Huh, interesting. I thought they'd save a bit more., maybe 25%.. Wow! didn't know TIFFS were THAT large relative to DNG!
Thanks for the info.
I think it's worth clarifying a couple of concepts as they are being mixed.
First, there's raw vs. not raw. Raw means that the image has not been converted from the (usually) green-blue-red-green bayer mosaic to the RGB pixels. Conversion from raw bakes in the white balance and color space, which while it can be edited later, you never have quite the leverage you do at the time of conversion. A non-destructive raw editor like Lightroom does the conversion every time, so you always preserve that so long as you (only) use non-destructive edits.
Then there's the container. A container is a file structure, so for example NEF and CR2 can hold raw data, but so can a DNG. A DNG could also hold non-raw, already converted data. A JPG and TIFF will hold already converted data (ok, I suspect TIFF in particular COULD hold raw data, but I am not aware of any programs that do so; the actual NEF (etc.) formats are a lot like TIFF's in structure, but from a practical standpoint when you see TIFF you are seeing an image already converted.
Conversion is also sometimes called de-mosaic'ing.
The edits that the OP was trying to preserve are non-destructive edits. These can occur on both converted and non-converted images, but in LR most of the time we are talking about non-destructive edits over top of a raw image. Not always, many of us do non-destructive edits over a converted TIFF for example.
Then there is bit depth. Raw has a baked in bit depth (usually 12 or 14), and when de-mosaiced is usually converted to 8, 16 or occasionally (e.g. if doing multiple images as HDR) 32 bits. The more bits the less likely you lose image quality purely from editing actions, so 16 bits is a good way to preserve the ability to keep editing, 8 bits is not. JPG can only do 8 bits, hence once reason it is not good for editing. And somewhere in here is compression: two kinds you need to know - lossless and lossy.
Then there is color space. Raw images do not have color space, period. They may have stored with them an intent to use a color space (e.g. Nikon you can set that), but it is just that -- a hint, not a color space. The color space is attached during raw conversion to some other format. sRGB holds less color breath than AdobeRGB (etc.), but not less colors (i.e. sRGB has the same NUMBER of colors, but it goes less far toward the edges of visible colors, so it is more precise in the center than AdobeRGB, but AdobeRGB can show colors sRGB can not).
Similar to color space, white balance is "baked in" when you do the conversion. in a 16 bit format you still have lots of ability to adjust this after conversion, but you just do not quite have the same leverage you do from raw (or maybe it is that the tools are more accurate?).
So doing editing without conversion from raw maintains a lot of flexibility, but REQUIRES non-destructive editing to preserve the ability to affect the raw conversion. And non-destructive editing is what is different between vendors.
So if you convert to 16 bit wide gamut TIFF you preserve a huge amount of ability to edit, but are never quite as flexible if you didn't convert. However, by converting you remove any dependency on Lightroom and the proprietary non-destructive edits.
DNG spans all of this as it can hold both raw and converted data (and more than that, original "pure" raw and slightly processed raw), and also converted (de-mosaiced and baked in) non-raw.
So you can preserve your ability to edit by using more bit depth and lossless compression (e.g. TIFF), as a substitute for using non-destructive edits on raw, but it is not quite the same.
In ALL of these formats you can save your metadata. There are slight variations but generally speaking all formats can house all (standard) metadata, like titles, keywords, captions, etc.
In ALL of these formats in lightroom you can save develop settings (in some inside, in some in side cars).
In NONE of these formats can you exchange substantive non-destructive edits between vendors. There are minor exceptions (e.g. Lightroom and PhotoMechanic can share crop/straighten data, as well as most metadata), but edits are pretty uniformly proprietary so long as non-destructive.
And in ALL of these formats that support them, destructive edits are baked in and can be shared, but are "done" and cannot be backed out.
Sorry for the long ramble, and apologies to all those for whom this is obvious.