Don't be too happy about the title. I'm not going to tell you the exact process to get rid of it. Call it a trade secret reserved for the persistent guys? I'll tell you what they are exactly referring to - fix it yourself. (I'm soo mean?)
I recently intentionally uploaded a set of overprocessed images. Okay, it might bring my AR to 30% or some day to a negative number...but it doesn't bother me much. Common sense says I'll be able to upload only 20 images a week for the lack of time. And getting some rejections to improve isn't a bad idea. I'll write about my experiment.
WARNING: If you're a professional, please close this article from right here or you may pass out within the next paragraph. I assume no responsibilities for a drastic reduction in AR or uploads/month if you adopt the procedure stated below. It is just for creating a test file. Don't do it often, obviously.
So here's what I did (and you can do too. But don't submit this thing):
- Shoot a photo in JPEG at ISO400 in a compact camera (or 1600 if you have a DSLR).
- Shoot the photo indoors under a fluorescent lamp at night but set the white balance manually to daylight.
- Make sure you expose the photo properly though. And try to photograph a barcode, the ones you see on a jar of jam or whatever.
- Now open the JPEG in some noise removal software and run a noise removal.
Settings for Noiseninja users: [generate noise profile from photo- All values like sharpening, color correction,etc at defaults]
- Now save the file at JPEG again.
- Open it in Photoshop and do whatever white balance method you use and get it right. Save as JPEG for the third time, at highest quality of 12 to retain the details in this perfect photo.
Again, if you're a confused, extremely new newbie, the above method is the worst workflow in the world. DO NOT use it ever.
Okay, so we're done with creating the file we would be studying.
Now you have to open the JPEG and look at the barcode at 300% size. Notice the region where black line meets white, i.e. the edge of the bars in barcode. You'd see little grey/white squares there and some pink/green colors too probably. That is distorted pixels - the noise plus the chromatic problem. If you shoot a proper photo in RAW with ISO100 in daylight with all correct settings and open it at 300% and compare with our test file, you'd notice the difference. The pink/green color might still be there but there would be no clearly visible square/rectangular grey-white pixels.
So the next time you get a rejection stating chromatic aberration and distorted pixels - look at regions of high contrast (or where "dark meets light" in newbie vocabulary). You'll need to fix the colored patches.
Or if you get a rejection with noise and distorted pixels, look for those white-grey pixels. They are clearly there on the edges, especially if you did noise removal improperly. They are not visible at 100% but they are clear after 200% or more.
So that's all for this article. I wouldn't tell you how to get rid of distorted pixels. Research on it. But what I'd like to say is...it IS POSSIBLE to fix it all within 60sec in Photoshop. Do some research and find out how. I did find stuff on chromatic problems in the message boards. And most people would tell you to reduce image size to get rid of distorted pixels but that's not really necessary. You don't need to downsize your photos. I tested my procedure and worked perfectly. Of course I had decent photos to start with. Forget the test file. It was intended to bring out extreme distortion.
The image in the beginning of this article had distorted pixels around the edges of the tiny black-grey leaves printed on the cups.
Keep thinking. ;)
The basic thing is you can use gaussian blur of radius