Magenta edge tinging at high ISO and a possible solution...

Started Aug 17, 2009 | Discussions thread
GordonBGood Veteran Member • Posts: 6,308
Re: Magenta edge tinging at high ISO and a possible solution...

dlacoutu wrote:

Great news, Gordon!

Yes, the project is moving along, but I did get interrupted today so I only got to work on it for about 4 hours. The user interface is complete (other than for adding any options ) and I am reading in the DNG files, parsing them a little to be sure of correct model number and etc., that the files haven't been changed by a DNG converter (I don't yet handle loss less compressed DNG's), unpacking the data, and writing the (presumed) corrected data back into the exact same formating of output file. Much of this was fairly easy as I just cribbed code from my raw conversion program, then stripped out quite a bit of parsing that isn't necessary as this program doesn't do any raw conversion whatsoever. I could have made it even simpler by looking for data at fixed file offsets but that wouldn't have been as universally friendly to different cameras such as the K-7 or if a firmware update changes location in the slightest way. When you see the project, you will see that it looks fairly professional, as I am so confident that this is going to work well that I have been coding for production.

Regarding the WB, I'd say I don't care (shooting UniWB, the "as shot" WB is pretty useless for me), but I think your second proposal is the best : modifying the "as-shot" WB to remove the green tinge would appeal to the majority of users, without messing the RAW data more than necessary.

Yes, I don't really want to modify the raw data any more than necessary and in fact there are ways not to modify it all all using horizontal and vertical gradient curves that are a feature of DNG. However, I want the technique to work for PEF files as well and they don't have such gradient definitions and there is a possibility that some raw converters may not fully implement the DNG specification and support gradients properly, so I'll stick to just subtracting variable offsets from the raw data and likely just smudging the As-Shot White Balance tags.

Maybe just a little request for your application : could you add an ISO threshold control? Meaning that only DNGs above the selected ISO will be treated. This would remove the need to move high-iso DNGs to a separate folder to be handled by you application.

As I told James above, this correction might still have a slight effect at ISO 800, so there is quite a wide range of ISO's where it would be desired. My thinking so far has been that it won't change the raw data at all or in very faint ways when the correction is not necessary as in ISO 400 and under, so it wouldn't hurt to apply it to all K20D (and maybe eventually K-7) raw files. There is a good possibility that one won't really need to keep the original files after the correction, as I am thinking that there may be no adverse effects on the image quality due to the correction. It shouldn't add any pattern noise or banding as my offset corrections are determined from a best fit straight line with all noise variations eliminated. But if necessary, I can add an ISO threshold for the correction, although this will make me have to parse the EXIF data for the ISO. I think we'll wait for past the alpha/beta stage to make a decision on this.

Thanks again for the effort!

You're most welcome!

Best regards, GordonBGood

Post (hide subjects) Posted by
Keyboard shortcuts:
FForum PPrevious NNext WNext unread UUpvote SSubscribe RReply QQuote BBookmark MMy threads
Color scheme? Blue / Yellow