Steve Spencer Posted December 17, 2015 Share #21 Posted December 17, 2015 Advertisement (gone after registration) I recognize that the embedded images to the DNG files are of worse quality than the ones I have now in my SL, coming from DNG+JPEG files, and... I understand your reasons and I only ask myself what hell they have done with the embedded JPEGs. inner my Leica M. Leica M (24 MG also) has the following options: JPEG fine, JPEG standard, DNG and all the combinatios of DNG with the other types of JPEGs? The same occurs with the M9 (CCD sensor) with still better enlarged image than the M one. It seems to me, Michael, that Leica reasons have more to do with the capacity of shutting many and fast images per second saved in the buffer. ??? Otherwise I can't understand why the problem of SL is different of the M(s). and the M9P(s). Francisco. As I understand it, and I could be wrong, what the M9 and M240 do is save a medium quality jpeg with the DNG and this medium quality jpeg is what they use for image review on the LCD (or the attached EVF with an M240). This works pretty well with the lower resolution of those monitors. It may work well for the EVF on the SL for some folks, but given the higher resolution I suspect many would prefer a large jpeg for review. Now when you select the image to be saved as a jpeg only on the M9 or M240, then I assume that they use whatever size jpeg you save. If you select DNG plus jpeg (or just DNG), however, I believe the M9 and M240 use the medium quality jpeg embedded in the DNG. Now on the SL all that is embedded in the DNG is a thumbnail. If you select DNG only then than thumbnail is used for image review. If you select DNG plus jpeg, however, whatever size jpeg you selected is used for image review. So, Leica could have just embedded a medium quality jpeg in the DNG, like I think they did with the M9 and M240, and always used this medium jpeg for image review, but I suspect at least some users would have been dissatisfied with that for image review using the EVF. They could have also just embedded a large jpeg with the DNG, but I suspect some people would not have liked that because it does increase the size of the DNG file by about 10 percent. They way they did it gives the user the choice of both what size jpeg to use for image review and whether to save it or not. I think once you understand it, the solution is a good one as I am all for choice over these matters. Link to post Share on other sites More sharing options...
Advertisement Posted December 17, 2015 Posted December 17, 2015 Hi Steve Spencer, Take a look here Quality of enlarged photos in DNG format.. I'm sure you'll find what you were looking for!
FMB Posted December 17, 2015 Author Share #22 Posted December 17, 2015 Scott, I've already realized the JPEG weight. and I'm happy with the very very good quality of the (JPEG) enlarged images, but I'm lazy and disturbs me a little to erase the files I need not in my computer. Thank you for your assistance. Michael, you have being very gentle and patient helping me to understand the "why" things are like are and repeat what I've said above: all understood. Francisco. Link to post Share on other sites More sharing options...
barjohn Posted December 17, 2015 Share #23 Posted December 17, 2015 Since the cameras all appear to be able to generate JPGs very quickly, why not generate the JPGs needed for review on the fly from the captured DNG files, hold them in the buffer while in review mode and discard once the camera is back to shooting? It could start generating with the most recent shot the moment the view button is pressed and keep generating in the background for other recent shots. Might be a slight lag on initial view but no wasted space. Link to post Share on other sites More sharing options...
ramarren Posted December 17, 2015 Share #24 Posted December 17, 2015 Since the cameras all appear to be able to generate JPGs very quickly, why not generate the JPGs needed for review on the fly from the captured DNG files, hold them in the buffer while in review mode and discard once the camera is back to shooting? It could start generating with the most recent shot the moment the view button is pressed and keep generating in the background for other recent shots. Might be a slight lag on initial view but no wasted space. There are only a few makes that offer post-capture raw conversion in-camera (Olympus, Pentax {i think}, ... others? I'm not sure). I suspect this has to do with how the hardware dataflow in the camera is structured. It seems that Olympus feels this is a signal feature and allows their cameras to be used as not only capture devices but post-processing image development tools as well, so they build the access to the processing chain into their cameras allowing the camera to re-process a raw file from storage. So ... If this supposition is correct, the only way Leica could offer "on the fly" raw processing would involve a redesign of the capture/storage hardware allow access to captured raw files in order to feed the data back into the rendering engine. That would not be a simple change. Link to post Share on other sites More sharing options...
barjohn Posted December 17, 2015 Share #25 Posted December 17, 2015 You may be right but I seem to recall Panasonic recently introduced a multi-focus capability and given Leica's partnership with Panasonic perhaps it is possible. Link to post Share on other sites More sharing options...
mjh Posted December 17, 2015 Share #26 Posted December 17, 2015 It seems to me, Michael, that Leica reasons have more to do with the capacity of shutting many and fast images per second saved in the buffer. ??? How so? Saving a DNG with an embedded low-quality JPEG would even be slightly faster than saving a DNG with just a thumbnail and a high-quality JPEG. Link to post Share on other sites More sharing options...
mjh Posted December 17, 2015 Share #27 Posted December 17, 2015 Advertisement (gone after registration) How does the M240 manage it, then? to my eyes, satisfactorily. The M and S do represent one way of handling this while the X, T, Q, and SL do represent another. Link to post Share on other sites More sharing options...
LocalHero1953 Posted December 17, 2015 Share #28 Posted December 17, 2015 The M and S do represent one way of handling this while the X, T, Q, and SL do represent another. Well, I find the M approach more intuitive: I just used the M without having to worry about it or read the manual, whereas I needed the help of this forum to work out why "DNG only" had the apparently unrelated consequence of not showing you an enlarged image. Hey ho, such are the ways firmware writers' minds work. Link to post Share on other sites More sharing options...
CheshireCat Posted December 18, 2015 Share #29 Posted December 18, 2015 Since the cameras all appear to be able to generate JPGs very quickly, why not generate the JPGs needed for review on the fly from the captured DNG files, hold them in the buffer while in review mode and discard once the camera is back to shooting? It is even easier than this. When the camera saves DNG + JPG: 1) Raw sensor data are converted to RGB. 2) RGB data are displayed as preview. 3) RGB data are rescaled and compressed to a smaller size thumbnail JPG image. 4) Raw sensor data are saved into a DNG file, embedding the thumbnail JPG image. 5) Original full-size RGB data are compressed to a full-size JPG image. 6) Full-size JPG image are saved to a JPG file. When the camera saves DNG only, then only steps 1 to 4 are required. For shot preview (and enlargement), step 2 is the same, so it does not make any sense for it to be different (low-quality or anything). ... Now, when playing back the images from the card, all good cameras do the following: 1a) If full-size JPG file is present, then read it and decode it to the RGB buffer. 1b) Else (only DNG file is present), read it and decode only the embedded JPG thumbnail into the RGB buffer. 2) Display the RGB buffer (note that in case 1b, it will be low-resolution for a few instants till step 5 below is complete). 3) In case only DNG was present, then read raw data from the DNG file. 4) In case only DNG was present, then convert raw data to a new full-resolution RGB buffer. 5) In case only DNG was present, then display the new full-resolution RGB buffer (if you zoomed-in meanwhile, you will see the resolution suddenly improving). If a camera does not perform steps 3, 4, 5 then the firmware is crap. Link to post Share on other sites More sharing options...
ramarren Posted December 18, 2015 Share #30 Posted December 18, 2015 ... Now, when playing back the images from the card, all good cameras do the following: 1a) If full-size JPG file is present, then read it and decode it to the RGB buffer. 1b) Else (only DNG file is present), read it and decode only the embedded JPG thumbnail into the RGB buffer. 2) Display the RGB buffer (note that in case 1b, it will be low-resolution for a few instants till step 5 below is complete). 3) In case only DNG was present, then read raw data from the DNG file. 4) In case only DNG was present, then convert raw data to a new full-resolution RGB buffer. 5) In case only DNG was present, then display the new full-resolution RGB buffer (if you zoomed-in meanwhile, you will see the resolution suddenly improving). If a camera does not perform steps 3, 4, 5 then the firmware is crap. I don' t know of any cameras that do 3, 4, or 5. Every camera I know of reads either a companion thumbnail JPEG, an embedded JPEG preview, or a companion JPEG when displaying raw files stored in the camera. Ipso facto, all camera firmware is crap. Link to post Share on other sites More sharing options...
ramarren Posted December 18, 2015 Share #31 Posted December 18, 2015 Well, I find the M approach more intuitive: I just used the M without having to worry about it or read the manual, whereas I needed the help of this forum to work out why "DNG only" had the apparently unrelated consequence of not showing you an enlarged image. Hey ho, such are the ways firmware writers' minds work. Most of the cameras I know of that have had only low-rez JPEG thumbnails in the raw files have no ability to store files as raw only. They always create a JPEG file, and when you turn on raw capture, they add it. The Ricoh GXR was like that. Later, they added a raw-only storage mode—and immediately a lot of people complained that they could no longer zoom in on the captured image files to check focus. Link to post Share on other sites More sharing options...
jaapv Posted December 18, 2015 Share #32 Posted December 18, 2015 Indeed. Like the M8. It calculates its own JPG from the raw data, but will display the embedded small JPG for a very short while whilst calculating. Very instructive to see the effect of the long exposure noise reduction and the difference in the highlight warning. Link to post Share on other sites More sharing options...
CheshireCat Posted December 18, 2015 Share #33 Posted December 18, 2015 Indeed. Like the M8. It calculates its own JPG from the raw data, but will display the embedded small JPG for a very short while whilst calculating. "whilst calculating" == steps 3, 4, 5. All half-decent cameras do this. (I don't own the GXR, but if it does not do this, it has a crappy firmware). Note also that it is incorrect to say that "it calculates its own JPG from the raw data", because the JPG is just a compression format used only for storage (not display). Instead of JPG, you mean RGB. What happens during the "calculation" is some loading time due to the raw file size on flash storage, and consequent raw development using the internal engine, which is basically a little Lightroom embedded in the camera firmware. After the "calculation" is complete, the resulting output RGB buffer replaces the low-resolution image calculated decoding the embedded JPG thumbnail, and then you can zoom in the image with full details. Link to post Share on other sites More sharing options...
Steve Spencer Posted December 18, 2015 Share #34 Posted December 18, 2015 Since the cameras all appear to be able to generate JPGs very quickly, why not generate the JPGs needed for review on the fly from the captured DNG files, hold them in the buffer while in review mode and discard once the camera is back to shooting? It could start generating with the most recent shot the moment the view button is pressed and keep generating in the background for other recent shots. Might be a slight lag on initial view but no wasted space. This would be useless for me. I do almost all my image review long after I am done shooting. While shooting I focus on the shot I am taking next and with an EVF that allows magnification before shooting I find little need to magnify and look around after I shoot. Link to post Share on other sites More sharing options...
ramarren Posted December 18, 2015 Share #35 Posted December 18, 2015 "whilst calculating" == steps 3, 4, 5. All half-decent cameras do this. (I don't own the GXR, but if it does not do this, it has a crappy firmware). Note also that it is incorrect to say that "it calculates its own JPG from the raw data", because the JPG is just a compression format used only for storage (not display). Instead of JPG, you mean RGB. What happens during the "calculation" is some loading time due to the raw file size on flash storage, and consequent raw development using the internal engine, which is basically a little Lightroom embedded in the camera firmware. After the "calculation" is complete, the resulting output RGB buffer replaces the low-resolution image calculated decoding the embedded JPG thumbnail, and then you can zoom in the image with full details. They're not doing the raw conversion and rendering the JPEG from that. The time is spent reading the embedded preview JPEG, the thumbnail, or the companion JPEG off the storage card, and displaying that. There's a HUGE difference. Link to post Share on other sites More sharing options...
CheshireCat Posted December 18, 2015 Share #36 Posted December 18, 2015 They're not doing the raw conversion and rendering the JPEG from that. The time is spent reading the embedded preview JPEG, the thumbnail, or the companion JPEG off the storage card, and displaying that. There's a HUGE difference. Maybe But all my cameras (including the M240) do what I have explained. C For the M240, it is easy to prove: - Set the camera to save DNG only - Take a photo of a detailed subject - Turn off the camera, so that any buffer in RAM is lost - Download the DNG to your computer - Extract from the DNG file the embedded preview JPEG data (1472x976 pixels) - Visualize the JPEG on your computer, zoom in till you see the pixels - Visualize the DNG on your M240, zoom in as much as the camera allows You will see that the file on the M240 has much more detail than the embedded preview. Hence, you are wrong. Link to post Share on other sites More sharing options...
jaapv Posted December 18, 2015 Share #37 Posted December 18, 2015 "whilst calculating" == steps 3, 4, 5. All half-decent cameras do this. (I don't own the GXR, but if it does not do this, it has a crappy firmware). Note also that it is incorrect to say that "it calculates its own JPG from the raw data", because the JPG is just a compression format used only for storage (not display). Instead of JPG, you mean RGB. What happens during the "calculation" is some loading time due to the raw file size on flash storage, and consequent raw development using the internal engine, which is basically a little Lightroom embedded in the camera firmware. After the "calculation" is complete, the resulting output RGB buffer replaces the low-resolution image calculated decoding the embedded JPG thumbnail, and then you can zoom in the image with full details. thank you for elaborating and convoluting Link to post Share on other sites More sharing options...
Steve Spencer Posted December 18, 2015 Share #38 Posted December 18, 2015 Maybe But all my cameras (including the M240) do what I have explained. C For the M240, it is easy to prove: - Set the camera to save DNG only - Take a photo of a detailed subject - Turn off the camera, so that any buffer in RAM is lost - Download the DNG to your computer - Extract from the DNG file the embedded preview JPEG data (1472x976 pixels) - Visualize the JPEG on your computer, zoom in till you see the pixels - Visualize the DNG on your M240, zoom in as much as the camera allows You will see that the file on the M240 has much more detail than the embedded preview. Hence, you are wrong. I think you are missing that the huge difference in screen size has a huge difference in what looks like detail. It may looks more detailed on the M240 but that is because the screen size is so small. Just like a small print often looks more detailed than a large print. I don't think this example proves anything and certainly not that CheshireCat is wrong. Link to post Share on other sites More sharing options...
ramarren Posted December 18, 2015 Share #39 Posted December 18, 2015 I think you are missing that the huge difference in screen size has a huge difference in what looks like detail. It may looks more detailed on the M240 but that is because the screen size is so small. Just like a small print often looks more detailed than a large print. I don't think this example proves anything and certainly not that CheshireCat is wrong. (bolded) agree. Link to post Share on other sites More sharing options...
CheshireCat Posted December 18, 2015 Share #40 Posted December 18, 2015 I think you are missing that the huge difference in screen size has a huge difference in what looks like detail. It may looks more detailed on the M240 but that is because the screen size is so small. Ah, the unbelievers ! I shot a newspaper page with my M240 and I can perfectly read the fine text on the M240 screen. Then I extracted the 1472x976 preview JPEG embedded in the resulting DNG, and the text is totally unreadable. Why would the camera not display the full-resolution data contained in the DNG ? Give me a reason that makes sense. Link to post Share on other sites More sharing options...
Recommended Posts
Archived
This topic is now archived and is closed to further replies.