RAW border

[tomczak]

I haven't noticed it before somehow, but there seem to be some 10px wide border around a processed RAW image (that's .CR2) that has a distinctly different tone. I tried to change RAW dialog settings in various ways, but it seems to be always there (example below, tone-stretched a little to show it better). Is that normal?

[Dieter Mayr]

Maciej,

I have never seen such border at either my Nikon or Fuji RAWs.
I also have few CR2s from my brother in law's 5DMkIII which don't show a border, too.
What camera are your files from ?

Dieter

[HanSch]

I don't see anything like that in my processed Pentax RAW files (PEF nor DNG).

[den]

This border does not show for Canon 350D CR2 file conversions to tiff performed by PWP7.0.11 or DPP3.13.0.1 or SNS-HDR1.4.22 or dcraw9.19...

1) Do you still have this border using "dcraw.exe" with options -v -w -6 -T?
2) Does it show when the embedded jpeg is extracted: "dcraw.exe" option -e?

...den...

[tomczak]

Thanks. The camera is Canon S95. Some .CR2 processed in PWP show the border, some don't. Embeded JPGs and .CR2 processed in DPP don't show the border. I'm trying to figure out what's different about those .CR2 that, processed in PWP, show the border, but I don't know yet.

[den]

Maciej...

...A possibility...

You may be seeing the difference of the published "effective" pixel image dimensions which is less than that produced with conversion software that is 'dcraw' based [or "actual pixels"] which is the case for PWP... ...there also may be a consideration here regarding various S95's image sizes and aspect ratios...

...for example: PWP/dcraw will produce a 3684x2760 image when the S95 setting is 3648x2736 and a 4:3 aspect ratio...

...if there is no offset than borders of 18 and 12 pixels widths could potentially be produced because sensor edge colors cannot be properly determined... ...these pixel widths being beyond the manufacturer's specification... ...yet DPP will most likely produce a 3648x2736 image size and show no borders rather than 3684x2760 as will PWP/dcraw...

...ie., something to do with "effective pixels" versus "actual pixels" of the S95's sensor and the camera's image size/aspect settings...

...den...

P.S. From previous off message board discussions with you, I have 3 of your S95 CR2 files and one will have borders when converted from raw to tiff with PWP/dcraw... ...the other two have CA corrections which makes the edge considerations mute.

[den]

Additional observations...

...With Canon 350D CR2 files, I can get one to one registration with a linear offset margin crop of dcraw, SNS-HDR, and PWP conversion tiffs with a DPP conversion tiff... ...but...

...When attempting to do the same with a Canon S95 CR2 file, there is one to one registration between dcraw, SNS-HDR, and PWP conversion tiffs... ...but the DPP conversion tiff apparantly also includes proprietary lens abberation corrections that not only eliminates the RAW edges ["actual pixels" minus "effective pixels"] but also corrects for lens distortions making it essentially impossible to easily one to one register the non-DPP tiffs...

...Maciej... perhaps the lack of DPP "proprietary lens corrections" may be also a contributing factor for the chroma bloom you have been experiencing with the S95's CR2 files that are converted with non-DPP software...

Illustration: ...left-dcraw-v-w-6-T; ...center-PWP7basic; ...right-DPPdefault

capture_29092013_082052.jpg (43.87 KiB) Viewed 4265 times

Notice the apparant 'close up' image area of the DPP image area [lens distortion corrections, perhaps] when compared to corresponding the dcraw and PWP image areas...

...den...