JPEG vs RAW

I would think before it was put in public they would clean it up, but we are talking about chrysler.

I don't think that negates the use of jpg though.

Douglas Kirkland is another that comes to mind, I doubt you'll find any dust spots on his work. He said in an article I read that he just hasn't seen enough difference in the quality to warrant the extra space and labor involved, but he is one of those types that has perfected getting it in the camera. He does admit to using it occasionally in difficult or tricky lighting situations though.

Please don't get me wrong, I'm not trying to beat up on RAW because I use it, but just because some don't, doesn't mean they don't care about quality, that's my only point.

I guess my point is why buy a camera with a 12-bit sensor, only to take 8-bit photos, compress it with a 4-bit algorithm and display it on 2-bit monitors? (I know, the last one is a stretch, but it seemed so poetic, somehow :lol

I think that by the time one 'graduates' to a DSLR, they need to be thinking seriously about RAW, or else wondering why they wasted all that money on a glorified digicam. Note that I will concede speed issues to those sports photographers that don't have cameras yet that can take large back-to-back buffer-fulls of RAWs. That situation is slowly being rectified. And I certainly sympathize with those that have picked up one of the new crop of 8 MP digi-wonders, only to find out that RAW mode is pretty useless on them, due to buffer hangs during the CF card write process.

theoldmoose wrote:
I guess my point is why buy a camera with a 12-bit sensor, only to take 8-bit photos, compress it with a 4-bit algorithm and display it on 2-bit monitors? (I know, the last one is a stretch, but it seemed so poetic, somehow :lol

I guess I'd say becasue sometimes, you just can't tell the difference. I totally agree with you from a totally technical standpoint but when I bought my 10D, it was because 1st I wanted the extra megapixels that I wasn't getting from my D30, secondly it was the digital SLR that had the shooting features I wanted, thirdly because the quality was as I desired.

There are many differences between a 10D and a P&S digital other than just RAW. I don't think I need to go into all those, but implying that somehow someone has wasted their money for not using RAW, is just wrong.

I've shot 6 jobs since this thread started and have used RAW on all of them, I have shot plenty of other jobs where I have shot both RAW & JPG and my cutomers did not suffer for quality in any of them. All the techno-speak in the world will not change the fact that you can get just as high of quality in print from either format, and that's what really matters, the final print. RAW is good if you need to do a lot of post work or are not sure about the lighting, exposure or color balance, if you are confident everything is right in that regards and you simply just make a print straight from the file, you will likely not see a difference.

This is only my opinion and though it may differ from others, that's all it is, just an opinion.

It sounds suspiciously like we are in violent agreement...

I like the idea of simultaneous RAW + JPEG capture, specifically because I don't wish to spend a lot of time on raw workflow processes, unless it is necessary. If I like the JPEG as is, then I'll go with that, instead, and keep the raw as an archive, "just in case".

As my skill with this camera increases, the number of keeper JPEGs has been steadily increasing, as well.

RAW resolution

one of the earlier posts to this thread referred to changing the resolution of the RAW file.
I there any difference in uping the resoulution of the RAW file vrs. changing it in photoshop?
assuming that is the only change made and saving as TIFF
...

The answer to this question is "because all digitally 'output' images are 8-bit images any way, and humans really can't perceive much more than this". (BTW, I'm not sure what you mean by "4-bit algorithm...")

Images captured with 12-bit precision are always converted down to 8 bits before a human views it (either in print or on screen), the advantage of the extra bits of precision comes when doing manipulation of the image. An image survives "extreme" post-processing better with the 12-bits of data intact. An image that isn't heavily edited in post-processing will really not display perceptible advantages from RAW capture. In some environments, that ability to do drastic post-processing may not be valuable, and the cost of missing a shot while changing flash cards might be high, or the cost of extra labor in post-processing might be high, so it may make more sense to shoot jpg.

There are plenty of reasons for shooting in RAW or in JPG, it's really not as cut 'n' dry as many people on the boards try to make it out to be.

All that said, I always shoot in RAW unless I'm worried about flash capacity, but then I tend to shoot low volume, purely as a hobby, and with a big, honking flash card.
-harry

I think you are asking about upressing a RAW vs. a JPG? For either, it must be done in software such as PS. (Assuming using LG/Fine jpg)

If so, I do believe there is a point to which there is a difference. I've had beautiful 16x20's made from rezed up jpgs, but have had some that started to show jaggies, it really all depends on the picture. Portraits seem to fair better in jpg than say a landscape. If you're shooting for really big enlargements, I would stick with RAW, even if it doesn't always make a noticable difference, that fact that it may is reason enough IMO.

RAW provides flexibility at a cost of memory size. "Usually", it needs more time to process RAW files. However, it is much easier and better to correct RAW than problematic JPEG file.

But if the photo is properly shot, I don't see much difference in JPEG vs RAW.

The presence of JPEG and RAW should have their reasons.

Assuming that you don't mean 'uprezzing', there is some confusion about the 'resolution' setting in image files. In essence, it really doesn't mean a thing, other than if you have a 500 x 700 pixel file at 100 pixels-per-inch (PPI) resolution, that you intend it to cover a 5 x 7 inch area of the screen.

This, of course, will only happen if your screen is running 100 PPI. If it is running 72 PPI, then the picture will appear larger on the screen, since the same number of pixels will be spread over a larger area. Changing the resolution of the image won't change the size on the screen, only changing the number of X,Y pixels will do so. That is because display screen are strictly pixel-oriented devices. If you don't match the resolution of the image to the display resolution, then you won't get the size you expected on the screen. Unfortunately, everyone seems to set up their displays with different PPI settings, and since the current of OSes don't do scaling of images for you 'under the covers', you are stuck at assuming that everyone has a 72 PPI display, and put out with the idea that at least you won't overflow their display area, and the folks with higher resolution displays (100 PPI or higher) can just go pound sand, and use a magnifier to see your images.

Now, other devices may care what the PPI setting of the file is, but typically only because some application like PhotoShop is interpreting the ratio of image PPI to printer DPI (which is a completely different beast) to fit the image to the paper size, but when resizing for printing and other dispositions, you will be better off paying attention to the actual final size, and check that as the PPI is re-calculated as you fit an image to a larger format device (such as a 13 x 19 print) that the calculated resolution doesn't drop much below about 180 PPI, or else you will get a poor print.

At that point, you might consider 'uprezzing', which actually increases the number of PPI without increasing the size of the output, by interpolating neighboring pixels with various algorithms. There is an art (and a science) to uprezzing images effectively, since you are trying to create something out of essentially nothing (just hints about might be appropriate given the pixels surrounding the one that you attempting to create).

What does this all mean? Well, for instance, a Digital Rebel with 6 MP sensor produces about 2048 x 3072 (I forget the exact numbers) pixel images. I believe the default resolution is 300 PPI, but you can essentially ignore that.
If you want to post the image for the web, commonly a maximum width of 700 pixels is chosen, so PhotoShop or whatever would need to downsample (throw out about 2/3 of the pixels) the image.

If you want to make a print with about 180 PPI, then you could make an approximate 11" x 17" image, before objectionable pixelization could be seen at normal viewing distances.

If you want a 16 x 20 from a 6 MP image, you will need to uprezz. You can get away with maybe up to about a 50% increase or so by uprezzing, but there are limits. You'll know when you hit them. QImage is mentioned a lot in regard to uprezzing photos, but you can do it with PhotoShop, but you should do it in small incremental steps, using the bicubic algorithm. The results are fairly decent, I've heard.

Just to add to the above excellent information:

I've tried QImage, Photoshop, Genuine Fractals and Extensis Smartpixel and I believe they all do a fine job. (Really couldn't see much of a difference in anyof them) C1 does a good as well when you uprez at conversion time.

I like Qimage because it does it automatically when you print only and leaves the file untouched. That way you don't have to make different versions of the same pic for different print sizes. Plus you don't have to store all these big files for print either. (But if you want to you can save them in QImage as well)

unprezzing etc

great explanation
i suspected the basic concepts but did not fully understand them
as i think of my post processing steps there remains some need for further clarification

1) i assume unprez and photoshop term resample are the same thing

2) the rebel JPG max defaults to 180@3072x2048 and RAW 240@6144x4096
i assume these are std defaults not something i set in photoshop
seems like the RAW are 5 times as large
is that due to a) JPG compression or b) "extra" RAW data re: RAW not presumeing any user preferences c) something else d) am i confused about these sizes

3) my workflow typically is to modify the image and save as uncompressed TIFF. I always resample to 300 pixels/in for prints. am i better off not resampling is at an acceptle size? What is that acceptable size rule of thumb? if i resamply what exactly is a "small increment"

well i am off to shoot some cyote kits living in a culvert behind my house. looking forward to any feedback...

1) yes, unless you are 'downrezzing', but then that is still re-sampling.

2) you might be confused about the actual sizes, but the 300D saves only a medium-sized JPEG when recording RAWs, so if you extract the JPEG from the RAW, it will be a smaller picture. If you convert the RAW to JPG or TIFF, you end up with the same size pic as the RAW (in terms of X,Y pixels). The firmware hack for the 300D lets you set the embedded JPEG to large, which is a nice feature, if you want to be able to exyract full-size JPEGs from the RAWs without doing a RAW to JPG or TIFF conversion.

3) 300 PPI could be considered overkill for prints, but harmless. I wouldn't bother re-sampling an image at all, until you decide what you are going to use it for. In other words, make all your color/contrast/etc. corrections, and save the file off as .PSD with adjustment layers, including some amount of medium to light sharpening. Then, when you want to print or publish (to the web, etc.) open the file resize (and/or resample as needed), do a final sharpening, and save it out with a modified file name, possibly as a JPG, if you are publishing for the web. So, an original shot might be IMG_1299.CRW, the full-size TIFF conversion might be IMG_1299.TIF, the full-size .PSD file might be IMG_1299.PSD, and a JPEG resized/re-sampled and sharpened for web display might be IMG_1299_500X700.JPG, etc.

Keeping the original .CRW (with the C1 .work file that contains all the develop tweaks), the .TIF, the .PSD, and any/all destination device files can add up to several tens of megabytes for a photo. That's why I archive to DVD, instead of CD. Tons more space to hold all this stuff.

You want to keep all this stuff around, though, so you can retrieve/rework/revisi​t images later, when you change your mind about cropping, etc., or need to re-purpose an image. Re-working a JPG will produce posterization, etc. problems. JPGs are final file formats, never intermediate or working file formats, IMO.

Oh yeah, I forgot about the question of RAW vs. JPG size.

RAWs contain 12-bit sensor data, and tend to be about 6 MB for a 6 MP sensor. The data is compressed, but using lossless compression, so no loss of quality results. Lossless compression, though, is not nearly as efficient, of course, if you aren't allowed to throw out anything you need.

JPEGs are compressed 8-bit images, so you've immediately tossed out 4-bits of sensor data for every pixel. On top of that, a theorectically perceptually lossless (but real in terms of data loss) compression is performed, which typically reduces the data by a factor of 8 for high quality images, or even more, if you can except some artifacts in the result.

As you can see, if you shoot JPEGs, you are best off nailing the exposure, etc. correctly the first time, and not attempting to make any post-shot corrections to the JPEGs. RAWs can be much more forgiving of white balance and/or exposure mistakes, but will typically take longer to write to the flash, take more flash storage space, and will take longer for you fiddle with during post processing.

But, if you aren't a 'deadeye' with the camera, RAWs can save your rear.

more on RAW

not sure if you are a 300D owner or if these terms are common across Canon products but
it appears that in what is referred to as the "basic zone" RAW is not an option only in the other half "creative zone"
when creating a RAW there is also a small file created called a THM
i assume that is only useful for the LCD monitor
The "imbedded JPG" is not intuitively obvious as a file
not sure I care since i only need to interface with my computer/photoshop/CS
it appears the photoshop/CS RAW interface remembers what RAW settings you made the last time you opened and adjusted things it so i will need to write down the original default CS settings if I am curious what the baseline was - it appears at least temp and tint are unique for each pic
check me on what get done in what format
CS/raw - probably only adjust exposure and white balance then save as PSD
cs/PSD - anything else that needs done as adjustment layers
cs/TIF - save from flattened PSD - crop, resample if needed then print
cs/JPG - save from flatteded PSD if you need to publish to web or email or

comments?
thanks for your patience...
PS: got a few pics of the cyote kits this AM before they saw me and dove back into the culvert to stay
i will be setting up a motion sensor/shutter release this PM
thats another topic...

How come no one ever posts "proof" of how RAW gives more room for post processing than JPEG? Why not take the same picture in both RAW and JPEG, then try to push the image as far as you can without really destroying it, then post the results from the RAW and from the JPEG. IMO, that'll be the true test.