On Jun 15, 2006, at 2:09 PM, keith_w wrote:

>> Given that something is lost on the initial jpeg capture,
>
> You know, I hate to be picky about this, but...nothing is actually  
> LOST, per
> se, on initial capture of a jpeg image.
> What's there is there, as your lens/camera system captures it and  
> delivers it
> to the sensor. Just because you have chosen to capture an image as  
> a jpeg
> doesn't mean you've selected an inferior image format.
>
> Whatever the sensor passes along by way of the camera's internal  
> software,
> which apparently identifies it as a jpeg, or tiff or raw image, is  
> what it is.
> How can any of that be classified as a "loss?"
>
> The image captured as a tiff or a jpeg is converted by the camera's  
> internal
> firmware (I suppose ?) to be what it is. Same with RAW.

This is incorrect.

What the camera captures on the sensor is RAW data, a 12bit deep  
intensity map in an RGB mosaic with one value for each photosite.

When you instruct the camera to save that data as RAW format, it  
writes it to a file on the camera's storage card, structured with the  
settings you had made to the camera's rendering engine (sharpness,  
contrast, saturation, colorspace, etc) as metadata. The camera  
performs none to little processing on RAW data itself, other than  
(for some cameras) doing lossless compression. It also renders a  
thumbnail and a preview image in highly compressed forms and includes  
that in the structured file.

When you tell the camera to save the data in TIFF format, the camera  
takes the RAW data and applies those settings, performs gamma  
conversion and chroma interpolation, and then interpolates the  
resulting RGB intensity map down to 8bits per channel. This is RAW  
conversion rendering to an 8bit RGB representation. It writes the  
resulting data out to a file in structured TIFF format. RAW  
conversion itself loses significant amounts of data through the gamma  
correction function, then the interpolation to [EMAIL PROTECTED] loses  
even more data. The result cuts the dynamic range by anywhere from  
3-4 stops.

When you tell the camera to save the data in JPEG format, it does  
everything it does for TIFF format and then applies a JPEG  
compression algorithm afterwards. Depending upon the subject matter,  
the implementation of the JPEG algorithm, the chosen quality level,  
and the data itself, this compression can lose very little to quite a  
bit over the TIFF. In general, comparing TIFF to JPEG highest quality  
format files, the difference is small and not significant to image  
quality. Comparing TIFF to highest compression JPEG format, the  
difference is substantial.

A conservative estimate is that the conversion from RAW to TIFF or  
JPEG high quality in-camera represents between 40-50% data loss over  
the original RAW sensor capture.

Godfrey

-- 
PDML Pentax-Discuss Mail List
[email protected]
http://pdml.net/mailman/listinfo/pdml_pdml.net

Reply via email to