From: Milton Aupperle <milton@outcastsoft.com>

Date: November 19, 2007 11:50:46 PM MST

To: Astro_IIDC@yahoogroups.com

Subject: Re: [Astro_IIDC] Re: Imaging Source Abandoning FireWire?

Hi Duane;

On 19-Nov-07, at 11:15 PM, Duane wrote:

Do you sell a different version of Astro IIDC for microscopy? At work we primarily use Zeiss 

driven cameras and software for light microscopy but I'm wondering what advantage Astro 

IIDC could offer. Our electron microscopy is pretty specialized (Gatan and AMT).

Same product.

For Microscopy it's primarily medial / biological, however there is no reason it can't be used for other things.

The more important things for microscopy are live flat field adjustment (Altivec or SSE accelerated), quick and non intrusive frame / movie grabs (in Astro IIDC 4 you can also re-name the movies/files you grab afterwards too), automautic settings saved in a text file for each file/ movie grabbed, selectable spot color balance, ability to balance to a specific color auto exposure and being able to change camera settings via a key stroke.

I'll be watching for ver 4. I for one know that you'll drudge through it and come out with 

an excellent upgrade. You have some competition out there but it doesn't come close. 

Astro IIDC makes it fun—now if only my sky would make it as fun.

Well it' is supposed to be about collecting images under "hostile" environments with minimal screwing around with hardware or software.

The main issue I've run into in porting to MacIntel is that SSE on x86 is a bad joke compared to PowerPC Altivec. Not just performance which Altivec kicks it butt all over the place, but also because it's missing instructions and the work around to solve this kill the performance to the point I may as well just write scalar.

For example, a lot of the math I do requires doing calculations and converting floating point values to unsigned 16 bit integers for images. On Altivec, I convert 8 floats into 8 saturated (ie. saturated means properly truncated so that values is in the range 0 to 65535) unsigned integers in 3 instructions. For scalar this same operation works out to being 56 instructions and for SSE, the best you can do it in is 32 instructions. If I add in vector load and unload time, SSE is only marginally faster than scalar and scalar is a lot easier to write and debug. And MMX doesn't do floats so it's useless.

Do you think it would be possible to have Astro IIDC judge shape while it judges 

sharpness? Sometimes there are very sharp images but the shape is distorted enough that 

stacking causes problems. Size is another issue. Sometimes the image bows (or bubbles) 

out and sometimes it does the opposite—making the disk change size. These things all 

together affect the stacked image. I may not explain that well, but you've done it enough, 

I'm sure you know what I mean. If these could be fixed (unbowed/resized) or tossed out, 

my sky wouldn't have to be quite so perfect to glean a good shot.

Your explanation is fine, I call it "boiling images" and it's common in poorer seeing.

Shape matching isn't going to be in there and I doubt it fit would work any better than the adjustments we have already. Basically the matching tolerance setting is what controls what it tolerates for matching to different shapes.

One thing you will probably like is that you can now select the sharpest frames visually in a frame by frame basis via a list of the sharpness values and can also select which frame you feel is the most representative for matching.

And after shooting 30 gigabyes of  Lunar video and moving it from the LapTop to the G5, I'm off to capture Mars for an hour as I'm sitting in a nice steady air spot right now , which the 300 mb NAM weather site correctly predicted.

TTYL..

Milton J. Aupperle