| Author | Thread |
|
|
10/03/2010 03:23:19 AM · #1 |
Instead of asking the question, let me explain in my terms first, the way I understand it.
Think the sensor cells as slots that holds colored papers. If you use Raw, the slot will have more shades of the color which exposed to. So, think algorithm as a guy who is smart which knows what colors should be added for each pixel at output. So, smarter the algo-guy gets, he will need more depth of the shades of each color collected. When he needs a color, probably 5th shade of green, and slot doesn't carry that color, it will take the closest one instead.
What all that means. Well, if you want to try to fix your bad exposed photos (hoping you don't have those a lot), since there are lots of colored papers in each slot, algorithm will be able to help you to fix it. Of course more papers in each slot means more space and more work.
I usually use RAW when I take photos, JPG if I only need faster burst sometimes. Or doing not too important shooting. I want to have that option if I do something wrong. I can always turn Raw into Jpg, but not the other way.
I worked for a document company over 10 years which I was introduced to a system called "Virtual Rescan". Virtual rescan means, if you scan a document, and you can't read some of the information because it came out too bright, or too dark, you were able to adjust it. The document output was 2bit, black and white, because they are smaller sizes of files, easier to process. However, scanner was scanning gray-scale actually. It was collecting more information for the document than the output we see. So, it was easy to fix a document without "re-scanning" it but it also was a slower scanning processes.
I see both "Virtual scanning" and "Raw images" the same. Collect as much as color shades possible for a cell.
If above explanation is wrong, I would like to know a correct explanation from you. I am not expert, however I have dealt with images and processes for a long time, I hope I am not saying something totally stupid. |
|
|
|
10/03/2010 12:03:20 PM · #2 |
What you said is completely accurate, but not quite complete. The RAW image (We will assume 14-bit RAW) has 64 times the number of shades for each pixel as JPEG. There are four other advantages:
1.) There is only lossless compression, so there is no data loss each time the file is adjusted and re-saved
2.) Adjustments to a RAW file are only applied as settings, so the original data are never changed. It's as if our buddy Al G. needs to do his job over again each time the file is opened. Poor guy.
3.) The RAW file is de-mosaiced by the converter (Al G.) so if Al's future replacement (sorry Al) is better at the job, you get a better conversion, even on an older file.
4.) Converting later lets you use all the horsepower of the computer, giving better conversion. Think about it this way: when poor Al has to convert to JPEG on the fly, he has to do the de-mosaicing task and apply all the in-camera settings and do compression, all before the next time the shutter button is pressed. Because he is in a hurry and working in a small space with limited power, he's sloppy. That's in-camera JPEG. |
|
|
|
10/03/2010 02:04:36 PM · #3 |
Originally posted by kirbic:
4.) Converting later lets you use all the horsepower of the computer, giving better conversion. Think about it this way: when poor Al has to convert to JPEG on the fly, he has to do the de-mosaicing task and apply all the in-camera settings and do compression, all before the next time the shutter button is pressed. Because he is in a hurry and working in a small space with limited power, he's sloppy. That's in-camera JPEG. |
Then why isn't the burst rate faster for RAW vs JPEG, since it only has to write but not process the data? |
|
|
|
10/03/2010 02:18:04 PM · #4 |
Originally posted by GeneralE:
Originally posted by kirbic:
4.) Converting later lets you use all the horsepower of the computer, giving better conversion. Think about it this way: when poor Al has to convert to JPEG on the fly, he has to do the de-mosaicing task and apply all the in-camera settings and do compression, all before the next time the shutter button is pressed. Because he is in a hurry and working in a small space with limited power, he's sloppy. That's in-camera JPEG. |
Then why isn't the burst rate faster for RAW vs JPEG, since it only has to write but not process the data? |
Size? A jpeg will be half the size of what a raw file is or smaller. |
|
|
|
10/03/2010 02:24:28 PM · #5 |
Yeah, but to get the file smaller the data has to be processed first. That plus writing the data seems like it ought to take longer than just saving the data -- if the JPEG is at the highest quality I think the file will only be half the size or so of the RAW file.
BTW: In old (pre-CS) versions of Photoshop (from before there were digital cameras) there is also a file format called "RAW" -- anyone know what it is or how to use it? |
|
|
|
10/03/2010 03:43:12 PM · #6 |
| Keep going, people. These are beautiful explanations/ways of thinking about it; and questions. |
|
|
|
10/03/2010 04:07:01 PM · #7 |
Originally posted by GeneralE:
Originally posted by kirbic:
|
Then why isn't the burst rate faster for RAW vs JPEG, since it only has to write but not process the data? |
The burst rate is typically determined by the rate at which the data can be read out and the mirror and shutter can reset. Now buffer depth, that's another matter. Buffer depth will be greater for JPEG; the roadblock is data transfer rate. For a typical DSLR you might be saving a 15MB RAW file. At 5 FPS, that's 75 MB of data per second. 8MB JPEGS require far less (40MB/s, or 47% less data). So the buffer fills slower, and empties faster. |
|
|
|
10/03/2010 05:11:06 PM · #8 |
| How fast is that 15MB of data processed? WB, color shift, sharpening, contrast, and compression all have to be applied before that 8MB can be written. It just seems intuitive to me that all that should take longer than writing the extra 7MB ... |
|
|
|
10/03/2010 07:19:20 PM · #9 |
Originally posted by GeneralE:
How fast is that 15MB of data processed? WB, color shift, sharpening, contrast, and compression all have to be applied before that 8MB can be written. It just seems intuitive to me that all that should take longer than writing the extra 7MB ... |
Card write speeds have caught up somewhat... there is at least one camera out now that has, effectively, an "infinite" buffer when shooting JPEG. In other words, the camera is processing and writing to JPEG faster than the frames can be acquired. This is a situation that is likely to become more common as write speeds increase... until the movable mirror goes the way of the dinosaur and frame rates go through the roof.
The italicized text is the key here. The camera can convert the RAW data to the final JPEG in less than 200 milliseconds, assuming the frame rate is 5 fps. This underscores how streamlined the in-camera processing has to be, and why it can be greatly beneficial to use a converter optimized for quality rather than, er, raw speed.
Message edited by author 2010-10-03 19:22:52. |
|
Home -
Challenges -
Community -
League -
Photos -
Cameras -
Lenses -
Learn -
Help -
Terms of Use -
Privacy -
Top ^
DPChallenge, and website content and design, Copyright © 2001-2025 Challenging Technologies, LLC.
All digital photo copyrights belong to the photographers and may not be used without permission.
Current Server Time: 08/22/2025 06:17:06 PM EDT.
