DSS taking ages registering files

[Ewan]

Just a quickie

Evening all (if your up still), at about 9.30pm loaded DSS up with 200 subs hit register & she's still going :-( 4.5 hours later & only on number 61, i think i made a boob somewhere ?

I have a decent pc, 1090T Hex core(6), 2 Graphics cards, 4GB RAM & 2TB hard drive space, also 6GB paging file plus using another 2.5GB Readyboost ram off my memory stick so the pc should be able to crunch data pretty quickly i guess, anyone have any idea what i might have 'pressed' wrong ? all subs are 5184 x 3456 (only jpeg though) 200 x 9sec subs on average.

Thank you for any help

[Black Knight]

I don't know if you would have "pressed" anything wrong. Although it's a long time, 4.5 hours doesn't surprise me. I got into the habit of dividing up my image files into chunks easily digestable by DSS a long time ago. By that I mean, say 25 image chunks, and place the files from each "chunk" into a separate group, and register them in batches. Then, when all the batches have been registered separately, load them into DSS and select Check all images. Then select Register and uncheck the Register already registered pictures option.

There may be other ways to do it, but this is how I used to do it.

[rfdesigner]

I use a different package, IRIS, and last night I tried using it to register 30 images, 7000x5000 pixels approx... it was taking so long I just went to bed.. this is not a shock.

It depends on the algoithum you're using. I use one that corrects for rotation and alignment which takes longer than just algnment... much longer.

It doesn't matter what you're image is, jpeg, tiff, fits. The problem the program has is:

A: find stars

B: match the stars in each image to the first image... if just one star you've told it to track, this is easy, if you're matching the entire star field (what I do) then it's going to take a while.

C: do some maths on the image to move it into position. If you're trying to rotate the image a small fraction there's a LOT of maths.

200 images is going to take a very long time.

Oh.. also it matters if the program you're using is multi-threaded, IRIS isn't, I don't know about DSS

If you want to make sure the algorithum is going to work, get it to align crops of your images.. then it will prove itself in half an hour or less.

Derek

[Ewan]

Thanks guys, the 'batch registering' sounds a good idea cheers Andy ill give it a go :-)

Derek, there is indeed a lot a data, more than i realised i will read up on a DSS tutorial asap, DSS is multi core compatible it uses all 6 of mine which is great.

Best way i have found is trial & error & a good way to learn but your input is invaluable cheers.

[Freddie]

If you haven't adjusted the detection level you may have a very high number of stars detected. I think this will slow things down.

[zakky2k]

I've noticed much better performance when just cropping/selecting a sub-portion of the image frame when at all possible.

also drizzle options seem to slow things down a lot, might be worth double checking.

[Black Knight]

[....] also drizzle options seem to slow things down a lot, might be worth double checking.

It certainly does. Having either x2 or x3 drizzle selected can also cause DSS to fail with out of memory errors, especially with DSLR subs with their high megapixel count.

[Ant]

Try Freddie's suggestion, I stacked about 20 frames from the dSLR the other evening and I had the star detection level set to 5% - it took ages and found around 30,000 stars in each sub... which obviously took a while.

Changing the detection level to 50% meant that it only found about about 500 stars and didn't take too long at all

With a hundred subs... it's still going to take a while though!

Ant

[martin_h]

DSS needs a minimum of 8 stars to register, I usualy set mine to find 50-80 stars, more than enough to work on.

[Ewan]

Sorry i havnt got back to the replies, i normally try various % of star detection & use between 80-120 for stacking, i guess it was because they were all RAW format as about 20mb each, sorted now though. Thankyou