From: Mark Gaffney <markgaffney@me.com>

Date: February 24, 2010 4:39:45 PM MST

To: Astro_IIDC@yahoogroups.com

Subject: Re: [Astro_IIDC] REPORT: Experiment with a pretty big focal length (C11 @ f.92 = 26 meters)


Hi Alberto, 

I`ve been able to open your new Mars images on my PC this morning! In the trade off between buying a TIS camera with a greater resolution (31 or 41 series) I decided nearly a year ago to go with a 7.4 micron PGR Flea 2, 640 x 480 pixel camera. It cost a cool $1400 new! There seemed little prospect I`d be beginning LRGB imaging any time soon back then so I chose the colour version. It gives a smaller but finer image. With the addition of the Scorpion 20SO to my stable of CCD`s I now have the option of monochrome & LRGB imaging although at only 4.4 microns square for this camera...A similar Flea B/W would be nice & there was one available from the eBay seller Tim posted the information about, however I`d run out of spending money by that time..

It is very interesting seeing you pushing your TIS cameras to the limit to of their endurance...!


Mark.

On 25/02/2010, at 7:07 AM, albe albo wrote:


Thank you Milton thank you Ray.


Milton the teoric limit is the one I'm trying to verify with such try.

I know the mathematical formulas: they seems absolute but... perhaps for a single shot without atmosphere.

The summing method and the seeing seems to modify a little that possibility.

I'd like to find that perhaps with such focal length i could resolve less than 10 pixels... perhaps 7 or 6.

Watching the video while i was grabbing it sometimes showed some nice detail.


For such reason i'd like to own a ideal camera that could allow me to go to 60 fps with good brightness and no noise.


The diagonal pattern is moving. Summing the frames UNALIGNED produce a pattern free stack while aligning even a gray flat patterned field produce a patterned stack.

Here the link to the files that i sent to TIS:

http://dl.dropbox.com/u/1094920/DMK31AF03.AS-Pattern-Mayer-NoMovies.zip

I have the 100 frames original movies too and it would be not important to recompress them if necessary.


About the sharpness/contrast.

Ok for what you say about smallest details but about the...macrosharpness (big details) i was thinking to the possibility to increase the contrast a lot (only temporarily during the estimation process)  and to assign a variable blur radius in order to cut off some  medium-high frequencies  (noise and not precious details in this case) allowing to choose among details of a certain given size.


Watching my movies:

What about downloading my IR movie (for thr moment) already cropped and aligned and recompressed APPLE-PIXLET highest quality?

I compared pixel x pixel the 2 compressions and the difference is really minimal so that i always recompress my movies after my processing including the DSO.

The ir movie is less than 500 megabytes....


TTYL

Alberto



Da: Milton Aupperle <milton@outcastsoft.com>

A: Astro_IIDC@yahoogroups.com

Inviato: Mer 24 febbraio 2010, 19:23:59

Oggetto: Re: [Astro_IIDC] REPORT: Experiment with a pretty big focal length (C11 @ f.92 = 26 meters)


 

Hi Alberto;


When your at those extreme focal lengths, why not try binning 2x2 (or even 4x4) when capturing? It will be 4x (or 16x)  brighter so you can reduce the noise levels or increase frame rate and I doubt your losing much detail. Your way over the theoretical resolution limits anyhow for F 92 on a per pixel basis. Your image scale is about 0.0345 arc seconds per pixel (about 376 pixel wide disk and about 13 Arc Seconds in real size). Smallest feature resolvable by a C11 is likely around 10 pixels or 0.345 arc seconds, so with 4x4 binning, you'd likely lose nothing. And let's not forget, there is no loss in size of the image with my "forward sum" binning method either so the image stays the same size and scale.


Something you really long focal length shooters need to remember is that sharpness is a measure of contrast (change in brightness as in light to dark) over distance, so the less contrast you have or the wider the spacing between the light  to dark area, the less sharpness there is.


With dark images (your blue image for example) that have basically no contrast, therefore sharpness is likely only influenced by noise which there is a lot of.


 In your brightest case (IR image) , the faint features on Mars are so diffuse that they don't even play any part in sharpness estimate and get lost in the noise. The only thing that will be perceived is the diffuse disk edge. And this "edge" is spread over about a 50 pixel distance and reflects a change of around 50 to 70 value over that distance, so that really isn't much gradient at all.


Add to this all the noise in the images which is causing each pixel to change value by up to 14 between frames and adjacent pixels  (I've measured this in the past and posted the Gain versus Noise statistic for these cameras on the list) and it's no wonder to me it can not figure out which frame is or isn't sharp.


The other issue is that what you consider to be blurry may not be from a contrast / over distance perspective. A double image due to turbulence can produce a better edge than what you consider being a sharp frame. One thing I have been seeing lately here is a very fast high frequency (small scale) ripple turbulence, which produces double features in the same frame (even at 5 ms exposures) , separated by 0.1 to 0.3 arc seconds. So a single sharp edge now has two sharp edges one offset from the other. Those frames actually come out as having a higher sharpness rating  because they have more edges. I weeded out most of them manually, but you can still see it in 2.5x barlow inset image of Aristarchus here:


http://www.outcasts oft.com/AstroIma ges/SRArisOcePro c_20100126. jpg


if you look closely.


I have been testing a way that allows you to select the spatial distance for measuring sharpness, but it takes a lot of time to work it all through, and I'm not yet convinced it will make any difference. For example, once you hit a certain size with some movies of different Astro targets, it makes no differences as far as what frames it selects for increasing distance (always the same order and same frames). And the sharpness algorithm is different for Lunar and Planetary sources too, as Lunar tends to have sharper edges where as Planetary doesn't, so I'm having to run a lot of tests before I can draw a conclusion as to if it's "better" or not.


If you want to send me your movies on a CD (or DVD as the case may be for movie sizes) be like Jim (and Allan has) so I can look at them and test with them, that would be great. If not, I have to make do with what I have shot and my seeing is nearly always turbulence limited, so I can't go beyond 5 meter focal lengths.


Lastly, your diagonal noise is referred to as Fixed Pattern Noise. If it doesn't move laterally (the Texas Instrument TLSV1501 chip set FW cameras had this very bad) and is constant in position, then stacking lots of frames will not remove it, it will make it far more noticeable as it will be perceived as "signal". You might be able to remove it using a say 255 averaged flat frame images of a uniformly gray area (wouldn't even need to be done on the telescope, as it's camera noise), but you need to have the camera at those same high gain levels and at that temperature to duplicate it. With the Texas Instruments based cameras, the pattern would change as the temperature changed, so for the first 5 minutes the line orientation was all over the place until it reached thermal equilibrium and then it was fairly constant. An example of it is shown in:


http://www.outcasts oft.com/AstroIma ges/MVMJS_ 04_03_28. jpg


where you can see it in the browner edge areas of Jupiter.


HTH..


Milton Aupperle


On 24-Feb-10, at 8:44 AM, albe albo wrote:



Hello to all,

some days before the 74" telescope image published by Jim,   i did my BIG experiment pushing  my new C11 at f. 92 .


http://tech. groups.yahoo. com/group/ Astro_IIDC/ files/Planetary/ 100207-BigMars- Mayer.jpg 


I know that i'm out of the theoretical resolution power for that OTA but i noticed that many amateur astronomers went beyond their OTA limit perhaps because of the multi image stacking that in the latest years changed the methods of creating lunar and planetary images freezing and then averaging the seeing deformations.

Perhaps we are "exploiting" the seeing which could behave like an additional lens modifying the characteristic of the OTA.

Even reading the book written by  Mobberley i noticed that he had the suspect about the different limits in webcam astronomy.


Considering the defects of my image (noise and blurriness) i was pleasantly surprised by it because there are lot of obstacles with such focal length.


1) The camera worked at 15 fps but the gain was almost at the upper limit for each channel with very high levels of  noise. I didn't want to lower the fps at 7.5. In addition my DMK31 shows a lot of diagonal banding when it's used at 15 fps with a very high gain.


2) even if the camera gain was at max the level of the CCD  illumination is still low.


3) With such magnifications i guess that even the eyepiece that i used for the the projection becomes a very critical component of the optical chain  because of its extreme usage  (9mm orthoscopic by University Optics).


4) Another serious problem is that it is almost impossible to perform a decent automatic selection between the good and the bad frames even if some good frames are present!  Often i find very good frames alternated with useless blobs both classified  at the top of the "sharpest frames". Since I took 3400 image x channel it's hard to select them manually!


5) an unexpected problem was represented by the big size of the planet that clearly shows the rotation more than smaller size not allowing very long sessions of IR-R-G-B sequences. In this case from the beginning to the end the sequence took 20 minutes!

Note: for this sequence I did an additional mistake that usually doesn't happens. I took this sequence as R-G-B-IR while usually i do IR-R-G-B. For Mars it is mandatory that IR & R must be very similar while in this case i noticed a big rotation between R (the first) and IR (the last)


All those factors together afflict such kind of experiments not allowing a "fair" comparison with shorter focals.

I was wondering and dreaming about a camera that could grab at 60 FPS even with such low condition of light but keeping a low noise but i guess that such camera don't exist or it is veeery expensive.


Example: being able to capture this huge disk with the same speed and sensitivity of a f. 20 (60 fps&low noise) i guess that the sharpness would increase and the noise would decrease a lot allowing better processing and a better evaluation.

Perhaps even using a more sophisticated "magnifier" (barlow, ocular, powermate, etc) could help for a better result.


Unfortunately i don't know a (cheap) camera that could perform such ideal sensitivity and low noise so... for the moment i must postpone this verification.


Nobody else experimented the TIS diagonal banding?

I'm in touch with TIS engineers and i sent them some stuff.

I noticed the banding only at 15 fps. with high gain (not at 7.5 or 3,75 or 30)

Such banding shift across the screen so that if you stack the frames without aligning them the banding seems to cancel itself while the banding itself seems to be used like an "hook" to align the images otherwise after the alignment process it should statistically disappear but, on the contrary, often it becomes more visible.


That's all for the moment (did someone resist to read all that?   )

Cheers


Alberto


http://tech. groups.yahoo. com/group/ Astro_IIDC/ files/Planetary/ 100207-BigMars- Mayer.jpg 









Milton J. Aupperle

President

ASC - Aupperle Services and Contracting

Mac Software (Drivers, Components and Application) Specialist

#1106 - 428 Chaparral Ravine View SE.

Calgary Alberta T2X 0N2

1-(403)-453- 1624

milton@outcastsoft. com

www.outcastsoft. com