I admit I haven't looked through every page in this forum but I did use 'search' and I can't find a discussion on how much difference there is between compressed and uncompressed RAW. Obviously compressed is smaller (~15MB against ~18MB in my test shots) but is there any noticeable difference in quality?
I believe the answer is "it depends". In compressed (not lossless) raw several nearby highlight pixel values are swept into single values, the theory being the eye can't distinguish these highlight differences. However, if one then decreases exposure substantially in pp, so that some of these grouped highlights (the compression actually starts in the mid-tones) are brought "down" then it is possible to observe posterization.
So for properly exposed images not requiring significant post processing one could probably never see the difference. IMO the storage savings isn't an important factor, so I go with lossless compressed, usually 14 bit (D7000). Cards and storage are relatively cheap, why throw away any detail?
>compressed (not lossless) raw several nearby highlight pixel >values are swept into single values, the theory being the eye >can't distinguish these highlight differences.
Yes, I too understand this is the technique used.
>Cards and >storage are relatively cheap, why throw away any detail?
Just speculation but I wonder if frames per second are affected by lossless or lossy configuration of NEFs? If time to write to memory cards is the limiting factor, perhaps larger files slow down FPS, perhaps only a small amount?
>Just speculation but I wonder if frames per second are >affected by lossless or lossy configuration of NEFs? If time >to write to memory cards is the limiting factor, perhaps >larger files slow down FPS, perhaps only a small amount? >
It depends on how many photos you're shooting continuously. Frames are first written to an internal buffer and from there to the memory card. The manual says that up to 100 photographs can be taken in succession (page 79), however, it doesn't give a file size for each photo.
Once the buffer is full the frame rate will be slowed down. The faster the card, the faster it drains the internal buffer and allows more continuous frames at full fps. And conversely, the larger the photo file size the fewer photos can be continuously shot before the frame rate is affected by slower card drainage.
There is a buffer capacity display (example - r28, means that the buffer can hold a maximum of 28 photos at current settings). You should be able to shoot close to that many photos in a burst before the fps slows down.
Set your camera for full raw and see what the "r" number is. That should be the approximate number of photos you can burst at full FPS. Then figure out how much data the buffer holds and calculate how fast the card can drain that much data. That should tell you how long you should wait between bursts based on the number of photos per burst.
Thu 06-Jan-11 03:40 AM | edited Thu 06-Jan-11 03:41 AM by DeanAZ
It seems from my testing that there are really two things that will affect the number of frames you can take in rapid succession. 1) Camera settings that affect the buffer number (rXX) 2) Card write throughput
Each of these is actually driven by multiple factors. 1) Camera settings
-RAW settings - The best RAW r number I could get was r11 It mattered very little whether or not 12 or 14 bit was selected or whether lossless was selected. Even RAW + JPEG made an insignificant impact on the buffer.
-Auto Distortion ON/OFF ON results in a r08 no mater what other settings are selected! (even for small/basic JPEGs)
I would expect that this would be a function of the file size and whether there was a RAW + JPEG shot. The card speed also comes into play here but if your camera settings are such that you never get a large r number the card speed is almost a trivial discussion as the buffer will fill and slow way down.
With an r08 in the display, I could usually get about 10 shots before the fps dropped as some data is cleared while shooting making room for a couple more frames.
Thu 06-Jan-11 10:43 AM | edited Thu 06-Jan-11 10:44 AM by PAStime
>It mattered very little whether or not 12 or 14 bit was >selected >or whether lossless was selected.
Thanks for sharing that interesting information. I agree with your reasoning: an r08 reading as compared to much higher numbers obtainable with other settings will likely be the limiting factor. And since you don't see differences in this value in between 12/14 bit and lossless or not, my question has been answered!
The effects of auto distortion is interesting. It must always operate on the raw data and require some signficant processing.
Auto distortion does not affect the raw data. However, in every Nikon raw data file is an embedded jpeg (full size, basic quality, about 1 mb or so size), and its generation will require auto distortion processing if turned on. I'm not sure as to significant processing, have not measured.