[hugin-ptx] Re: Speed reference...

ur external drive is USB 2.x ? or eSata? anything not eSata, I'm suprised you are able to function at all Dale Date: Mon, 27 Jul 2009 20:07:07 -0400 Subject: [hugin-ptx] Speed reference... From: nicolas.pellet...@gmail.com To: hugin-ptx@googlegroups.com Curious to get a speed comparison

[hugin-ptx] Re: Speed reference...

USB 2.0 I'll copy to local disk and retry. Thanks for the input. nick Stephen Leacock - "I detest life-insurance agents: they always argue that I shall some day die, which is not so." On Mon, Jul 27, 2009 at 8:39 PM, Dale Beams

[hugin-ptx] Re: Speed reference...

Update, still goes about same speed. What I don't like is seeing enfuse work for hours, with 1.3 to 1.8 gigs of ram, and nothing more than 2-4% cpu (out of 25% for a single thread). Any advice for speeding this up? And the question is still open for those who make gigapixels panorama... how log do

[hugin-ptx] Re: Speed reference...

> > USB 2.0 > > I'll copy to local disk and retry. Thanks for the input. > I don't think it matters much, unless you keep your swap file there. I didn't notice you mentioned how much ram you have installed? That would be the bottleneck for sure. In any case, you can run enblend withe the -v (ve

[hugin-ptx] Re: Speed reference...

Zoran Zorkic wrote: > I saw it go up to 8500mb for my last pano (47000x12000, 75x13mp) so if > you have less it will take ages while it swaps. > > Though I'd be very much interested in a way to speed up or optimize > the process. The problem is that writing this kind of image processing software

[hugin-ptx] Re: Speed reference...

Try running enblend with -a -l 20. I got a nice 3x speed up compared to defaults. On Jul 28, 2:07 am, Nicolas Pelletier wrote: > Curious to get a speed comparison from other people... specially those who > generated giant panorama. > For these 2 examples, this is only the stitch part. The contro

[hugin-ptx] Re: Speed reference...

I'll try the -v -a and -l 20 To answer your question, I have 8 gig of ram on a 64 bit box... but I have hugin in 32 bit, so limited to 2 gig per process (i.e. passing the -m2000 value). Paul, I would imagine that if I run in 64 bits, and let enblend use 6-7 gigs of ram, speed should improve dramat

[hugin-ptx] Re: Speed reference...

On Wed 29-Jul-2009 at 11:23 -0400, Nicolas Pelletier wrote: > > BTW anyone know why if I ask for the 3 exposure images + the > enfused result, it will generate the 3 exposures. It will NOT > enfuse them but enfuse each and every warped image, and then > emblend them. seems like double the work.

[hugin-ptx] Re: Speed reference...

Sorry, was not clear enough on last point What the system did: Take the 36 images (12 per exposure). Enblend the first exposure Enblend the second exposure Enblend the third exposure Enfuse the first 3 images Enfuse the next 3 images ... [12 times total] Enfuse the last 3 images Enblend the 12 enf

[hugin-ptx] Re: Speed reference...

On 3 aug, 02:04, Nicolas Pelletier wrote: > What I expected: > Take the 36 images (12 per exposure). > Enblend the first exposure > Enblend the second exposure > Enblend the third exposure > Enfuse the 3 exposures By taking the following approach, you allow for the bracketed images to be aligned

[hugin-ptx] Re: Speed reference...

On Sun 02-Aug-2009 at 20:04 -0400, Nicolas Pelletier wrote: > >What I expected: >Take the 36 images (12 per exposure). >Enblend the first exposure >Enblend the second exposure >Enblend the third exposure >Enfuse the 3 exposures > >I thing it would have skipped a whole lot of processing. And, unles

[hugin-ptx] Re: Speed reference...

On Mon, Aug 3, 2009 at 5:54 AM, Bart van Andel wrote: > > On 3 aug, 02:04, Nicolas Pelletier > wrote: > > What I expected: > > Take the 36 images (12 per exposure). > > Enblend the first exposure > > Enblend the second exposure > > Enblend the third exposure > > Enfuse the 3 exposures > > By tak

[hugin-ptx] Re: Speed reference...

On Mon 03-Aug-2009 at 09:04 -0400, Nicolas Pelletier wrote: >Let's say a fish eye, so 3 pictures to do a 360 degree (0, 120 and 240). > >I do 9 shots >0 ev at 0 120 240 >+2 ev at 40 160 280 >-2 ev at 80 200 320 > >All the information to do a properly enfused 360 degree is available, but >would the

[hugin-ptx] Re: Speed reference...

> Then there is something I probably don't understand... > > I thought that once nona was run, enblend and enfuse had no impact to > alignment, everything had already been calculated\preprocessed. enblend and > enfuse would only use this information. I think you understand that part correctly. H

[hugin-ptx] Re: Speed reference...

I see what you mean. We have no guarantee that the seam will land exactly in the same place for all exposures... leaving fusing them beforehand the best option. Thanks for the explanation. nick On Tue, Aug 4, 2009 at 9:14 PM, Bart van Andel wrote: > > > Then there is something I probably don't