Experience with Extended Range Imaging Technology-JPEG?

Does anyone have experience and opinions about Kodak's <a

href="http://www.kodak.com/global/en/professional/products/cameras/eri

t/erit.jhtml;jsessionid=LZ5BPLV1FL0ZDQHIO3JHWGI">ERI-JPEG</a>

Technology? Is it marketing hype or does it really add to accuracy

and streamlined digitial workflow?

From the press release, it sounded to me like a jpeg with 12 bit depth (for at least the lightness channel). It would seem like a good thing for exposure lattitude (when you don't want to burn the space for a raw file).

From what the Kodak people have told me, ERI is a lossless compression that maintains the full alterability of a tagged DCR file. Fiddle all you want, and the raw data remains intact. I've heard it can pare a 20mb DCR file to about 4mb, which would be phenomenal - not so sure I believe it.

 

It's available as a firmware update for some older Kodak digital cameras, and you can download the File Format Module (for Photosho) from the Kodak Professional website - not sure if you need to be registered as a professional user or not..

The best explanation for this is at Imaging Resources (bottom of the page): http://www.imaging-resource.com/EVENTS/PKNA02/1033204252.html

<p>

Essentially, the ERI-JPEG, captures the data that would have been thrown away when going from a RAW -> JPG and stuffs it back into the JBG EXIF header. This allows you to capture most of the information that would have been lost in creating a JPG file.

<p>

I will quote a small section here about how this is created: <i>

<p>

<ol> <li>Start with a RAW image, process it normally to a JPEG.</li> <li>Map the RAW data into an expanded 36-bit rendering space, using logarithmic magnitude coordinates for the RGB values. (Actually, I'm not sure, it may be some variation on an Lab-type color space, the key elements are that it's a space with an expanded range, and that has logarithmic coordinates. - This last basically means that a one-bit change corresponds to the same percentage change in brightness, across something like an 11 or 12 f-stop range.</li> <li>Now, take your JPEG version of the file, and convert it "backwards" into the logarithmic rendering space.</li> <li>Subtract the two images, working in the expanded rendering space.</li> <li>In a lot of places, you'll find that JPEG version of the file does indeed match the data from the RAW file exactly. Cool. For those parts of the image, you don't have to store any correction data in the EXIF header.</li> <li>With a conservative JPEG compression ratio, you'll end up a lot of the image area with no correction needed to get back to the RAW data. This means that you can compress the heck out of the correction data, even with lossless compression.</li> <li>Go ahead and do a lossless compression on the "difference image" and stuff it into the EXIF header.</li> </ol>

<p></i>