rosuna Posted April 25, 2007 Share #181 Posted April 25, 2007 Advertisement (gone after registration) Another point is noise. I don't know if it is possible to reduce noise in the DNG files by means of firmware fine tuning. Can be expected an improvement here sometime in the future? Link to post Share on other sites More sharing options...
Advertisement Posted April 25, 2007 Posted April 25, 2007 Hi rosuna, Take a look here Official Response from Leica on Laundry List. I'm sure you'll find what you were looking for!
rosuna Posted April 25, 2007 Share #182 Posted April 25, 2007 The Leica M8 has an internat interpreter of RAW data, employed in the "development" of JPG files. I suppose this interpreter also affects how the DNG files are written. I would like to ask Leica if future improvements are possible here. Link to post Share on other sites More sharing options...
eronald Posted April 25, 2007 Share #183 Posted April 25, 2007 I consider the white balance written into Raw files with 1.102 as BROKEN. I have tested two cameras and found this. It was hit and miss before, now it's just plain bad. Edmund Link to post Share on other sites More sharing options...
elopezso Posted April 25, 2007 Share #184 Posted April 25, 2007 Edmund, I gather you mean the auto white balance, correct? I think the specific ones, daylight, cloudy, etc are much better. But the auto white balance is indeed a travesty! Edwin Link to post Share on other sites More sharing options...
wparsonsgisnet Posted April 25, 2007 Share #185 Posted April 25, 2007 Well, clearly Leica is messing with the raw file -- that's the only way that we would require new profiles with each firmware release. This so-called raw file is supposed to be what the sensor sees, not what Leica thinks they want this time. Link to post Share on other sites More sharing options...
eronald Posted April 25, 2007 Share #186 Posted April 25, 2007 Edwin - Yes, I mean the auto white balance is broken for the Raw files as read into C1. Bill - Yes,indeed, they seem to be changing things around a lot. With the profiling I'm chasing a rapidly moving target. Edmund Link to post Share on other sites More sharing options...
wlaidlaw Posted April 25, 2007 Share #187 Posted April 25, 2007 Advertisement (gone after registration) Edwin - Yes, I mean the auto white balance is broken for the Raw files as read into C1. Bill - Yes,indeed, they seem to be changing things around a lot. With the profiling I'm chasing a rapidly moving target. Edmund \ The difference between the preset white balances seems wider than before. For example if you took a daylight shot on 1.09 when the WB was set to tungsten, it came out, as you would expect, with a blue tint. Now on 1.102, if you do the same, it comes out bright blue - almost a blue and white image. I would agree with Edmund, that on 1.09 or 1.092, if you used AWB on a lens without a UV/IR filter, on daylight shots in reasonable light, the AWB on both DNG and JPEG was within easy correction limits. Now it seems all over the place. I have given up AWB altogether. This is fine if you are in relatively consistent lighting but if you are in an old city environment, where your lighting is changing from shot to shot, it is an absolute pain having to keep resetting manual WB. I think this, together with the actual bugs, is by far the most urgent thing to get right on the next firmware issue. Wilson Link to post Share on other sites More sharing options...
wparsonsgisnet Posted April 25, 2007 Share #188 Posted April 25, 2007 Wilson, I always use AWB (for the jpg's) and also shoot raw. That is, I store both a hi-res jpg and the dng. And, I whitebalance the dng in C1, when I process it. I guess Leica can do whatever they want with the AWB. Certainly they're getting enuf complaints about it, so they must be working on it. Also, the notes for the release said they tried to make fixes. Why is the raw file changing???!!! This makes NO sense. Not to make too big a deal about it, it's like the pharmacist deciding to change my life-saving meds, away from the dose the doctor has prescribed. [Disclaimer: This is not my particular problem ... it's really empty space between the ears.] Maybe I don't understand the basic concept of a raw file. I thot it was what the sensor saw at the time of image capture. If Leica is transforming it during compression, ok, they can do that if they unwind it again (in some form). And, if they changed the compression "formula" at version 1.09 or 1.091, then the colors in the "raw" file should not be changing in the 2 subsequent releases. GUY, can you talk to Stefan about this? Link to post Share on other sites More sharing options...
Chuck Hatcher Posted April 25, 2007 Share #189 Posted April 25, 2007 ...Maybe I don't understand the basic concept of a raw file. I thot it was what the sensor saw at the time of image capture. If Leica is transforming it during compression, ok, they can do that if they unwind it again (in some form)... I actually wondered about this regarding vignetting/cyan shift compensation for coded lenses. Is the compensation applied to both jpegs and raw files? If not, it would seem to be of little use. If so, it means raw files aren't all that raw. Link to post Share on other sites More sharing options...
ho_co Posted April 25, 2007 Share #190 Posted April 25, 2007 The graph wasn't mine, but it was made only for illustrating the basic idea of the non-linear 8-bits compression. The small differences with the original graph are meaningless. Rubén-- Sorry to have annoyed by saying that your graph wasn't the same as the one in LFI. I think there are several meaningful points of difference, but you are correct that they are all minor. Your new posting of charts is excellent and makes the point very strongly that (as Jamie said) things are "fishy fishy fishy." What you have shown is completely accurate in regard to the M8's output. But remember that Leica is relying heavily on Bayer interpolation to fill in the absent levels, and the charts do not show this. Indeed, the M8 does produce better images than these stepped graphs would indicate, doesn't it? If you didn't know how the compression algorithm worked, would you be able to castigate the M8 image quality? Personally, I think Leica has given us a very innovative and successful compromise in the M8's compression scheme. Nonetheless, the only way they will put this ongoing debate behind them is not by showing us why their way is fine, but by including the option to write out full 16 bit data. BTW--I'm with you in heart and practice: Even though I know that Nikon's compressed NEFs are "virtually lossless," I shoot only the uncompressed version. --HC Link to post Share on other sites More sharing options...
rosuna Posted April 25, 2007 Share #191 Posted April 25, 2007 Even Nikon's compression has much more tonal variations than the DNGs. I would use compressed files without hesitation, if Leica provides a safe margin in their compression scheme. Link to post Share on other sites More sharing options...
farnz Posted April 26, 2007 Share #192 Posted April 26, 2007 ... the basic concept of a raw file. I thot it was what the sensor saw at the time of image capture. If Leica is transforming it during compression, ok, they can do that if they unwind it again (in some form)... Bill, to my mind this is not radically different from any other camera that captures raw, it's just that the conversion from raw to (in this instance) DNG is done in-camera rather than in proprietary raw file conversion software. My apologies if this is old hat to you but for others who may not be familiar, the sensor comprises a bunch of photodiodes that each emit an analogue voltage when light is shone on them. The voltage output from the photodiode is proportional to the intensity of the light and is fed into an analogue to digital convertor (ADC) which assigns it a digital value (word) in the range 0 to 16385 (if it's a 14-bit system). The raw file is composed of all of these assigned digital values and in fact is just a series of 14-bit words. Put that on screen and obviously all we'd see is a stream of ones and zeros. Raw file conversion software looks at each word and decides whether it comes from a photodiode with a red, green or blue bayer filter and assembles them all into the image that you see on your screen. The M8 does this assembly in-camera and produces it in a DNG file (and also compresses the 14-bit data into 8-bit data). The point is that raw file conversion software will be making similar changes to the raw data that the M8 is making in-camera except we're less aware of it. That's my simplistic understanding of how it works but I'm sure others will correct me if I've missed the mark. Pete. Link to post Share on other sites More sharing options...
wparsonsgisnet Posted April 26, 2007 Share #193 Posted April 26, 2007 Pete, nice description. Thanks for that. I would like the data, in whatever fashion a particular manufacturer captures it, to be saved and unwound, if necessary, when I work on the image. The raw version of the image needs to be accessible to all other pieces of sw that any of us uses. What I am confused about is: 1. We know that Leica compresses the tile -- and perhaps this is responsible for opening up the shadows. This camera has more shadow detail than any other device I have ever used. 2. We know that Leica uses a conversion scheme to transform the data from 14-bits to 8, and that this table has changed, once so far, I believe. 3. We know that the color balance changes with each firmware release. I can see it in the jpegs, and Edmund is having to make new profiles for each release. Others may be changing their profiles, as well. 4. This means that the so-called raw image is changing all the time. I'm not excited about having the colors change every time Leica updates the firmware. I'm being razor-bladed to death by having to buy new profiles all the time. We know that the C1 profile for the original M8 was not satisfactory. Those guys aren't even trying to bring out new profiles! There is a similar problem for ACR. The images from this device are superb. In my case, I attribute this to 3 things: 1. Lenses to die for, 2. Leica's software that gives wonderful color and shadow detail, and 3. Edmund's profiles that give me the correct color. I have used Edmund's profiles for every firmware version, and they are CRITICAL to my getting good color rendition. Maybe others, Guy, Sean, et al, have another way to solve this problem, but I don't understand why the color balance keeps changing. As I said, the raw file is supposed to represent what the sensor sees, with no in-camera "correction." Link to post Share on other sites More sharing options...
Guest guy_mancuso Posted April 26, 2007 Share #194 Posted April 26, 2007 Bill can't count on it but i think leica is done tweaking color which this latest firmware is tweaked for the Leica IR filters. So I don't think we will see any more changes in this for awhile if ever. So doing profiles now is probably pretty safe. Link to post Share on other sites More sharing options...
wparsonsgisnet Posted April 26, 2007 Share #195 Posted April 26, 2007 Thanks for the info, Guy. I'm looking forward to Sean's comprehensive analysis of different filters. Since I own a bunch of 486's, and since Leica has some new formulation, I will be interested which ones I will have to replace. I work in software testing and expect to see a document with each software release I work with. That document usually has a section called something like New Features. I would really like to see something like that from Leica. They know what they are doing to the color renditions. It would be very helpful to know what changes to expect in our images. I don't want to whine (after all, I'm not allowed to at home...) and I certainly don't want to be a complainer. I really like the control we have in digital photography and I don't miss the current set of films. DON'T even ask about panatomic-X and Dupont Varilour papers. Damn. I've got prints that can never even be reprinted. Great lenses. Great camera. In one sense the M8 is new technology because of the additional IR sensitivity. One expects a learning curve and growing pains with new technology. It'll be nice to have the color rendition stabilized. My single unhappiness with this camera has to do with the outlawed 50mm DR. When the pocketbook has recovered, I'm going to buy a second one and have its close-focus flange removed. Link to post Share on other sites More sharing options...
Guest guy_mancuso Posted April 26, 2007 Share #196 Posted April 26, 2007 LOL i have been asking for a new feature in firmware list since the beginning. I agree it would be nice to know what was done. On the 486 besides the cyan shift difference between leica and B+W filters should be fairly small on most lenses but the color may have a slight difference which I noticed the leica produces a slightly warmer tone or more normal than the B+W. So we may need to adjust there . I see more green in the B+W but most likely need to run more tests on this which i have not done but casual comparisions is what i have noticed Link to post Share on other sites More sharing options...
rosuna Posted April 27, 2007 Share #197 Posted April 27, 2007 I don't know if things look "fishy fishy fishy", but I observe better image quality in the DMR files (sharpness, color, tonal variation). I don't know what tweaks could improve this. A wider storing space can be a positive input (not necessarily 16bits uncompressed), but maybe it doesn't make difference. However, I think it could be an important factor, because the sensor is similar. It could improve color and tonal variation. The internal noise "handling" for DNG files should improve too. I would ask Leica (more than suggest) if these improvements can be done in the future. Link to post Share on other sites More sharing options...
sdai Posted April 27, 2007 Share #198 Posted April 27, 2007 Another point is noise. I don't know if it is possible to reduce noise in the DNG files by means of firmware fine tuning. Can be expected an improvement here sometime in the future? Most other contemporary digital SLRs do noise attenuation either on the sensor board (such as Canon) or through an add-on module/processor upstream the DSP but downstream the sensor output (such as Nikon). It seems that the M8's DSP isn't powerful enough to handle 16 bit data crunching, I highly doubt it's capable of intensive noise reduction calculation either. Link to post Share on other sites More sharing options...
Philinflash Posted April 27, 2007 Share #199 Posted April 27, 2007 I find the AWB function on the M8 to be distinctly sub-par. It compares unfavorably to $200 purse-type digital cameras; in fact the M8 does not cope with changing light conditions nearly as well as those do. Complicating the situation, it is difficult to overcome the problem by selecting alternate presets on the in-camera menu unless you know what the color temperature is on the Kelvin scale, which requires a separate color temperature meter that I do not want to buy or carry. For instance, what is the difference between the "Cloudy" and "Shadow" pre-settings on the menu? I cannot find what the Kelvin equivalent of these settings are, does anyone know? The manual is nearly mute on the subject of WB. Why can't Leica publish that? One does not always have the time to experiment with different settings on the trot. And why can't the camera remember Kelvin settings, instead of just starting off at the lowest color temperature, 2000K? P. S. I do not have the promised filters, so I am shooting unfiltered and the firmware is fully up to date with 1.102. Philip Kozloff Link to post Share on other sites More sharing options...
wparsonsgisnet Posted April 27, 2007 Share #200 Posted April 27, 2007 I find the AWB function on the M8 to be distinctly sub-par. It compares unfavorably to $200 purse-type digital cameras ... This comparison is not strictly fair. The so-called $200 digicam benefits from the awb circuitry designed by giant firms like Canon. They spend megabucks on a wb circuit for the top-of-the-line camera and it gets ported to the toy. There is a perfectly good solution to awb. Shoot raw and take a calibration picture during your session. You can carry a small WhiBal card in your pocket. Result: WB:always correct. Cost: insignificant. Ease-of-mind: incalculable. Link to post Share on other sites More sharing options...
Recommended Posts
Archived
This topic is now archived and is closed to further replies.