StarTools

I am a new user of StarTools and have been very pleased with it. However, I have a couple of questions relating to Masks.

First, I am using the 1.3 Alpha 32 bit version with only 2 gig memory. This is only temporary as I will purchase a new 64bit machine next month when Win 8 is out.

When I use Decon and try to use a star mask, I cannot get the auto stars feature to work. The same feature seems to work ok when I'm trying things other than Decon.

Also, when I mask an item (M51) in Life I do it by clicking numerous times on the same colour choice, as M51 has numerous colours in it. Finally I get the whole object green and I keep it. However, Life does not always act on the whole mask but often on only a small part of it - usually just one of my earlier clicks.

Thanks again Ivo for creating such a great product.

Hi Tim,

When I use Decon and try to use a star mask, I cannot get the auto stars feature to work. The same feature seems to work ok when I'm trying things other than Decon.

Hmmm... that sounds and smells like a bug. Thanks for reporting this. I investigated this and found a bug that could potentially have been causing this. I just published a new alpha version of 1.3. Can you check for me if this still has the same problem?

Also, when I mask an item (M51) in Life I do it by clicking numerous times on the same colour choice, as M51 has numerous colours in it. Finally I get the whole object green and I keep it. However, Life does not always act on the whole mask but often on only a small part of it - usually just one of my earlier clicks.

When you return to the Life module, does the correct mask blink 3 times upon return?

I downloaded the latest alpha 1.3.164 from the site but it says 1.3.165 on my copy when I hit the i for about. However, I can't find the stretch command - it seems to have disappeared. So, I can't get to Decon to try the fix.

Regarding the problem with the mask when using Life - yes the whole green area does flash 3 times. However, when I increase the saturation it only seems to be applied to a small part of the image. I got around this by using a different setting "lighter then". That got most of the galaxy green so I was ok. It seems that it takes many many clicks to get the whole area green when using the same colour command. I wonder if it only works for a set number of clicks, regardless of the actual green?

Tim

I downloaded the latest alpha 1.3.164 from the site but it says 1.3.165 on my copy when I hit the i for about. However, I can't find the stretch command - it seems to have disappeared. So, I can't get to Decon to try the fix.

The stretch command is gone because it is not needed anymore! There is no difference anymore between linear and non-linear processing, and therefore screenstretching is no longer required.
Note, though, that 'tracking' needs to be engaged for Decon to be selectable. Go ahead and try applying stretching, contrast enhancement, HDR optimisation, Wavelet Sharpening, etc. and then apply Decon. It will go back in time and apply it retroactively - it will yield completely valid results, even if you data is stretched to all heaven.

Regarding the problem with the mask when using Life - yes the whole green area does flash 3 times. However, when I increase the saturation it only seems to be applied to a small part of the image. I got around this by using a different setting "lighter then". That got most of the galaxy green so I was ok. It seems that it takes many many clicks to get the whole area green when using the same colour command. I wonder if it only works for a set number of clicks, regardless of the actual green?

Yes, the 'same colour' selection mode can be a bit tempramental and could use some improvement. For galaxies, Flood Fill Lighter Pixels is your friend - just click next to any bright object and it should select it.

Hope this helps!

Thanks Ivo, you are really doing some great stuff here.

Yes, the problem with the auto star mask in Decon has been fixed. Also, thanks for the information on galaxy masks. Much better than the colour choice I was making.

Just for your information. I have an older laptop with one of the first dual core processors with 2 GB of memory running 32 bit Windows 7.

I tried loading files direct from DSS. I could load those 90mb-95mb without a problem. I opened them and did an autodev and then cropped to about 40-50mb. I continued with files that size without a problem. I did not bin.

I tried using Track and non-track. Using Track was slower with some commands but there were no crashes or out of memory issues.

However, when I tried to load files greater than that mentioned above the program crashed or I got out an of memory error before I could reach bin.

So, I think you fixed the crashes that some people mentioned on "cloudy nights". However, some of those had much older computers with less memory running XP. So, I assume their maximum file sizes would be much smaller.

I think its great the direction that you are taking to develop StarTools. However, as you use more processing power to simplify and improve the process, people that have older 32 bit machines that aren't up to the task (processing or memory) won't be able to use StarTools without encountering problems. When 1.3 becomes the production system, it may be worth making 1.2 or an earlier less processing intense version of 1.3 available to those that need it. I don't think you would support that version into the future but it would provide a bridge until those people move to 64 bit machines.

I posted most of the above information on "Cloudy Nights" as there were some other members following this discussion. I have since added the following.

It's been raining here today so I've had more time to play with StarTools. In spite of what I just posted, the file sizes in MB don't appear to be the factor that causes "out of memory" issues etc. Some files work great and others cause issues. However, I still think if you bin you can process ok.

I have tested the same file in the 3 versions that I have. 1.2, 1.3.160 Alpha and 1.3.164 Alpha. Ivo has certainly added some neat stuff to the latest version but it definitely runs much slower and has more memory issues etc than the earlier versions. I can always process files straight from DSS with 1.2 and 1.3.160 Alpha. They are snappier and more fun to work with. I don't think I'll upgrade to later versions until I get a new more powerful laptop.

--------------------
Tim

TimN wrote:Thanks Ivo, you are really doing some great stuff here.

Yes, the problem with the auto star mask in Decon has been fixed. Also, thanks for the information on galaxy masks. Much better than the colour choice I was making.

Just for your information. I have an older laptop with one of the first dual core processors with 2 GB of memory running 32 bit Windows 7.

I tried loading files direct from DSS. I could load those 90mb-95mb without a problem. I opened them and did an autodev and then cropped to about 40-50mb. I continued with files that size without a problem. I did not bin.

I tried using Track and non-track. Using Track was slower with some commands but there were no crashes or out of memory issues.

When tracking is engaged, you should not notice any slowdowns in the modules themselves. Only when pressing the 'keep' button you may notice a delay in some modules. The worst delays are probably in the bin and crop modules, as it needs to retroactively crop and bin a number of images to preserve sequence-independency.

There are now no extra memory requirements to use the tracking feature, as I swap to HDD to access non-sequential image information (despite saying so, the 25% message is no longer true in the latest alpha).

There are some optimisations I can perform yet, to speed things up a little, so watch this space.

However, when I tried to load files greater than that mentioned above the program crashed or I got out an of memory error before I could reach bin.

That is rather annoying and I'll investigate this - you should at least be able to load an image, reach and complete the Bin module without that dreaded message popping up.

So, I think you fixed the crashes that some people mentioned on "cloudy nights". However, some of those had much older computers with less memory running XP. So, I assume their maximum file sizes would be much smaller.

I think its great the direction that you are taking to develop StarTools. However, as you use more processing power to simplify and improve the process, people that have older 32 bit machines that aren't up to the task (processing or memory) won't be able to use StarTools without encountering problems. When 1.3 becomes the production system, it may be worth making 1.2 or an earlier less processing intense version of 1.3 available to those that need it. I don't think you would support that version into the future but it would provide a bridge until those people move to 64 bit machines.

Supporting 32-bit operating systems is hard because of their memory constraints. It does not really matter if you have 1Gb, 2Gb or 4Gb memory, when a program can only (roughly) access 2Gb, no matter how much virtual memory (e.g. a swap file on the HDD) you have to supplement your RAM.
The real constraint is that 2Gb-per-program limit, which is a pain to deal with, especially if your algorithms are elaborate and require multiple copies/iterations of the image in memory.

Computer upgrades are the single cheapest and most versatile upgrade you could ever make to your astronomy gear. Lighting fast, state-of-the art 64-bit operating systems like Ubuntu or Linux Mint are free, and you could even run them alongside your WinXP install, just for processing, if you wish.

That said, StarTools is still committed to supporting 32-bit operating systems, and quite a bit of development time goes into finding and implementing optimisations to reduce CPU and memory utilisation.

I have gone back to the latest Alpha with Track and tried some of the images again to see where my earlier problems actually occurred.

When I first load my 95mb image I choose track. The screen goes light and I see an indication "StarTools Not Responding". However, I wait a bit and everything is ok. I then hit autodev and I see the image. It will occasionally crash here. I then leave ROI on and cover the galaxy in green assuming that makes the galaxy the ROI. It may or may not crash here. If I make it this far, I can crop and continue ok. You are correct in that the slower speeds in this version are only after I hit "Keep".

I have reprocessed my images In StarTools 32 bit and I am very pleased. Can't wait until I get a new computer. Ivo, you have done a fantastic job and I'm surprised more people aren't aware of this great product. I will mention it on Forums when it's appropriate. With the new "Track" Feature is there still a suggested processing order? If so, has it changed from what you posted on "cloudy Nights"?

I agree that the best thing I could now do would be to get a 64bit computer with at least 4 GiG of memory. We are expecting a lot of new laptops, tablets and hybrids to be released over the next month or so with Windows 8. So, I figure I'll wait a bit and see what's available. Otherwise, I'd buy something now. I may use Linux to tide me over.

Anyway, thanks again Ivo for creating a great product and thinking ahead of your time.

TimN wrote:I have gone back to the latest Alpha with Track and tried some of the images again to see where my earlier problems actually occurred.

When I first load my 95mb image I choose track. The screen goes light and I see an indication "StarTools Not Responding". However, I wait a bit and everything is ok.

I will be adding a dialog so the user knows what is happening. The alpha that is currently up is far from feature complete unfortunately!

I then hit autodev and I see the image. It will occasionally crash here. I then leave ROI on and cover the galaxy in green assuming that makes the galaxy the ROI. It may or may not crash here. If I make it this far, I can crop and continue ok. You are correct in that the slower speeds in this version are only after I hit "Keep".

I'm a little worried about this crashing behaviour - that should not happen. I have fixed a few bugs since, so I'm hoping the next alpha version will be more stable.

I have reprocessed my images In StarTools 32 bit and I am very pleased. Can't wait until I get a new computer. Ivo, you have done a fantastic job and I'm surprised more people aren't aware of this great product. I will mention it on Forums when it's appropriate.

Thanks Tim - that is very nice to hear!
StarTools is a loss-making not-for-profit endeavour, so I rely on word-of-mouth only. More and more people seem to find StarTools as time goes on though. I just figure that if I do something right and the software is worthwhile, then people will come! I'm not going to do a hard sell to accomplish that though. Your contributions and feedback mean a lot - it really helps knowing how users experience the software, even if it is an alpha version.

With the new "Track" Feature is there still a suggested processing order? If so, has it changed from what you posted on "cloudy Nights"?

As more and more modules become tracking enabled and context aware, the processing order becomes less and less important. The next alpha will introduce more tracking enabled and context aware modules, but hopefully you have a taste now of what the whole 'tracking' thing is about.

I agree that the best thing I could now do would be to get a 64bit computer with at least 4 GiG of memory. We are expecting a lot of new laptops, tablets and hybrids to be released over the next month or so with Windows 8. So, I figure I'll wait a bit and see what's available. Otherwise, I'd buy something now. I may use Linux to tide me over.
Anyway, thanks again Ivo for creating a great product and thinking ahead of your time.

My pleasure - couldn't have done it without the positive and constructive feedback that I get from people like yourself!