RAW format.

[RioRico]

I have only developed RAWs using the PentaxPhotoLab (PPL) program (a version of SilkyPix) that came with my K20D, so I can't talk about other warez. I typically adjust the in-camera JPEG settings to what I would use if I weren't shooting RAWs. Even though I shoot RAW, those settings carry over as defaults into PPL. So, when I develop a RAW in PPL, I can leave those settings as they are, or I can tweak them as needed. PPL also allows more adjustments than just the camera settings, such as histogram editing and shadow compensation. Having saved the developed image as a JPEG, I then use PaintShopPro (PSP) to crop, straighten, distort, and otherwise torture the picture.

In PPL, I may develop a RAW in several-to-many ways, especially when saving B&W converted versions. For instance, the output looks quite different when applying red, green, cyan, or infrared-simulated filters. Such filtration would be more difficult to apply to a JPEG in PSP -- not impossible, but in-camera JPEG processing reduces my editing options. Here's an analogy: You could shoot film and send it to a lab to be developed and printed by machine. That's like in-camera JPEG rendering. But if you develop and print that work yourself, YOU get to control what it looks like. Luckily, RAW development doesn't smell nearly as bad as darkroom work, depending of course on your sanitary habits.

[camvard]

Originally Posted by VTphotog

To my way of thinking, if you are going to use the default settings of your Raw developer, and batch process, you really aren't any better off than if you shoot jpeg, unless you are overexposing and need to recover some highlights - and there is only a little bit to be gained from that.

What if the camera uses a bit too heavy compression for my liking? The Canon S90/S95 produces smaller JPEG files than an old 6mp Canon. That can't be right! And according to Canon, it's likely because they've turned up the compression rate.

[RioRico]

Originally Posted by camvard

What if the camera uses a bit too heavy compression for my liking? The Canon S90/S95 produces smaller JPEG files than an old 6mp Canon. That can't be right! And according to Canon, it's likely because they've turned up the compression rate.

I'm not familiar with your camera, but can't you set image quality level? I've been able to do that since my first 1mpx P&S almost a decade ago, with Sony, Minolta and Olympus P&S's and my Pentax dSLR. If your Canon doesn't let you change compression levels, you need a less crippled camera.

[camvard]

Originally Posted by RioRico

I'm not familiar with your camera, but can't you set image quality level? I've been able to do that since my first 1mpx P&S almost a decade ago, with Sony, Minolta and Olympus P&S's and my Pentax dSLR. If your Canon doesn't let you change compression levels, you need a less crippled camera.

Yes, you can set the compression level. But even at the lowest compression/highest quality level, it still produces smaller JPEG images than an old 6 megapixel camera. I don't know how many levels there are on the S95, but the images that I linked to were at the highest quality level.

I wonder why Canon couldn't just include an additional lower level of compression and simply default to the current level...

[peripatetic]

Or maybe the compression algorithms they are using are much better than they used to be.

Are you finding compression artifacts? If not I'd suggest you celebrate rather than bemoan the fact that the file sizes are smaller.

[tclune]

Originally Posted by peripatetic

Or maybe the compression algorithms they are using are much better than they used to be.

Are you finding compression artifacts? If not I'd suggest you celebrate rather than bemoan the fact that the file sizes are smaller.

In JPEG, there are basically two ways that Canon could improve the compression rate without materially undermining the quality of the images. First, if they had been using a default Huffman table in the past and have increased their processor's speed to the point that they could calculate an optimal Huffman table from the actual image data for each image, they could reasonably expect to gain about 25% on average in their compression.

Second, if they had been using suboptimal quantization tables, they could improve the performance of the JPEG algorithm by tuning the quant tables to the particular frequency performance of their equipment (or the particular images if that was reasonably constant, but this is virtually impossible with a general-use device like a commercial camera.) However, it is unlikely that there would be much to gain in tuning the quant tables unless there was something anomalous in their hardware -- if, for example, the camera had an abnormally poor frequency response. But that would presumably mean that they would be selling a product that was not as good a camera as competitors in the marketplace. So this is an unlikely way to gain with JPEG file size.

Given that the camera's native resolution has increased from 6MP to 10MP, it is very unlikely that Canon could have increased the performance of such a mature technology as JPEG compression enough to make the file sizes actually shrink without materially degrading the quality of the compression. So, in my professional opinion (as a medical imaging software engineer), I would expect that there is good reason to worry that Canon has lowered the quality of the images that they are saving.

With lossy JPEG, the effects of the lossy compression will be noticeable on some images and not on others. A good candidate for showing artifacts would be images with long expanses of slowly-changing color (sky is a good place to look, especially if it varies from one edge of the photo to the other. You will see posterization in the sky as a compression artifact.) Now, you will very likely get posterization sometimes with JPEG under that circumstance. You would want to take photos of the same scene with both the old and new cameras to determine how pronounced the degradation in JPEG quality is with the new camera. FWIW