I'm having a problem. I went out shooting tonight on my 5d3, 16-35 and shot 270 shots and stacked them of the stars. I batched out low res the first time and stacked using startrails and got this checkerboard pattern at 100%. I re-batched full res copies out using LR4.1 and stacked manually using photoshop and got the same problem. See the examples below. Can you offer any advice?
from first batch: http://dl.dropbox.com/u/1637319/Capture3.JPG
from the second batch: http://dl.dropbox.com/u/1637319/Capture4.JPG
final image merged: http://dl.dropbox.com/u/1637319/capture5.jpg
(notice in the final image the banding bottom right)
Settings for these shots were f/2.8 iso 640, 10 seconds per shot
ANY ideas at all? Is this my post processing? Individual shots look perfectly clear and normal, unless there is one that is adding a checkerboard pattern to the image as a whole but i looked at a ton of them and nothing.
I haven't done any of the kind of P-P work you've done with these images, I might try it sometimes. Looks kinda cool.
However, I gripe about noise issues of this very type affecting cameras, to the discontent of many, no doubt.
However, it's those of us pushing the equipment that run into problems like this.
from the last image you posted (cropped?) the neat little grid that shows up seems to be exacly 8 pixels across. Makes me suspicious that it's a hardware-based artifact. But still hard to say if it's your camera/sensor or the PP going on that causes it. Also seems to affect primarily the green channel.
Most of the noise I've looked at from Canon cameras predominantly affects the red channel, altho B & G can be similarly affected.
have you tried processing a similar batch of dark-frame shots with no image data, maybe lens cap on to check for similar results? Done it with a different camera model and same process?
It's possible that your camera is generating a very slight intensity bump on the green channel, every 8 pixels, and the stacking procedure is causing an additive effect. Or it could be a processing algorithm error in the software.
A few too many variables yet, you may have to try some other tests to narrow it down unless someone can recognize it from experience.