Thomas Geist wrote:

 

All i'm asking for is more convenience with it.

It's too dangerous to be that convenient.

Thomas Geist wrote:

 

You think so? I'd assume that users employing the "Convert to DNG" dialog or using DNG converter are more of the savvy kind.

Convert to DNG is essentially a way to destroy your originals.  I could imagine a bunch of people downsizing and compressing their originals without truly understanding what they were doing.

Lee Jay wrote:

 

Thomas Geist wrote:

 

You think so? I'd assume that users employing the "Convert to DNG" dialog or using DNG converter are more of the savvy kind.

 

Convert to DNG is essentially a way to destroy your originals.  I could imagine a bunch of people downsizing and compressing their originals without truly understanding what they were doing.

Convert to DNG isn’t a way to destroy your originals until you define the way one converts to DNG. Prior to LR4, convert to DNG doesn’t destroy an original. It might destroy some proprietary metadata that only the manufacturers converter could use. The newer Lossy DNG conversion is a different story as pointed out, it is partially processed data.

I’d agree it would be super dangerous on import if someone picked such an option without thinking (then yes, if they didn’t back up their original raws, it would destroy a lot of data).

But the OP is also asking for a conversion on EXPORT. How is that dangerous? 

Right. I'm not into the whole lossy compression thing. I like the idea of making smaller raw files, but I want the megapixels which remain to be good megapixels.

Mark Alan Thomas wrote:

 

Right. I'm not into the whole lossy compression thing. I like the idea of making smaller raw files, but I want the megapixels which remain to be good megapixels.

To get from raw data to lossy compressed DNG, you demsoaic and then compress with JPEG compression.  If you downsample, you do that first.

The JPEG compression is efficient and of high quality (similar to PS JPEG level 10) so the loss associated with the compression is minimal.  Downsampling is a far larger loss of image quality.

You'd be far better off with a lossy-compressed full-res file than a losslessly compressed down-sampled file of the same final size.

I guess the appeal to me of pixel-binning (if that's the right term) a 36 MP raw file down to 18 or 9 MP would be in taking relatively soft Bayer-interpolated pixels and packing them down into fewer, but purer, pixels à la Foveon. You'd still be getting some of the resolution benefits of the initial capture having been 36 MP, but without the unnecessarily large sizes and processing overhead you get with the full size files. It seems to me that a 16 MP raw file generated from a 36 MP capture would be of higher quality than a native 16 MP raw file in terms of per-pixel quality.

Or maybe this is just silliness.

That makes sense. I suppose I personally wouldn't care if the data had to be demosaiced first. When I think of the benefits of raw, I'm thinking about white balance and exposure and stuff like that, not whether the data has been demosaiced or not. At any rate, if if the pixels are being binned down to a quarter their size, they may as well be demosaiced first. To me this would still qualify as raw since I'd still be able to glean all the benefits of raw which are important to me. I wonder what Canon does to the raw files to yield its sRAW format. Have sRAWs been demosaiced?