|
Re: 6D - sensor measurements and comparison with other Canon FF
In reply to rrccad,
5 months ago
|
rrccad wrote:
Thierry Legault wrote:
rrccad wrote:
actually yes / no - the average black point is in the EXIF - however, the masked pixels contain the row / column offsets from the averaged black point. canon i believe if I recall correctly does not use the EXIF based row/column black point offsets but uses only the masked pixels.
your images look suspiciously like images that don't have that taken into account.
The banding crops are just here to give an idea of the intensity and shape of banding in comparison with other cameras. Balck point level is just an offset and in Iris, one can adjust the black and white levels at any values, to make the background darker or brighter as desired.
which is only the "average" black point value - not the actual row / column offset.
You know, in astronomy with CCD cameras we are used to deal with raw 16 bits files and offset frames and dark frames. And we are doing that since more than 20 years now, long time before the first consumer digital camera appeared
yes, however we also create bias offset black frames to automatically take into account row / column shifts in actual black point.
in theory the masked row / column values should remove any banding - the fact it's there is indicating it's not.
I think that Canon uses the masked photosites to decrease the banding level, but my tests show that to significantly decrease it furthermore, you would need hundreds of masked lines and columns, instead of a few dozens currently. This is clearly not the good solution.
they do.
canon has the entire right side of the sensor and the entire top of the sensor masked.
your banding values aren't accurate / real world because they aren't taking that into account.
as far as I recall, adobe will take into account the row data values.
got my left / right top/bottom mixed up.
3 columns on the left hand side, two rows on the bottom form the offset data for the CR2 data.
as far as comparing to other sensors / cameras - you are exasperating data error that doesnt really exist because the CR2 format has embedded data to shift the black point on a row / column basis.
To note, and not just you actually - the images you showed on banding triggered all this in my noggin ..
but - it's always one thing I wondered about DxOMark - if they are taking the raw data, and making some assumptions on processing and how that RAW is processed - it's not exactly accurate to how the CR2 file is engineered to be interpreted.
In actuality - if it's not handled - against the equivalent nikon NEF, a canon CR2 would indicate twice as much deviation off the black point as a nikon NEF would. simply because of math.
| Post (hide subjects) | Posted by | When | |
|---|---|---|---|
| 5 months ago | 5 | ||
| 5 months ago | |||
| 5 months ago | |||
| 5 months ago | |||
| 5 months ago | |||
| 5 months ago | |||
| 5 months ago | |||
| 5 months ago | |||
| 5 months ago | |||
| 5 months ago | |||
| 5 months ago | |||
| 5 months ago | |||
| 5 months ago | |||
| 5 months ago | |||
| 5 months ago | 2 | ||
| 5 months ago | |||
| 5 months ago | |||
| 5 months ago | |||
| 5 months ago | |||
| 5 months ago | |||
| 5 months ago | |||
| 5 months ago | 1 | ||
| 5 months ago | 1 | ||
| 5 months ago | |||
| 5 months ago | |||
| 5 months ago | |||
| 5 months ago | |||
| 5 months ago | |||
| 5 months ago | |||
| 5 months ago | |||
| 5 months ago | |||
| 5 months ago | |||
| 5 months ago | |||
| 5 months ago | |||
| 5 months ago | 1 | ||
| 5 months ago | |||
| 5 months ago | |||
| 5 months ago | |||
| 5 months ago | |||
| 5 months ago | |||
| 5 months ago |
   |
|
Connect with dpreview
|
