Just had another thought, do you think onion rings may also be exacerbated by a jittery drive? My LXD75 can sometimes run smooth as silk, but other times the image jitters around a little (more than just seeing as it stops if you tunr the drive off)????
i believe it is a combo of bad seeing and underexposure.
ok, i am thinking too.
double slit experiment, diffraction rings.
if we have a circular object morphing into ellipses etc with bad seeing, then as registax is in the process of beta testing multipoint alignment, then with the current version the edges of these bad circles will combine during stacking to form these onion rings like a diffraction pattern.
i have notice i still get diffraction rings when star testing even though the seeing is bad. not airy patterns, just diffraction patterns???
I have stacked at lowest quality 95% & when reviewing the frames it has picked, a lot of my worst frames will be in there. If I let it stack all those images, I will get a crap image one way or another....That's why I ended up with RSI from hand picking frames & stacking!
Asi, now that's onion rings! Great test, and definitely proves the theory of underexposure. I think I deleted an image of Saturn of mine from the other week where it was very underexposed, and had obvious onion rings too.
Robert, I just want to check that you're not confusing onion rings with the "ring" that you get from oversharpening, like your far left image on your recent (great) jupiter shots.
Robert, I just want to check that you're not confusing onion rings with the "ring" that you get from oversharpening, like your far left image on your recent (great) jupiter shots.
might be a bit of both Mike, but I'm also definitely getting multiple onion rings lines (very fine and much fainter) further in on both sides of many of my shots... I'm thinking I've got a few issues working together. I'll try and be a bit scientific on the next run to isolate a few other variables.
These Onion rings look like lack of bit depth. The larger the stack the greater the bit depth. Make sure all processing steps are done in 16bpp. Only at the very last step before saving to jpeg should they be converted to 8bpp.
Tests done last night: 9/03/06 The first image: taken using 80% gamma 55% gain & 35% brightness. 1/33rd sec exposure @ 5FPS 600 frames captured/120 stacked.
The second image: zero gamma zero gain 100% brightness. 1/25th sec exposure @ 5FPS 600 frames captured/120 stacked.
Identical heavy waveletts used on both images. Identical processing in picture publisher (unsharp mask) Identical noise reduction/sharpening in neatimage.
6" refractor & toucam. Please ignore image quality, shot in terrible seeing conditions for test purposes only.
No further tests required.
Asi/Guys, it's just dawned on me from Asi's test (thanks Asi) that I'm generally running brightness a lot higher (~75%), and so I'm getting burn-out before I get sufficient gain/gamma to kill the onions.... hope I have the chance to try this out tonight. Might have to eat my words
Ah..No word eating warranted. Just 'cos 99.9% of the astrophotography population told us so is no reason, lol. Besides, not sure about you, but it's in my nature to prove it for myself rather than read 'this is how it's done, just do it'.
There is plenty of evidence to suggest that other factors
may also contribute to the onion ring effect. I have experienced
this effect with properly exposed Jupiter images.
One indication that it may be a registration problem is that one can select
the Registax prefilter option (any low pass filter) in the Optimize
(not Stack) stage and they all but disappear. Unfortunately some
detail is also lost when this is done which does not appear to be recoverable in the wavelets (I use dyadic after resampling 1.7-1.8x Mitchell).
I notice this particularly when the 'good' images of the planet seem to
oscillate between an ellipsoid and a more circular shape.
If this is not the case, could it be some kind of aliasing effect from the
Bayer pattern of the color CCD? Has anybody seen these darn onion rings that are doing tri-color imaging?
There is plenty of evidence to suggest that other factors
may also contribute to the onion ring effect. I have experienced
this effect with properly exposed Jupiter images.
One indication that it may be a registration problem is that one can select
the Registax prefilter option (any low pass filter) in the Optimize
(not Stack) stage and they all but disappear. Unfortunately some
detail is also lost when this is done which does not appear to be recoverable in the wavelets (I use dyadic after resampling 1.7-1.8x Mitchell).
I notice this particularly when the 'good' images of the planet seem to
oscillate between an ellipsoid and a more circular shape.
If this is not the case, could it be some kind of aliasing effect from the
Bayer pattern of the color CCD? Has anybody seen these darn onion rings that are doing tri-color imaging?
Glenn
Arizona USA
Thanks Glenn, good to hear it's not just me experiencing Onioning with good exposure levels. My last two Jupiter posts on this forum being clear examples. I'll have to try the low-pass filter thingy - is there a default on Registax? Can you think of a reason that under-exposure would also contribute?
The default prefilter for Registax/Optimize is off. When checked the first filter that pops up is 'flat' - same as lowpass. I found all the lowpass filters result in essentially the same result.
Underexposure certainly does limit the number of bits that will cover the limb shadows of Jupiter, perhaps gamma adjustment would correct that. With my NexImage camera I rarely get onion rings even when I expose at half saturation (128 bits) with ET 1/10s Brt 50% Sat 100% Gamma 0 (1.0) Gain 40% at F25 or so. It is with my Firewire camera (Unibrain 501c) that I get onion rings even with pels at 220.
I just did some extensive testing with IC Capture and a DFK21au04 (color) and for me on Jupiter, brightness wasn't the issue - the deciding factor was having Gamma at or above 120 resulted in onion skinning. Below is my last round of data. G=Gain (0-1024), Exp=frame exposure (seconds), Gamma (0-255), Brightness (0-63). For reference, on my system C8sct with 2.5x barlow, Exp of 1/30s and Gain of around 500 makes jupiter almost completely invisible (dark) so my tests of G=650 mad for a pretty underexposed image.
