." Link to post Share on other sites More sharing options...
diogenis Posted July 24, 2008 Share #40 Posted July 24, 2008 The lookup table isn't dynamic - it remains the same for all images. Although there is no reason why Leica couln't change that if they wanted. It is easily extractable from the DNG. In fact, If you set CornerFix to "verbose" mode, you will see the first part of the table displayed. Sandy Sandy, We are interested in the actual data. I guess it must be somewhat easy to validate if it's dynamic or not. And there is no reason not to make it dynamic and to focus on a portion of the range of values instead of the whole spectrum, unless this live processing requires a lot of computation power Link to post Share on other sites More sharing options...
Recommended Posts
Archived
This topic is now archived and is closed to further replies.