Imaging Source DMK21 - grainy planetary images

[Matt Wastell]

Hi all

I have started to use my Imaging Source DMK21 USB on Jupiter - I usually use it for Lunar and Solar.
I have noticed that I obtain 'grainy' images - see attached images. I image through an 8" LX90 with a 2.5 Powermate.

Any ideas - my conditions have been poor - lots of atmospheric disturbance - so I have found it difficult to focus. I do not recall my settings but any feedback would be much appreciated.

Thanks in advance

[matt]

Hi Matt.

First question - how many frames are you stacking?

Second - Are you using Registax? If so, which sliders do you use in wavelet sharpening and how far are you pushing the sliders?

[Matt Wastell]

Hi Matt

I got about 2000 frames. I used 85% quality in Registax and the stack size was about 700.
I usually only use the top slider and only give it a nudge - I will take notes next time.

But that's it - a touch of contrast and gamma.

Thanks for your help!

[sheeny]

G'Day Matt,

I'm not sure what you mean by grainy... I've had a look at the images you've attached. They don't look grainy to me, perhaps a little soft or in need of sharpening.

A couple of things to watch to control noise:


try not to use too much gain. More gain = more noise, but sometimes its worth using more gain to get exposure times down in bad seeing.
make sure you are saving the video uncompressed.
if you use more gain, and get more noise you will need more frames to stack to smooth out the noise (of course darks could help too)

I expect you probably know all that anyway, but it doesn't hurt to get a reminder if you know it:P.

Al.

[Matt Wastell]

Thanks Al

I have taken notes and will use those tips - I know very little with this imager - it has been unassisted trial and error for me - every little bit helps.

Thanks for the prompt response.

[matt]

Matt.

Stacking 700 frames should provide a very smooth image...even with the gain fully maxed.

Try using sliders 3, 4, and 5 for now. Everyone has different techniques for wavelets, but I've found sliders 1 and 2 introduce a fair bit of speckle if not used very carefully.

But that's just my opinion. I've read lots of people only use sliders 1 and 2. Each to their own, and it's a case of whatever works best for you in your processing routine:)

How about telling us your routine?

[Dennis]

Hi Matt W

I can see some evidence of “onion rings”; concentric rings bunched up close to the limb of Jupiter. These usually indicate noise from high Gain settings or having your data bunched up to the left of the histogram where noise can become an issue in the dark tones.

Have a look at your Histogram when recording. The bottom axis shows 0, 127 and 255. If the main “hump” of data is between 0 and 127, then you will tend to suffer the effects of noise.

I try to optimize fps, Exposure and Gain so that the main “hump” in the histogram is pushed to the right hand side, between 127 and 255.

Cheers

Dennis

[Matt Wastell]

Hi Dennis

I have never looked at the histogram visual - in fact I never knew where it was until I looked after your post! I will give this a go when imaging next - thanks for the tip!

Hi Matt

What do you mean by routine?

Thanks

[matt]

What do you do after you've captured the AVIs???

For example...do you follow this 'routine': AVIs into Registax...save final RGBs as TIFFs, and then into Photoshop for combining, aligning and processing?

And what processing do you do in Photoshop or whatever your image processing software of choice is?

What's your processing 'routine'?????:D

[Matt Wastell]

Ahh

I follow this routine for all my imaging.

Registax5
-Select AVI's
-I use the slider along the bottom to find the best frame
-Set lowest quality to about 85%
-Select a feature and put an alignment box on / around it (sometimes I use multi alignment points)
-Press align...wait
-Press Limit...wait
-Stack and optimise
-Wavelet sliders
-Brightness and contrast
-Gamma panel
Final Tab - lightness and save

No photoshop, just Registax - simple.

If you have any extra Registax tips please post!

[sheeny]

Just for comparison, Matt, my routine is similar to yours except:



I often use multipoint alignment particularly for lunar (but only a single point for spectra;)) and I use "autopick" for the alignment points usually. If it picks too many points, I clear it and increase the alignment frame size. I aim for 3 - 15 points... it depends on the image.
Instead of optimise and stack, I choose "Create reference frame". That will create a stack of say the best 50 frames (based on the best frame you selected during alignment), which you can process and sharpen in wavelets page - then press "continue" and it comes back and optimises all your frames to the new sharpened reference frame. It helps to get the absolute best frames in your stack:thumbsup:.
On the stacking page, I display the stack graph. Depending on the shape of the quality curve, I'll limit the number of frames I stack. If the curve drops steeply from the LHS in an S shape, I minimise the number of frames I stack. I have stacked as few as 10 frames. If the stack graph quality curve is pretty flat and high from the left hand side I'll stack more frames (say 50 to 150). For stacking spectra, I use the frame list instead of the stacking graph, and check every frame is aligned on the same spot exactly and delete any frames that aren't.;)
In the wavelets page, I use Gaussian Linear wavelets, and I have a saved wavelet set called "Dennis Wavelets" (onya Dennis!:thumbsup:) which has wavelet1 = about 20, and wavelet 2 = about 10. I have a second wavelet set called "Light Dennis" which is about 1/2 that.;) I use these as a starting point. If the data is good, that'll be all you need. If the image is still soft after applying these wavelets I bump up the higher order wavelets a bit. I don't use wavelets on spectra. Really good data doesn't need much in the way of wavelets:). Experimentation is good:thumbsup: so play around...
After registax I often use PS just to crop any wavelet artifacts at the image's edge and add a border and title, etc, but you can crop the edge in the Finalise page of Registax.

Play around a bit and see what works.:thumbsup:

Al.

[iceman]

The images look quite good to me, Matt, and not grainy.

I think you'll find in better conditions, you'll get the sort of sharper detail you're hoping to find.

[Matt Wastell]

Thanks all!

I have some new things to try.

I will let you know how it goes next time out!